field_dictionary.csv

field_name,example_path,description
ai_tutor_message_count,4_supplementary_and_excluded_research_data/quant_exploratory_prompt_optimization_pilot/*prompt_analysis*.csv,"Historical aggregate AI-tutor event count retained from the pilot source; row-level trajectories are not released here, so blank-payload status cannot be independently determined and this must not be treated as the canonical nonblank profile field. Historical prompt-pilot session summaries only."
student_chat_message_count,4_supplementary_and_excluded_research_data/quant_exploratory_prompt_optimization_pilot/*prompt_analysis*.csv,"Historical aggregate student chat-event count retained from the pilot source; row-level trajectories are not released here, so blank-payload status cannot be independently determined and this must not be treated as the canonical nonblank profile field. Historical prompt-pilot session summaries only."
total_message_count,4_supplementary_and_excluded_research_data/quant_exploratory_prompt_optimization_pilot/*prompt_analysis*.csv,Historical total chat-event aggregate; blank-payload status cannot be independently determined from released pilot files. Historical prompt-pilot session summaries only.
ai_content_delivered,**/README.json,Ai content delivered recorded for this observation; interpret it within the file and parent-record context shown by the example path.
primary_main_leaderboard_exclusion_reason,**/README.json,Plain-language primary reason the record is outside the final main leaderboard.
production_evidence.ai_model_api_provider,**/README.json,"API-provider family used for the AI tutoring condition, such as anthropic, gemini, openai, openai-responses, or openrouter."
production_evidence.ai_model_endpoint_id,**/README.json,Provider-facing model or endpoint identifier used for calls; it does not encode the full experimental reasoning preset.
production_evidence.ai_tutor_messages_sent_to_student,**/README.json,Number of nonblank AI-tutor messages sent to the student in the canonical interaction trajectory; blank or whitespace-only payload rows are excluded.
production_evidence.practice_answer_submissions_count,**/README.json,"Total number of submitted practice answers, including repeated attempts on the same practice problem."
production_evidence.practice_problems_answered_count,**/README.json,Number of distinct practice problems with was_answered=true.
purpose,**/README.json,Purpose recorded for this observation; interpret it within the file and parent-record context shown by the example path.
student_experimental_arm,**/README.json,Student experimental arm recorded for this observation; interpret it within the file and parent-record context shown by the example path.
tutoring_arm,**/README.json,"Assigned tutoring modality: ai, control, or human. 'Human' refers to the tutor modality, not whether the student is human."
unexpected_ai_tutoring_content_delivered,**/README.json,Whether AI tutoring content was unexpectedly delivered during a nominally human-arm study session.
age_band,**/age_distribution_by_section.csv,Age band in the privacy-safe aggregate population table.
age_lower_inclusive,**/age_distribution_by_section.csv,Age lower inclusive in the privacy-safe aggregate population table.
age_upper_inclusive,**/age_distribution_by_section.csv,Age upper inclusive in the privacy-safe aggregate population table.
n,**/age_distribution_by_section.csv,N in the privacy-safe aggregate population table.
percent,**/age_distribution_by_section.csv,Percent in the privacy-safe aggregate population table.
population,**/age_distribution_by_section.csv,Population in the privacy-safe aggregate population table.
age_18_24_n,**/age_distribution_summary.csv,Age 18 24 n in the privacy-safe aggregate population table.
age_18_24_percent,**/age_distribution_summary.csv,Age 18 24 percent in the privacy-safe aggregate population table.
age_missing_n,**/age_distribution_summary.csv,Age missing n in the privacy-safe aggregate population table.
age_observed_n,**/age_distribution_summary.csv,Age observed n in the privacy-safe aggregate population table.
max_age,**/age_distribution_summary.csv,Max age in the privacy-safe aggregate population table.
mean_age,**/age_distribution_summary.csv,Mean age in the privacy-safe aggregate population table.
median_age,**/age_distribution_summary.csv,Median age in the privacy-safe aggregate population table.
min_age,**/age_distribution_summary.csv,Min age in the privacy-safe aggregate population table.
n,**/age_distribution_summary.csv,N in the privacy-safe aggregate population table.
population,**/age_distribution_summary.csv,Population in the privacy-safe aggregate population table.
q1_age,**/age_distribution_summary.csv,Q1 age in the privacy-safe aggregate population table.
q3_age,**/age_distribution_summary.csv,Q3 age in the privacy-safe aggregate population table.
sd_age,**/age_distribution_summary.csv,Sd age in the privacy-safe aggregate population table.
assessment_form_order,**/human_grading_sessions.csv,Randomized PQ or QP assessment-form counterbalancing assignment.
assigned_elapsed_seconds,**/human_grading_sessions.csv,event time when this expert lesson-plan review assignment was recorded. It comes from the assignment record; it is not the response submission event time. Measured in seconds from the shared origin described in README.
comparison_id,**/human_grading_sessions.csv,Comparison id recorded for one completed paid human lesson-plan comparison.
comparison_task_architecture,**/human_grading_sessions.csv,Comparison task architecture recorded for one completed paid human lesson-plan comparison.
comparison_type,**/human_grading_sessions.csv,Comparison type recorded for one completed paid human lesson-plan comparison.
completed_elapsed_seconds,**/human_grading_sessions.csv,event time recorded when this expert lesson-plan review assignment was completed. It comes from the assignment record and is distinct from the response submitted_elapsed_seconds event time. Measured in seconds from the shared origin described in README.
duration_seconds,**/human_grading_sessions.csv,Duration seconds recorded for one completed paid human lesson-plan comparison.
general_notes,**/human_grading_sessions.csv,General notes recorded for one completed paid human lesson-plan comparison.
grading_id,**/human_grading_sessions.csv,Grading id recorded for one completed paid human lesson-plan comparison.
human_lesson_plan_created_during,**/human_grading_sessions.csv,Human lesson plan created during recorded for one completed paid human lesson-plan comparison.
human_reviewer_anon_id,**/human_grading_sessions.csv,Human reviewer anon id recorded for one completed paid human lesson-plan comparison.
instrument_id,**/human_grading_sessions.csv,Assessment instrument identifier: gre_quant or gre_verbal.
lesson_plan_a_ai_model_preset_id,**/human_grading_sessions.csv,Stable identifier for lesson plan a ai model preset.
lesson_plan_a_human_author_anon_id,**/human_grading_sessions.csv,Lesson plan a human author anon id recorded for one completed paid human lesson-plan comparison.
lesson_plan_a_id,**/human_grading_sessions.csv,Lesson plan a id recorded for one completed paid human lesson-plan comparison.
lesson_plan_a_kind,**/human_grading_sessions.csv,Lesson plan a kind recorded for one completed paid human lesson-plan comparison.
lesson_plan_b_ai_model_preset_id,**/human_grading_sessions.csv,Stable identifier for lesson plan b ai model preset.
lesson_plan_b_human_author_anon_id,**/human_grading_sessions.csv,Lesson plan b human author anon id recorded for one completed paid human lesson-plan comparison.
lesson_plan_b_id,**/human_grading_sessions.csv,Lesson plan b id recorded for one completed paid human lesson-plan comparison.
lesson_plan_b_kind,**/human_grading_sessions.csv,Lesson plan b kind recorded for one completed paid human lesson-plan comparison.
linked_ssot_student_id,**/human_grading_sessions.csv,Linked ssot student id recorded for one completed paid human lesson-plan comparison.
pretest_maximum_score_points,**/human_grading_sessions.csv,Canonical field for pretest maximum score points.
pretest_score_points,**/human_grading_sessions.csv,Number of pretest response rows scored correct under the deployed scoring contract.
review_display_variant,**/human_grading_sessions.csv,Review display variant recorded for one completed paid human lesson-plan comparison.
review_slot,**/human_grading_sessions.csv,"Recorded assignment slot within a lesson-plan comparison. Values 1, 2 and 3 identify standard reviewer slots; 0 is a source-system special value used for nonstandard assignments or slots released during assignment changes. A zero alone does not establish that a completed review is invalid, an admin review, or excluded from analysis."
review_subject_id,**/human_grading_sessions.csv,Review subject id recorded for one completed paid human lesson-plan comparison.
rubric_version,**/human_grading_sessions.csv,Rubric version recorded for one completed paid human lesson-plan comparison.
student_id,**/human_grading_sessions.csv,Linked ssot student id recorded for one completed paid human lesson-plan comparison.
submitted_elapsed_seconds,**/human_grading_sessions.csv,"Expert review submission in seconds from assignment when recorded, otherwise from submission minus the recorded duration. If both assignment and duration are missing, submission defines zero; zero is not an observed duration. Missing source times remain empty."
client_seq,**/interaction_events.csv,Client-recorded event sequence value in supplementary interaction logs. Missing values mean the source did not provide a client sequence; not a global row number.
event_type,**/interaction_events.csv,"Application event category. visibility records document visibility changes; blur and focus record browser-window focus changes. These raw events are distinct from filtered test_focus_lost/test_focus_returned integrity events and do not by themselves prove a tab switch, inattention, or misconduct."
payload_json,**/interaction_events.csv,"JSON object containing the recorded payload for this application event after release redaction. For visibility events, state records the browser document visibility state (for example hidden or visible); raw blur/focus events may have an empty object. Interpret fields with event_type and phase."
phase,**/interaction_events.csv,"Study phase active when the telemetry event occurred, such as pretest, intervention, or posttest."
posttutoring_survey_lesson_plan_helpfulness_rating,**/interaction_events.csv,How helpful was the lesson plan (concepts covered) for your learning? Integer rating 1–5: 1 = Not at all helpful; 5 = Extremely helpful. Collected after tutoring and before the posttest. Event histories may differ from the final profile; neither source is overwritten.
posttutoring_survey_practice_problem_helpfulness_rating,**/interaction_events.csv,How helpful were the practice problems the tutor created for your learning? Integer rating 1–5: 1 = Not at all helpful; 5 = Extremely helpful. Collected after tutoring and before the posttest. Event histories may differ from the final profile; neither source is overwritten.
posttutoring_survey_tutor_helpfulness_rating,**/interaction_events.csv,"Student 1-to-5 rating of tutor helpfulness, collected after the intervention and before the posttest. AI-arm tutor_feedback_submitted events when the student completed the survey; inapplicable to control and human arms."
route,**/interaction_events.csv,Application route recorded with a supplementary interaction event; not a research outcome or a participant identity.
client_elapsed_seconds,**/interaction_events.csv,Client-reported event time for a supplementary interaction event. May be absent or differ from server time; use server_elapsed_seconds for the paper timing rules. Measured in seconds from the shared origin described in README.
server_elapsed_seconds,**/interaction_events.csv,Authoritative event time at which the server recorded the telemetry event. Measured in seconds from the shared origin described in README.
concepts[].allotted_minutes,**/lesson_plan.json,"Allotted minutes stored in the prepared lesson plan, its generation metadata, or its usage record."
concepts[].concept_id,**/lesson_plan.json,"Concept id stored in the prepared lesson plan, its generation metadata, or its usage record."
concepts[].misconception,**/lesson_plan.json,"Misconception stored in the prepared lesson plan, its generation metadata, or its usage record."
concepts[].name,**/lesson_plan.json,"Name stored in the prepared lesson plan, its generation metadata, or its usage record."
concepts[].practice_problems[],**/lesson_plan.json,"Practice problems stored in the prepared lesson plan, its generation metadata, or its usage record."
concepts[].practice_problems[].format,**/lesson_plan.json,Recorded assessment or practice-problem response format.
concepts[].practice_problems[].practice_question_text,**/lesson_plan.json,Full question or practice-problem stem.
concepts[].pretest_evidence[],**/lesson_plan.json,"Pretest evidence stored in the prepared lesson plan, its generation metadata, or its usage record."
concepts[].priority_rank,**/lesson_plan.json,"Priority rank stored in the prepared lesson plan, its generation metadata, or its usage record."
concepts[].why_it_matters,**/lesson_plan.json,"Why it matters stored in the prepared lesson plan, its generation metadata, or its usage record."
summary,**/lesson_plan.json,"Summary stored in the prepared lesson plan, its generation metadata, or its usage record."
ai_model_api_provider,**/lesson_plan_generation_metadata.json,"API-provider family used for the AI tutoring condition, such as anthropic, gemini, openai, openai-responses, or openrouter."
ai_model_endpoint_id,**/lesson_plan_generation_metadata.json,Provider-facing model or endpoint identifier used for calls; it does not encode the full experimental reasoning preset.
cache_read_tokens,**/lesson_plan_generation_metadata.json,"Provider-reported tokens read from a prompt cache during lesson-plan generation. Lesson-plan generation calls for which the provider reported cache-read token usage; JSON null means unavailable or not reported, not zero."
cache_write_tokens,**/lesson_plan_generation_metadata.json,"Provider-reported tokens written to a prompt cache during lesson-plan generation. Lesson-plan generation calls for which the provider reported cache-write token usage; JSON null means unavailable or not reported, not zero."
completion_tokens,**/lesson_plan_generation_metadata.json,Provider-reported output-token count for the lesson-plan generation call. Every released lesson-plan generation metadata record.
cost_usd,**/lesson_plan_generation_metadata.json,"Recorded USD cost attributable to the lesson-plan generation call. Lesson-plan generation calls with a recorded call-level USD cost; JSON null means unavailable, not zero cost."
delivered_to_student,**/lesson_plan_generation_metadata.json,"Delivered to student stored in the prepared lesson plan, its generation metadata, or its usage record."
finish_reason,**/lesson_plan_generation_metadata.json,Provider-reported reason the model response ended.
latency_ms,**/lesson_plan_generation_metadata.json,Elapsed generation-call latency in milliseconds. Every released lesson-plan generation metadata record.
lesson_plan_generation_recorded_elapsed_seconds,**/lesson_plan_generation_metadata.json,"Time lesson-plan generation was recorded, in seconds from the study-session origin."
lesson_plan_id,**/lesson_plan_generation_metadata.json,Lesson plan id for one causally ordered tutoring interaction event.
number_of_lesson_plans_used,**/lesson_plan_generation_metadata.json,"Number of lesson plans used stored in the prepared lesson plan, its generation metadata, or its usage record."
prompt_tokens,**/lesson_plan_generation_metadata.json,Provider-reported input-token count for the lesson-plan generation call. Every released lesson-plan generation metadata record.
reasoning_tokens,**/lesson_plan_generation_metadata.json,"Provider-reported reasoning-token usage for the lesson-plan generation call when exposed separately. Lesson-plan generation calls for which reasoning-token usage was available; JSON null means unavailable or not reported, not zero."
selected_for_practice,**/lesson_plan_generation_metadata.json,Whether this generated lesson plan was selected for the student's tutoring practice session.
source_lesson_plan_id,**/lesson_plan_generation_metadata.json,"Source lesson plan id stored in the prepared lesson plan, its generation metadata, or its usage record."
validation_status,**/lesson_plan_generation_metadata.json,Structural/parser validation status; it does not certify pedagogical or mathematical correctness.
concepts[].allotted_minutes,**/lesson_plan_including_practice_problems.json,"Allotted minutes stored in the prepared lesson plan, its generation metadata, or its usage record."
concepts[].concept_id,**/lesson_plan_including_practice_problems.json,"Concept id stored in the prepared lesson plan, its generation metadata, or its usage record."
concepts[].name,**/lesson_plan_including_practice_problems.json,"Name stored in the prepared lesson plan, its generation metadata, or its usage record."
concepts[].practice_problems[].answer,**/lesson_plan_including_practice_problems.json,Recorded answer in a supplementary or excluded lesson-plan practice problem. The corresponding main-cohort field is correct_answer; preserve the original answer representation.
concepts[].practice_problems[].answer_original,**/lesson_plan_including_practice_problems.json,Original answer representation preserved before answer normalization in a supplementary lesson-plan practice problem.
concepts[].practice_problems[].correct_answer,**/lesson_plan_including_practice_problems.json,Answer key stored with the practice problem.
concepts[].practice_problems[].difficulty,**/lesson_plan_including_practice_problems.json,Recorded difficulty label for the assessment item or practice problem.
concepts[].practice_problems[].format,**/lesson_plan_including_practice_problems.json,Recorded assessment or practice-problem response format.
concepts[].practice_problems[].options,**/lesson_plan_including_practice_problems.json,Answer-choice map for the question; keys vary by item and should be treated as choice labels rather than a fixed A-D schema.
concepts[].practice_problems[].options.<choice_key>,**/lesson_plan_including_practice_problems.json,Answer-choice map for the question; keys vary by item and should be treated as choice labels rather than a fixed A-D schema.
concepts[].practice_problems[].original_unnormalized_answer,**/lesson_plan_including_practice_problems.json,Original unnormalized answer value retained before canonical answer cleanup.
concepts[].practice_problems[].practice_problem_id,**/lesson_plan_including_practice_problems.json,"Practice problem id stored in the prepared lesson plan, its generation metadata, or its usage record."
concepts[].practice_problems[].practice_question_text,**/lesson_plan_including_practice_problems.json,Full question or practice-problem stem.
concepts[].practice_problems[].solution,**/lesson_plan_including_practice_problems.json,Worked solution stored with the practice problem or lesson plan.
concepts[].practice_problems[].source_problem_id,**/lesson_plan_including_practice_problems.json,"Source problem id stored in the prepared lesson plan, its generation metadata, or its usage record."
concepts[].practice_problems[].stem,**/lesson_plan_including_practice_problems.json,Practice-question text in a supplementary or excluded lesson plan. The corresponding main-cohort field is practice_question_text.
concepts[].pretest_evidence[],**/lesson_plan_including_practice_problems.json,"Pretest evidence stored in the prepared lesson plan, its generation metadata, or its usage record."
concepts[].priority_rank,**/lesson_plan_including_practice_problems.json,"Priority rank stored in the prepared lesson plan, its generation metadata, or its usage record."
concepts[].why_it_matters,**/lesson_plan_including_practice_problems.json,"Why it matters stored in the prepared lesson plan, its generation metadata, or its usage record."
lesson_plan_id,**/lesson_plan_including_practice_problems.json,"Lesson plan id stored in the prepared lesson plan, its generation metadata, or its usage record."
lesson_plan_recovery_note,**/lesson_plan_including_practice_problems.json,Explicit provenance note describing recovery of this lesson plan from the authoritative source.
source_lesson_plan_id,**/lesson_plan_including_practice_problems.json,"Source lesson plan id stored in the prepared lesson plan, its generation metadata, or its usage record."
student_summary,**/lesson_plan_including_practice_problems.json,"Student summary stored in the prepared lesson plan, its generation metadata, or its usage record."
total_minutes,**/lesson_plan_including_practice_problems.json,"Total duration specified in the recorded lesson plan, in minutes; this is planned time, not observed tutoring duration."
ai_model_api_provider,**/llm_cost_records.csv,"API-provider family used for the AI tutoring condition, such as anthropic, gemini, openai, openai-responses, or openrouter."
ai_model_endpoint_id,**/llm_cost_records.csv,Provider-facing model or endpoint identifier used for calls; it does not encode the full experimental reasoning preset.
ai_model_preset_id,**/llm_cost_records.csv,"Experimental AI condition identifier combining the model family with the configured reasoning-effort or mode, for example opus-4.8-off versus opus-4.8-xhigh."
cache_read_tokens,**/llm_cost_records.csv,Cache read tokens for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
cache_write_tokens,**/llm_cost_records.csv,Cache write tokens for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
completion_tokens,**/llm_cost_records.csv,"Provider-reported completion token count for this request. Empty means not recorded, not an observed zero. Some reasoning-only responses omit completion counts."
latency_ms,**/llm_cost_records.csv,Latency ms for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
lesson_plan_usage,**/llm_cost_records.csv,Lesson plan usage for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
lesson_plan_validation_status,**/llm_cost_records.csv,Canonical field for lesson plan validation status.
occurred_elapsed_seconds,**/llm_cost_records.csv,"AI provider request time, in seconds from the study-session origin. Missing source times are empty."
prompt_tokens,**/llm_cost_records.csv,"Provider-reported input token count for this request. Empty means not recorded, not an observed zero. Some reasoning-only responses omit completion counts."
provider_finish_reason_raw,**/llm_cost_records.csv,Provider finish reason raw for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
reasoning_tokens,**/llm_cost_records.csv,"Provider-reported reasoning-token usage when exposed separately; null means unavailable, not zero."
recorded_cost_usd,**/llm_cost_records.csv,Recorded cost usd for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
request_completion_status,**/llm_cost_records.csv,Request completion status for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
request_id,**/llm_cost_records.csv,Request id for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
request_type,**/llm_cost_records.csv,Request type for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
student_id,**/llm_cost_records.csv,"Globally unique anonymous released study-session identifier, including instrument and release scope."
token_usage_recorded,**/llm_cost_records.csv,True when at least one provider token count was recorded. Does not imply every token category is present. Missing token fields remain missing; replay estimates price the categories observed.
other_lesson_plans_used_during_tutoring[].generated_elapsed_seconds,**/other_lesson_plans_used_during_tutoring.json,"Generation time of this alternative lesson plan, in seconds from the study-session origin."
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].allotted_minutes,**/other_lesson_plans_used_during_tutoring.json,"Allotted minutes stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].concept_id,**/other_lesson_plans_used_during_tutoring.json,"Concept id stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].name,**/other_lesson_plans_used_during_tutoring.json,"Name stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].practice_problems[].answer,**/other_lesson_plans_used_during_tutoring.json,Recorded answer in a supplementary or excluded lesson-plan practice problem. The corresponding main-cohort field is correct_answer; preserve the original answer representation.
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].practice_problems[].correct_answer,**/other_lesson_plans_used_during_tutoring.json,Answer key stored with the practice problem.
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].practice_problems[].difficulty,**/other_lesson_plans_used_during_tutoring.json,Recorded difficulty label for the assessment item or practice problem.
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].practice_problems[].format,**/other_lesson_plans_used_during_tutoring.json,Recorded assessment or practice-problem response format.
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].practice_problems[].options,**/other_lesson_plans_used_during_tutoring.json,Answer-choice map for the question; keys vary by item and should be treated as choice labels rather than a fixed A-D schema.
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].practice_problems[].options.<choice_key>,**/other_lesson_plans_used_during_tutoring.json,Answer-choice map for the question; keys vary by item and should be treated as choice labels rather than a fixed A-D schema.
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].practice_problems[].original_unnormalized_answer,**/other_lesson_plans_used_during_tutoring.json,Original unnormalized answer value retained before canonical answer cleanup.
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].practice_problems[].practice_problem_id,**/other_lesson_plans_used_during_tutoring.json,"Practice problem id stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].practice_problems[].practice_question_text,**/other_lesson_plans_used_during_tutoring.json,Full question or practice-problem stem.
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].practice_problems[].solution,**/other_lesson_plans_used_during_tutoring.json,Worked solution stored with the practice problem or lesson plan.
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].practice_problems[].source_problem_id,**/other_lesson_plans_used_during_tutoring.json,"Source problem id stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].practice_problems[].stem,**/other_lesson_plans_used_during_tutoring.json,Practice-question text in a supplementary or excluded lesson plan. The corresponding main-cohort field is practice_question_text.
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].pretest_evidence[],**/other_lesson_plans_used_during_tutoring.json,"Pretest evidence stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].priority_rank,**/other_lesson_plans_used_during_tutoring.json,"Priority rank stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].lesson_plan.concepts[].why_it_matters,**/other_lesson_plans_used_during_tutoring.json,"Why it matters stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].lesson_plan.lesson_plan_id,**/other_lesson_plans_used_during_tutoring.json,"Lesson plan id stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].lesson_plan.source_lesson_plan_id,**/other_lesson_plans_used_during_tutoring.json,"Source lesson plan id stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].lesson_plan.student_summary,**/other_lesson_plans_used_during_tutoring.json,"Student summary stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].lesson_plan.total_minutes,**/other_lesson_plans_used_during_tutoring.json,"Total duration specified in the recorded lesson plan, in minutes; this is planned time, not observed tutoring duration."
other_lesson_plans_used_during_tutoring[].lesson_plan_id,**/other_lesson_plans_used_during_tutoring.json,Lesson plan id for one causally ordered tutoring interaction event.
other_lesson_plans_used_during_tutoring[].source_lesson_plan_id,**/other_lesson_plans_used_during_tutoring.json,"Source lesson plan id stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].usage_summary.first_used_elapsed_seconds,**/other_lesson_plans_used_during_tutoring.json,"First recorded use of this alternative lesson plan, in seconds from the study-session origin."
other_lesson_plans_used_during_tutoring[].usage_summary.last_used_elapsed_seconds,**/other_lesson_plans_used_during_tutoring.json,"Last recorded use of this alternative lesson plan, in seconds from the study-session origin."
other_lesson_plans_used_during_tutoring[].usage_summary.number_answered,**/other_lesson_plans_used_during_tutoring.json,"Number answered stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].usage_summary.number_of_answer_attempts,**/other_lesson_plans_used_during_tutoring.json,"Number of answer attempts stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].usage_summary.number_of_conversation_messages_referencing_plan_problems,**/other_lesson_plans_used_during_tutoring.json,"Number of conversation messages referencing plan problems stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].usage_summary.number_of_practice_problems_in_plan,**/other_lesson_plans_used_during_tutoring.json,"Number of practice problems in plan stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].usage_summary.number_of_practice_problems_instantiated,**/other_lesson_plans_used_during_tutoring.json,"Number of practice problems instantiated stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].usage_summary.number_shown,**/other_lesson_plans_used_during_tutoring.json,"Number shown stored in the prepared lesson plan, its generation metadata, or its usage record."
other_lesson_plans_used_during_tutoring[].usage_summary.number_skipped,**/other_lesson_plans_used_during_tutoring.json,"Number skipped stored in the prepared lesson plan, its generation metadata, or its usage record."
answer_changes,**/posttest_responses.csv,Number of recorded answer changes before the assessment response was finalized.
correct_answer,**/posttest_responses.csv,Canonical answer key used by the deployed scoring or grading record.
difficulty,**/posttest_responses.csv,Recorded difficulty label for the assessment item or practice problem.
format,**/posttest_responses.csv,Recorded assessment or practice-problem response format.
is_correct,**/posttest_responses.csv,Whether the recorded submitted answer matches the deployed answer key under the applicable scoring contract.
presented_position,**/posttest_responses.csv,One-based position at which the item was presented; blank means the item was never opened before test submission.
question_id,**/posttest_responses.csv,Public assessment-item identifier using P/Q form prefixes. Join to question_id in the matching Quant or Verbal question bank; the same IDs are reused across instruments.
question_open_event_received_server_elapsed_seconds,**/posttest_responses.csv,Question open event received at server for one pre-test or post-test item response. Measured in seconds from the shared origin described in README.
question_opened_client_elapsed_seconds,**/posttest_responses.csv,Question opened at client for one pre-test or post-test item response. Measured in seconds from the shared origin described in README.
response_recorded_server_elapsed_seconds,**/posttest_responses.csv,Response recorded at server for one pre-test or post-test item response. Measured in seconds from the shared origin described in README.
skill,**/posttest_responses.csv,Top-level GRE skill category assigned to the item.
skipped,**/posttest_responses.csv,Whether the item was submitted without a student answer.
student_answer,**/posttest_responses.csv,Student answer recorded for the assessment item; blank represents unanswered when applicable.
time_spent_seconds,**/posttest_responses.csv,Client-accumulated item-view time in seconds; it is not a complete wall-clock trace and may be zero for event-log recoveries.
concept_id,**/practice_problems_created_during_tutoring.csv,Concept id for a practice-problem definition generated during live tutoring.
correct_answer,**/practice_problems_created_during_tutoring.csv,Canonical answer key used by the deployed scoring or grading record.
created_elapsed_seconds,**/practice_problems_created_during_tutoring.csv,"Practice-problem creation time, in seconds from the study-session origin."
difficulty,**/practice_problems_created_during_tutoring.csv,Recorded difficulty label for the assessment item or practice problem.
format,**/practice_problems_created_during_tutoring.csv,Recorded assessment or practice-problem response format.
generation_context_lesson_plan_id,**/practice_problems_created_during_tutoring.csv,Generation context lesson plan id for a practice-problem definition generated during live tutoring.
generation_cost_usd,**/practice_problems_created_during_tutoring.csv,Generation cost usd for a practice-problem definition generated during live tutoring.
generation_latency_ms,**/practice_problems_created_during_tutoring.csv,Generation latency ms for a practice-problem definition generated during live tutoring.
generation_recorded_elapsed_seconds,**/practice_problems_created_during_tutoring.csv,"Time practice-problem generation was recorded, in seconds from the study-session origin."
generation_trigger,**/practice_problems_created_during_tutoring.csv,Generation trigger for a practice-problem definition generated during live tutoring.
options_json,**/practice_problems_created_during_tutoring.csv,Strict JSON object containing separately recorded answer choices; an empty object means choices were not stored separately.
ordinal,**/practice_problems_created_during_tutoring.csv,"Recorded practice-problem order within the source session and concept. It is not the CSV row number and need not restart for a new lesson-plan context. Preserve recorded ordinals, including gaps; do not recompute them from the released row order."
practice_problem_id,**/practice_problems_created_during_tutoring.csv,Practice problem id for a practice-problem definition generated during live tutoring.
practice_question_text,**/practice_problems_created_during_tutoring.csv,Full text of a practice problem generated during tutoring.
solution,**/practice_problems_created_during_tutoring.csv,Worked solution stored with the practice problem or lesson plan.
source_problem_id,**/practice_problems_created_during_tutoring.csv,Source problem id for a practice-problem definition generated during live tutoring.
answer_changes,**/pretest_responses.csv,Number of recorded answer changes before the assessment response was finalized.
correct_answer,**/pretest_responses.csv,Canonical answer key used by the deployed scoring or grading record.
difficulty,**/pretest_responses.csv,Recorded difficulty label for the assessment item or practice problem.
format,**/pretest_responses.csv,Recorded assessment or practice-problem response format.
is_correct,**/pretest_responses.csv,Whether the recorded submitted answer matches the deployed answer key under the applicable scoring contract.
presented_position,**/pretest_responses.csv,One-based position at which the item was presented; blank means the item was never opened before test submission.
question_id,**/pretest_responses.csv,Public assessment-item identifier using P/Q form prefixes. Join to question_id in the matching Quant or Verbal question bank; the same IDs are reused across instruments.
question_open_event_received_server_elapsed_seconds,**/pretest_responses.csv,Question open event received at server for one pre-test or post-test item response. Measured in seconds from the shared origin described in README.
question_opened_client_elapsed_seconds,**/pretest_responses.csv,Question opened at client for one pre-test or post-test item response. Measured in seconds from the shared origin described in README.
response_recorded_server_elapsed_seconds,**/pretest_responses.csv,Response recorded at server for one pre-test or post-test item response. Measured in seconds from the shared origin described in README.
skill,**/pretest_responses.csv,Top-level GRE skill category assigned to the item.
skipped,**/pretest_responses.csv,Whether the item was submitted without a student answer.
student_answer,**/pretest_responses.csv,Student answer recorded for the assessment item; blank represents unanswered when applicable.
time_spent_seconds,**/pretest_responses.csv,Client-accumulated item-view time in seconds; it is not a complete wall-clock trace and may be zero for event-log recoveries.
ai_model_api_provider,**/prompt_analysis_sessions.csv,"API-provider family used for the AI tutoring condition, such as anthropic, gemini, openai, openai-responses, or openrouter."
ai_model_endpoint_id,**/prompt_analysis_sessions.csv,Provider-facing model or endpoint identifier used for calls; it does not encode the full experimental reasoning preset.
ai_model_preset_id,**/prompt_analysis_sessions.csv,"Experimental AI condition identifier combining the model family with the configured reasoning-effort or mode, for example opus-4.8-off versus opus-4.8-xhigh."
ai_tutor_completion_tokens,**/prompt_analysis_sessions.csv,Canonical field for ai tutor completion tokens.
ai_tutor_message_count,**/prompt_analysis_sessions.csv,Number of ai tutor message represented by this record.
ai_tutor_reasoning_tokens,**/prompt_analysis_sessions.csv,Canonical field for ai tutor reasoning tokens.
assessment_bank_version_id,**/prompt_analysis_sessions.csv,Immutable identifier and content fingerprint for the assessment bank used by the student.
assessment_form_order,**/prompt_analysis_sessions.csv,Randomized PQ or QP assessment-form counterbalancing assignment.
assigned_intervention_ended_elapsed_seconds,**/prompt_analysis_sessions.csv,Recorded end marker for the assigned intervention phase. It does not necessarily mark the end of active tutoring engagement. Measured in seconds from the shared origin described in README.
assigned_intervention_started_elapsed_seconds,**/prompt_analysis_sessions.csv,"Recorded start marker for the assigned AI, control, or human intervention phase. Measured in seconds from the shared origin described in README."
assigned_intervention_wall_clock_minutes,**/prompt_analysis_sessions.csv,"Assigned intervention wall clock, measured in minutes."
assigned_intervention_wall_clock_seconds,**/prompt_analysis_sessions.csv,"Elapsed wall-clock time between intervention phase markers; this is not active tutoring time, video-watch time, or on-task engagement."
completed_lesson_plan_concept_count,**/prompt_analysis_sessions.csv,Number of completed lesson plan concept.
extra_practice_problem_count,**/prompt_analysis_sessions.csv,Number of extra practice problem.
feedback_confusing,**/prompt_analysis_sessions.csv,Feedback confusing recorded for this observation; interpret it within the file and parent-record context shown by the example path.
feedback_helpful,**/prompt_analysis_sessions.csv,Feedback helpful recorded for this observation; interpret it within the file and parent-record context shown by the example path.
feedback_other,**/prompt_analysis_sessions.csv,Feedback other recorded for this observation; interpret it within the file and parent-record context shown by the example path.
feedback_rating,**/prompt_analysis_sessions.csv,Feedback rating recorded for this observation.
first_attempt_student_id,**/prompt_analysis_sessions.csv,Stable identifier for first attempt student in this record.
lesson_plan_concept_count,**/prompt_analysis_sessions.csv,Number of lesson plan concept.
lesson_plan_generation_completion_tokens,**/prompt_analysis_sessions.csv,Canonical field for lesson plan generation completion tokens.
lesson_plan_generation_fallback_used,**/prompt_analysis_sessions.csv,Canonical field for lesson plan generation fallback used.
lesson_plan_generation_prompt_tokens,**/prompt_analysis_sessions.csv,Canonical field for lesson plan generation prompt tokens.
lesson_plan_generation_provider_latency_seconds,**/prompt_analysis_sessions.csv,"Lesson plan generation provider latency, measured in seconds."
lesson_plan_generation_reasoning_tokens,**/prompt_analysis_sessions.csv,Canonical field for lesson plan generation reasoning tokens.
lesson_plan_generator_ai_model_api_provider,**/prompt_analysis_sessions.csv,Canonical field for lesson plan generator ai model api provider.
lesson_plan_generator_ai_model_endpoint_id,**/prompt_analysis_sessions.csv,Stable identifier for lesson plan generator ai model endpoint.
lesson_plan_prompt_version_id,**/prompt_analysis_sessions.csv,Exact prompt-version identifier used to generate the AI lesson plan.
lesson_plan_validation_status,**/prompt_analysis_sessions.csv,Canonical field for lesson plan validation status.
lesson_plan_wait_seconds,**/prompt_analysis_sessions.csv,"Lesson plan wait, measured in seconds."
posttest_attempted_item_count,**/prompt_analysis_sessions.csv,Number of posttest attempted item.
posttest_ended_elapsed_seconds,**/prompt_analysis_sessions.csv,Server event time marking the end of the posttest phase. Measured in seconds from the shared origin described in README.
posttest_maximum_score_points,**/prompt_analysis_sessions.csv,Canonical field for posttest maximum score points.
posttest_score_points,**/prompt_analysis_sessions.csv,Number of posttest response rows scored correct under the deployed scoring contract.
posttest_started_elapsed_seconds,**/prompt_analysis_sessions.csv,Server event time marking the start of the posttest phase. Measured in seconds from the shared origin described in README.
posttest_wall_clock_seconds,**/prompt_analysis_sessions.csv,Elapsed wall-clock seconds from posttest start marker to posttest end marker; distinct from summed item-view time.
practice_problem_definitions_count,**/prompt_analysis_sessions.csv,"Total defined practice problems available in the retained primary lesson plan plus definitions created during tutoring, regardless of whether shown."
practice_problems_answered_count,**/prompt_analysis_sessions.csv,Number of distinct practice problems with was_answered=true.
practice_problems_final_correct_count,**/prompt_analysis_sessions.csv,Number of answered practice problems whose final recorded correctness is true.
practice_problems_skipped_count,**/prompt_analysis_sessions.csv,Number of practice problems skipped.
pretest_and_posttest_focus_lost_event_count,**/prompt_analysis_sessions.csv,"Count of recorded test_focus_lost events accumulated across both timed assessments; it is a focus-loss telemetry count, not proof of a literal browser-tab switch."
pretest_attempted_item_count,**/prompt_analysis_sessions.csv,Number of pretest attempted item.
pretest_ended_elapsed_seconds,**/prompt_analysis_sessions.csv,Server event time marking the end of the pretest phase. Measured in seconds from the shared origin described in README.
pretest_maximum_score_points,**/prompt_analysis_sessions.csv,Canonical field for pretest maximum score points.
pretest_score_points,**/prompt_analysis_sessions.csv,Number of pretest response rows scored correct under the deployed scoring contract.
pretest_started_elapsed_seconds,**/prompt_analysis_sessions.csv,Server event time marking the start of the pretest phase. Measured in seconds from the shared origin described in README.
pretest_wall_clock_seconds,**/prompt_analysis_sessions.csv,Elapsed wall-clock seconds from pretest start marker to pretest end marker; distinct from summed item-view time.
prompt_analysis_exclusion_reason,**/prompt_analysis_sessions.csv,Prompt analysis exclusion reason recorded for this observation; interpret it within the file and parent-record context shown by the example path.
prompt_analysis_included,**/prompt_analysis_sessions.csv,Prompt analysis included recorded for this observation; interpret it within the file and parent-record context shown by the example path.
prompt_condition_id,**/prompt_analysis_sessions.csv,Stable identifier for prompt condition.
rushed_submits,**/prompt_analysis_sessions.csv,Rushed submits recorded for this observation; interpret it within the file and parent-record context shown by the example path.
score_lift_points,**/prompt_analysis_sessions.csv,Raw change in correct-answer points from pretest to posttest.
screener_correct,**/prompt_analysis_sessions.csv,Screener correct recorded for this observation; interpret it within the file and parent-record context shown by the example path.
student_chat_message_count,**/prompt_analysis_sessions.csv,Number of student chat message represented by this record.
student_consented_to_research,**/prompt_analysis_sessions.csv,Whether the student gave research-participation consent.
student_id,**/prompt_analysis_sessions.csv,"Globally unique anonymous released study-session identifier, including instrument and release scope."
student_practice_answer_key_dispute_count,**/prompt_analysis_sessions.csv,Number of student_disagreed events indicating the student clicked that they disagreed with a practice answer key.
student_survey_education_major,**/prompt_analysis_sessions.csv,Student's selected or normalized education-major category; it is not necessarily an undergraduate major.
student_survey_has_taken_gre_before,**/prompt_analysis_sessions.csv,Canonical field for student survey has taken gre before.
student_survey_prior_tutoring_hours_category,**/prompt_analysis_sessions.csv,"Student's selected category for prior tutoring exposure: none, fewer than 10 hours, 10 to 50 hours, or more than 50 hours."
student_survey_quant_comfort_areas,**/prompt_analysis_sessions.csv,"Student-selected Quant areas in response to “Which of these are you comfortable with?”; six-area multi-select, with none selected allowed. Collected during intake in both Quant and Verbal sessions during this release; these are Quant comfort areas, not a Verbal-skill measure. An empty selection is an observed response; an absent key means no value was retained at that location. Events may contain responses absent from the final profile or earlier different submissions. Allowed values: arithmetic, algebra, geometry, trigonometry, data_analysis, word_problems."
study_participation_attempt_number,**/prompt_analysis_sessions.csv,Source participation-attempt ordinal for the account; it is not itself the final same-instrument repeat-adjudication result.
study_session_record_created_elapsed_seconds,**/prompt_analysis_sessions.csv,"Study-session record creation, in seconds from the shared session origin; zero in sessions with a recorded creation time. This is distinct from pretest start."
study_total_wall_seconds,**/prompt_analysis_sessions.csv,"Study total wall, measured in seconds."
total_message_count,**/prompt_analysis_sessions.csv,Number of total message.
tutor_prompt_version_id,**/prompt_analysis_sessions.csv,Exact prompt-version identifier used by the interactive AI tutor.
direction,**/rating_scales.csv,"Direction stored in the prepared lesson plan, its generation metadata, or its usage record."
label,**/rating_scales.csv,"Label stored in the prepared lesson plan, its generation metadata, or its usage record."
notes,**/rating_scales.csv,"Notes stored in the prepared lesson plan, its generation metadata, or its usage record."
scale_id,**/rating_scales.csv,"Scale id stored in the prepared lesson plan, its generation metadata, or its usage record."
score,**/rating_scales.csv,"Score stored in the prepared lesson plan, its generation metadata, or its usage record."
review_subject_id,**/repeat_participant_review_subjects.csv,Existing anonymous expert-review context excluded in the repeat-participant sensitivity. Membership only; no person identity or paired-session link.
student_id,**/repeat_participant_sessions.csv,Existing anonymous main-study session excluded in the repeat-participant sensitivity. Membership only; no person identity or paired-session link.
answer_changes,**/review_subject_pretest_responses.csv,Number of recorded answer changes before the assessment response was finalized.
correct_answer,**/review_subject_pretest_responses.csv,Answer key stored with the practice problem.
difficulty,**/review_subject_pretest_responses.csv,Recorded difficulty label for the assessment item or practice problem.
format,**/review_subject_pretest_responses.csv,Recorded assessment or practice-problem response format.
instrument_id,**/review_subject_pretest_responses.csv,Assessment instrument identifier: gre_quant or gre_verbal.
is_correct,**/review_subject_pretest_responses.csv,Whether the recorded submitted answer matches the deployed answer key under the applicable scoring contract.
linked_ssot_student_id,**/review_subject_pretest_responses.csv,Anonymous public study-session identifier linking this review subject to released student-session data when that link is available. Blank means no such link is recorded; review_subject_id remains the review-context key.
opened_elapsed_seconds,**/review_subject_pretest_responses.csv,"Recorded event time when this assessment item was opened; this describes the student pretest response, not the lesson-plan review. Measured in seconds from the shared origin described in README."
options_json,**/review_subject_pretest_responses.csv,Strict JSON object containing separately recorded answer choices; an empty object means choices were not stored separately.
presented_position,**/review_subject_pretest_responses.csv,One-based position at which the item was presented; blank means the item was never opened before test submission.
question_text,**/review_subject_pretest_responses.csv,"Question text stored in the prepared lesson plan, its generation metadata, or its usage record."
review_subject_id,**/review_subject_pretest_responses.csv,Anonymous identifier for the student context used in expert review. Joins this subject's pretest responses to human_grading_sessions.csv within the same study and instrument; it is distinct from student_id.
skill,**/review_subject_pretest_responses.csv,Top-level GRE skill category assigned to the item.
source_question_id,**/review_subject_pretest_responses.csv,"Historical application question identifier. Join to source_question_id in the matching instrument question bank (M/N for Quant; Mv/Nv for Verbal), not to its public P/Q question_id."
student_answer,**/review_subject_pretest_responses.csv,Student answer recorded for the assessment item; blank represents unanswered when applicable.
student_id,**/review_subject_pretest_responses.csv,Anonymous public study-session identifier linking this review subject to released student-session data when that link is available. Blank means no such link is recorded; review_subject_id remains the review-context key.
submitted_elapsed_seconds,**/review_subject_pretest_responses.csv,"Recorded submission event time for this student pretest response; this describes the assessment item, not submission of the expert lesson-plan review. Measured in seconds from the shared origin described in README."
subskill,**/review_subject_pretest_responses.csv,Subskill in the immutable assessment question bank.
time_spent_seconds,**/review_subject_pretest_responses.csv,Client-accumulated item-view time in seconds; it is not a complete wall-clock trace and may be zero for event-log recoveries.
note,**/reviewed_lesson_plans.json,"Note stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].ai_model_api_provider,**/reviewed_lesson_plans.json,"API-provider family used for the AI tutoring condition, such as anthropic, gemini, openai, openai-responses, or openrouter."
plans[].ai_model_endpoint_id,**/reviewed_lesson_plans.json,Provider-facing model or endpoint identifier used for calls; it does not encode the full experimental reasoning preset.
plans[].ai_model_preset_id,**/reviewed_lesson_plans.json,"Experimental AI condition identifier combining the model family with the configured reasoning-effort or mode, for example opus-4.8-off versus opus-4.8-xhigh."
plans[].ai_model_reasoning_effort,**/reviewed_lesson_plans.json,Canonical field for ai model reasoning effort.
plans[].completion_tokens,**/reviewed_lesson_plans.json,"Completion tokens stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].created_elapsed_seconds,**/reviewed_lesson_plans.json,"Lesson-plan creation time, in seconds from the reviewed student reference origin described in README."
plans[].finish_reason,**/reviewed_lesson_plans.json,Provider-reported reason the model response ended.
plans[].generation_latency_ms,**/reviewed_lesson_plans.json,"Generation latency ms stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].human_lesson_plan_created_during,**/reviewed_lesson_plans.json,"Human lesson plan created during stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].human_lesson_plan_creation_duration_seconds,**/reviewed_lesson_plans.json,"Human lesson plan creation duration seconds stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].human_plan_author_anon_id,**/reviewed_lesson_plans.json,"Human plan author anon id stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].lesson_plan_id,**/reviewed_lesson_plans.json,"Lesson plan id stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].lesson_plan_kind,**/reviewed_lesson_plans.json,"Lesson plan kind stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_shown_in_ai_vs_human_review.concepts[].allotted_minutes,**/reviewed_lesson_plans.json,"Allotted minutes stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_shown_in_ai_vs_human_review.concepts[].concept_id,**/reviewed_lesson_plans.json,"Concept id stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_shown_in_ai_vs_human_review.concepts[].name,**/reviewed_lesson_plans.json,"Name stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_shown_in_ai_vs_human_review.concepts[].practice_problems[],**/reviewed_lesson_plans.json,"Practice problems stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_shown_in_ai_vs_human_review.concepts[].practice_problems[].answer,**/reviewed_lesson_plans.json,Answer key stored with the practice problem.
plans[].plan_as_shown_in_ai_vs_human_review.concepts[].practice_problems[].difficulty,**/reviewed_lesson_plans.json,Recorded difficulty label for the assessment item or practice problem.
plans[].plan_as_shown_in_ai_vs_human_review.concepts[].practice_problems[].format,**/reviewed_lesson_plans.json,Recorded assessment or practice-problem response format.
plans[].plan_as_shown_in_ai_vs_human_review.concepts[].practice_problems[].options,**/reviewed_lesson_plans.json,Answer-choice map for the question; keys vary by item and should be treated as choice labels rather than a fixed A-D schema.
plans[].plan_as_shown_in_ai_vs_human_review.concepts[].practice_problems[].options.<choice_key>,**/reviewed_lesson_plans.json,Answer-choice map for the question; keys vary by item and should be treated as choice labels rather than a fixed A-D schema.
plans[].plan_as_shown_in_ai_vs_human_review.concepts[].practice_problems[].solution,**/reviewed_lesson_plans.json,Worked solution stored with the practice problem or lesson plan.
plans[].plan_as_shown_in_ai_vs_human_review.concepts[].practice_problems[].stem,**/reviewed_lesson_plans.json,Full question or practice-problem stem.
plans[].plan_as_shown_in_ai_vs_human_review.concepts[].pretest_evidence[],**/reviewed_lesson_plans.json,"Pretest evidence stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_shown_in_ai_vs_human_review.concepts[].priority_rank,**/reviewed_lesson_plans.json,"Priority rank stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_shown_in_ai_vs_human_review.student_summary,**/reviewed_lesson_plans.json,"Student summary stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_shown_in_ai_vs_human_review.total_minutes,**/reviewed_lesson_plans.json,"Total minutes stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_stored.concepts[].allotted_minutes,**/reviewed_lesson_plans.json,"Allotted minutes stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_stored.concepts[].concept_id,**/reviewed_lesson_plans.json,"Concept id stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_stored.concepts[].misconception,**/reviewed_lesson_plans.json,Recorded description of a student misconception addressed by the lesson-plan concept; preserve it as source-authored material rather than a verified diagnosis.
plans[].plan_as_stored.concepts[].name,**/reviewed_lesson_plans.json,"Name stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_stored.concepts[].practice_problems[],**/reviewed_lesson_plans.json,"Practice problems stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_stored.concepts[].practice_problems[].answer,**/reviewed_lesson_plans.json,Answer key stored with the practice problem.
plans[].plan_as_stored.concepts[].practice_problems[].difficulty,**/reviewed_lesson_plans.json,Recorded difficulty label for the assessment item or practice problem.
plans[].plan_as_stored.concepts[].practice_problems[].format,**/reviewed_lesson_plans.json,Recorded assessment or practice-problem response format.
plans[].plan_as_stored.concepts[].practice_problems[].options,**/reviewed_lesson_plans.json,Answer-choice map for the question; keys vary by item and should be treated as choice labels rather than a fixed A-D schema.
plans[].plan_as_stored.concepts[].practice_problems[].options.<choice_key>,**/reviewed_lesson_plans.json,Answer-choice map for the question; keys vary by item and should be treated as choice labels rather than a fixed A-D schema.
plans[].plan_as_stored.concepts[].practice_problems[].solution,**/reviewed_lesson_plans.json,Worked solution stored with the practice problem or lesson plan.
plans[].plan_as_stored.concepts[].practice_problems[].stem,**/reviewed_lesson_plans.json,Full question or practice-problem stem.
plans[].plan_as_stored.concepts[].pretest_evidence[],**/reviewed_lesson_plans.json,"Pretest evidence stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_stored.concepts[].priority_rank,**/reviewed_lesson_plans.json,"Priority rank stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_stored.concepts[].why_it_matters,**/reviewed_lesson_plans.json,"Why it matters stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_stored.student_summary,**/reviewed_lesson_plans.json,"Student summary stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_stored.summary,**/reviewed_lesson_plans.json,"Summary stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].plan_as_stored.total_minutes,**/reviewed_lesson_plans.json,"Total minutes stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].prompt_tokens,**/reviewed_lesson_plans.json,"Prompt tokens stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].reasoning_tokens,**/reviewed_lesson_plans.json,"Provider-reported reasoning-token usage when exposed separately; null means unavailable, not zero."
plans[].review_subject_ids[],**/reviewed_lesson_plans.json,"Review subject ids stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].section_ids[],**/reviewed_lesson_plans.json,"Section ids stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].source_review_subject_id,**/reviewed_lesson_plans.json,"Source review subject id stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].submitted_elapsed_seconds,**/reviewed_lesson_plans.json,"Lesson-plan submission time, in seconds from the reviewed student reference origin described in README."
plans[].used_in_comparison_types[],**/reviewed_lesson_plans.json,"Used in comparison types stored in the prepared lesson plan, its generation metadata, or its usage record."
plans[].validation_status,**/reviewed_lesson_plans.json,"Parser, required-field, schema, and configured leakage-check result for the plan artifact."
schema_version,**/reviewed_lesson_plans.json,"Schema version stored in the prepared lesson plan, its generation metadata, or its usage record."
comparison_id,**/rubric_grades.csv,Comparison id recorded for one rubric question within a completed human lesson-plan comparison.
comparison_type,**/rubric_grades.csv,Comparison type recorded for one rubric question within a completed human lesson-plan comparison.
grading_id,**/rubric_grades.csv,Grading id recorded for one rubric question within a completed human lesson-plan comparison.
instrument_id,**/rubric_grades.csv,Assessment instrument identifier: gre_quant or gre_verbal.
lesson_plan_a_rating,**/rubric_grades.csv,Lesson plan a rating recorded for one rubric question within a completed human lesson-plan comparison.
lesson_plan_a_rating_label,**/rubric_grades.csv,Lesson plan a rating label recorded for one rubric question within a completed human lesson-plan comparison.
lesson_plan_b_rating,**/rubric_grades.csv,Lesson plan b rating recorded for one rubric question within a completed human lesson-plan comparison.
lesson_plan_b_rating_label,**/rubric_grades.csv,Lesson plan b rating label recorded for one rubric question within a completed human lesson-plan comparison.
preference_key,**/rubric_grades.csv,Preference key recorded for one rubric question within a completed human lesson-plan comparison.
preference_label,**/rubric_grades.csv,Preference label recorded for one rubric question within a completed human lesson-plan comparison.
preference_score,**/rubric_grades.csv,"Pairwise lesson-plan preference: positive favors plan A, negative favors plan B, and zero is a tie."
question_position,**/rubric_grades.csv,Question position recorded for one rubric question within a completed human lesson-plan comparison.
question_variant_id,**/rubric_grades.csv,Question variant id recorded for one rubric question within a completed human lesson-plan comparison.
reviewer_reasoning,**/rubric_grades.csv,Reviewer reasoning recorded for one rubric question within a completed human lesson-plan comparison.
rubric_question_id,**/rubric_grades.csv,Rubric question id recorded for one rubric question within a completed human lesson-plan comparison.
rubric_version,**/rubric_grades.csv,Rubric version recorded for one rubric question within a completed human lesson-plan comparison.
source_rubric_question_id,**/rubric_grades.csv,Source rubric question id recorded for one rubric question within a completed human lesson-plan comparison.
comparison_type,**/rubric_questions.csv,Comparison type recorded for one completed paid human lesson-plan comparison.
instrument_id,**/rubric_questions.csv,Assessment instrument identifier: gre_quant or gre_verbal.
notes,**/rubric_questions.csv,"Notes stored in the prepared lesson plan, its generation metadata, or its usage record."
observed_grades,**/rubric_questions.csv,Number of released rubric-grade rows associated with this rubric variant. Stored in the supplementary human-versus-AI review data; main-study counts can be derived from rubric_grades.csv.
plan_rating_scale,**/rubric_questions.csv,"Plan rating scale stored in the prepared lesson plan, its generation metadata, or its usage record."
preference_scale_id,**/rubric_questions.csv,"Preference scale id stored in the prepared lesson plan, its generation metadata, or its usage record."
question_label,**/rubric_questions.csv,"Question label stored in the prepared lesson plan, its generation metadata, or its usage record."
question_position,**/rubric_questions.csv,"Question position stored in the prepared lesson plan, its generation metadata, or its usage record."
question_text,**/rubric_questions.csv,"Question text stored in the prepared lesson plan, its generation metadata, or its usage record."
question_variant_id,**/rubric_questions.csv,"Question variant id stored in the prepared lesson plan, its generation metadata, or its usage record."
rubric_question_id,**/rubric_questions.csv,"Rubric question id stored in the prepared lesson plan, its generation metadata, or its usage record."
rubric_version,**/rubric_questions.csv,Rubric version recorded for one completed paid human lesson-plan comparison.
source_rubric_question_id,**/rubric_questions.csv,"Source rubric question id stored in the prepared lesson plan, its generation metadata, or its usage record."
denominator,**/section_participant_overlap.csv,Denominator in the privacy-safe aggregate population table.
metric,**/section_participant_overlap.csv,Metric in the privacy-safe aggregate population table.
n,**/section_participant_overlap.csv,N in the privacy-safe aggregate population table.
percent,**/section_participant_overlap.csv,Percent in the privacy-safe aggregate population table.
human_tutoring_transcript_available,**/session_metadata.json,Whether at least one human-tutoring transcript source is retained for this study session.
human_tutoring_transcript_file_count,**/session_metadata.json,Number of retained transcript files represented by the sessions list.
human_tutoring_transcript_source_description,**/session_metadata.json,Historical export note describing the transcript collection used for release. It may contain an inherited or stale instrument label; use the containing cohort/instrument folder and structured session/profile fields as authoritative. Preserve this original provenance text rather than rewriting its label.
human_tutoring_transcript_unavailable_reason,**/session_metadata.json,Recorded reason no human-tutoring transcript source was available for this study session.
sessions[],**/session_metadata.json,One retained human-tutoring recording/transcript session represented in this study session's metadata.
sessions[].human_tutoring_session_started_elapsed_seconds,**/session_metadata.json,Recorded start event time for the human-tutoring session associated with this transcript. Measured in seconds from the shared origin described in README.
sessions[].transcript_filename,**/session_metadata.json,Name of the associated transcript text file in the same folder as session_metadata.json.
sessions[].tutor_anon_id,**/session_metadata.json,Anonymous tutor identifier recorded for this transcript session; preserve its source namespace and do not assume it equals an identifier from another collection.
ai_model_api_provider,**/sessions_excluded_from_prompt_analysis.csv,"API-provider family used for the AI tutoring condition, such as anthropic, gemini, openai, openai-responses, or openrouter."
ai_model_endpoint_id,**/sessions_excluded_from_prompt_analysis.csv,Provider-facing model or endpoint identifier used for calls; it does not encode the full experimental reasoning preset.
ai_model_preset_id,**/sessions_excluded_from_prompt_analysis.csv,"Experimental AI condition identifier combining the model family with the configured reasoning-effort or mode, for example opus-4.8-off versus opus-4.8-xhigh."
ai_tutor_completion_tokens,**/sessions_excluded_from_prompt_analysis.csv,Canonical field for ai tutor completion tokens.
ai_tutor_message_count,**/sessions_excluded_from_prompt_analysis.csv,Number of ai tutor message represented by this record.
ai_tutor_reasoning_tokens,**/sessions_excluded_from_prompt_analysis.csv,Canonical field for ai tutor reasoning tokens.
assessment_bank_version_id,**/sessions_excluded_from_prompt_analysis.csv,Immutable identifier and content fingerprint for the assessment bank used by the student.
assessment_form_order,**/sessions_excluded_from_prompt_analysis.csv,Randomized PQ or QP assessment-form counterbalancing assignment.
assigned_intervention_ended_elapsed_seconds,**/sessions_excluded_from_prompt_analysis.csv,Recorded end marker for the assigned intervention phase. It does not necessarily mark the end of active tutoring engagement. Measured in seconds from the shared origin described in README.
assigned_intervention_started_elapsed_seconds,**/sessions_excluded_from_prompt_analysis.csv,"Recorded start marker for the assigned AI, control, or human intervention phase. Measured in seconds from the shared origin described in README."
assigned_intervention_wall_clock_minutes,**/sessions_excluded_from_prompt_analysis.csv,"Assigned intervention wall clock, measured in minutes."
assigned_intervention_wall_clock_seconds,**/sessions_excluded_from_prompt_analysis.csv,"Elapsed wall-clock time between intervention phase markers; this is not active tutoring time, video-watch time, or on-task engagement."
completed_lesson_plan_concept_count,**/sessions_excluded_from_prompt_analysis.csv,Number of completed lesson plan concept.
extra_practice_problem_count,**/sessions_excluded_from_prompt_analysis.csv,Number of extra practice problem.
feedback_confusing,**/sessions_excluded_from_prompt_analysis.csv,Feedback confusing recorded for this observation; interpret it within the file and parent-record context shown by the example path.
feedback_helpful,**/sessions_excluded_from_prompt_analysis.csv,Feedback helpful recorded for this observation; interpret it within the file and parent-record context shown by the example path.
feedback_other,**/sessions_excluded_from_prompt_analysis.csv,Feedback other recorded for this observation; interpret it within the file and parent-record context shown by the example path.
feedback_rating,**/sessions_excluded_from_prompt_analysis.csv,Feedback rating recorded for this observation.
first_attempt_student_id,**/sessions_excluded_from_prompt_analysis.csv,Stable identifier for first attempt student in this record.
is_admin,**/sessions_excluded_from_prompt_analysis.csv,Whether admin is true for this record.
lesson_plan_concept_count,**/sessions_excluded_from_prompt_analysis.csv,Number of lesson plan concept.
lesson_plan_generation_completion_tokens,**/sessions_excluded_from_prompt_analysis.csv,Canonical field for lesson plan generation completion tokens.
lesson_plan_generation_fallback_used,**/sessions_excluded_from_prompt_analysis.csv,Canonical field for lesson plan generation fallback used.
lesson_plan_generation_prompt_tokens,**/sessions_excluded_from_prompt_analysis.csv,Canonical field for lesson plan generation prompt tokens.
lesson_plan_generation_provider_latency_seconds,**/sessions_excluded_from_prompt_analysis.csv,"Lesson plan generation provider latency, measured in seconds."
lesson_plan_generation_reasoning_tokens,**/sessions_excluded_from_prompt_analysis.csv,Canonical field for lesson plan generation reasoning tokens.
lesson_plan_generation_status,**/sessions_excluded_from_prompt_analysis.csv,Canonical field for lesson plan generation status.
lesson_plan_generator_ai_model_api_provider,**/sessions_excluded_from_prompt_analysis.csv,Canonical field for lesson plan generator ai model api provider.
lesson_plan_generator_ai_model_endpoint_id,**/sessions_excluded_from_prompt_analysis.csv,Stable identifier for lesson plan generator ai model endpoint.
lesson_plan_prompt_version_id,**/sessions_excluded_from_prompt_analysis.csv,Exact prompt-version identifier used to generate the AI lesson plan.
lesson_plan_validation_status,**/sessions_excluded_from_prompt_analysis.csv,Canonical field for lesson plan validation status.
lesson_plan_wait_seconds,**/sessions_excluded_from_prompt_analysis.csv,"Lesson plan wait, measured in seconds."
posttest_attempted_item_count,**/sessions_excluded_from_prompt_analysis.csv,Number of posttest attempted item.
posttest_ended_elapsed_seconds,**/sessions_excluded_from_prompt_analysis.csv,Server event time marking the end of the posttest phase. Measured in seconds from the shared origin described in README.
posttest_maximum_score_points,**/sessions_excluded_from_prompt_analysis.csv,Canonical field for posttest maximum score points.
posttest_score_points,**/sessions_excluded_from_prompt_analysis.csv,Number of posttest response rows scored correct under the deployed scoring contract.
posttest_started_elapsed_seconds,**/sessions_excluded_from_prompt_analysis.csv,Server event time marking the start of the posttest phase. Measured in seconds from the shared origin described in README.
posttest_wall_clock_seconds,**/sessions_excluded_from_prompt_analysis.csv,Elapsed wall-clock seconds from posttest start marker to posttest end marker; distinct from summed item-view time.
practice_problem_definitions_count,**/sessions_excluded_from_prompt_analysis.csv,"Total defined practice problems available in the retained primary lesson plan plus definitions created during tutoring, regardless of whether shown."
practice_problems_answered_count,**/sessions_excluded_from_prompt_analysis.csv,Number of distinct practice problems with was_answered=true.
practice_problems_final_correct_count,**/sessions_excluded_from_prompt_analysis.csv,Number of answered practice problems whose final recorded correctness is true.
practice_problems_skipped_count,**/sessions_excluded_from_prompt_analysis.csv,Number of practice problems skipped.
pretest_and_posttest_focus_lost_event_count,**/sessions_excluded_from_prompt_analysis.csv,"Count of recorded test_focus_lost events accumulated across both timed assessments; it is a focus-loss telemetry count, not proof of a literal browser-tab switch."
pretest_attempted_item_count,**/sessions_excluded_from_prompt_analysis.csv,Number of pretest attempted item.
pretest_ended_elapsed_seconds,**/sessions_excluded_from_prompt_analysis.csv,Server event time marking the end of the pretest phase. Measured in seconds from the shared origin described in README.
pretest_maximum_score_points,**/sessions_excluded_from_prompt_analysis.csv,Canonical field for pretest maximum score points.
pretest_score_points,**/sessions_excluded_from_prompt_analysis.csv,Number of pretest response rows scored correct under the deployed scoring contract.
pretest_started_elapsed_seconds,**/sessions_excluded_from_prompt_analysis.csv,Server event time marking the start of the pretest phase. Measured in seconds from the shared origin described in README.
pretest_wall_clock_seconds,**/sessions_excluded_from_prompt_analysis.csv,Elapsed wall-clock seconds from pretest start marker to pretest end marker; distinct from summed item-view time.
prompt_analysis_exclusion_reason,**/sessions_excluded_from_prompt_analysis.csv,Prompt analysis exclusion reason recorded for this observation; interpret it within the file and parent-record context shown by the example path.
prompt_analysis_included,**/sessions_excluded_from_prompt_analysis.csv,Prompt analysis included recorded for this observation; interpret it within the file and parent-record context shown by the example path.
prompt_condition_id,**/sessions_excluded_from_prompt_analysis.csv,Stable identifier for prompt condition.
rushed_submits,**/sessions_excluded_from_prompt_analysis.csv,Rushed submits recorded for this observation; interpret it within the file and parent-record context shown by the example path.
score_lift_points,**/sessions_excluded_from_prompt_analysis.csv,Raw change in correct-answer points from pretest to posttest.
screener_correct,**/sessions_excluded_from_prompt_analysis.csv,Screener correct recorded for this observation; interpret it within the file and parent-record context shown by the example path.
student_chat_message_count,**/sessions_excluded_from_prompt_analysis.csv,Number of student chat message represented by this record.
student_consented_to_research,**/sessions_excluded_from_prompt_analysis.csv,Whether the student gave research-participation consent.
student_id,**/sessions_excluded_from_prompt_analysis.csv,"Globally unique anonymous released study-session identifier, including instrument and release scope."
student_practice_answer_key_dispute_count,**/sessions_excluded_from_prompt_analysis.csv,Number of student_disagreed events indicating the student clicked that they disagreed with a practice answer key.
student_survey_education_major,**/sessions_excluded_from_prompt_analysis.csv,Student's selected or normalized education-major category; it is not necessarily an undergraduate major.
student_survey_has_taken_gre_before,**/sessions_excluded_from_prompt_analysis.csv,Canonical field for student survey has taken gre before.
student_survey_prior_tutoring_hours_category,**/sessions_excluded_from_prompt_analysis.csv,"Student's selected category for prior tutoring exposure: none, fewer than 10 hours, 10 to 50 hours, or more than 50 hours."
student_survey_quant_comfort_areas,**/sessions_excluded_from_prompt_analysis.csv,"Student-selected Quant areas in response to “Which of these are you comfortable with?”; six-area multi-select, with none selected allowed. Collected during intake in both Quant and Verbal sessions during this release; these are Quant comfort areas, not a Verbal-skill measure. An empty selection is an observed response; an absent key means no value was retained at that location. Events may contain responses absent from the final profile or earlier different submissions. Allowed values: arithmetic, algebra, geometry, trigonometry, data_analysis, word_problems."
study_participation_attempt_number,**/sessions_excluded_from_prompt_analysis.csv,Source participation-attempt ordinal for the account; it is not itself the final same-instrument repeat-adjudication result.
study_session_record_created_elapsed_seconds,**/sessions_excluded_from_prompt_analysis.csv,"Study-session record creation, in seconds from the shared session origin; zero in sessions with a recorded creation time. This is distinct from pretest start."
study_total_wall_seconds,**/sessions_excluded_from_prompt_analysis.csv,"Study total wall, measured in seconds."
total_message_count,**/sessions_excluded_from_prompt_analysis.csv,Number of total message.
tutor_prompt_version_id,**/sessions_excluded_from_prompt_analysis.csv,Exact prompt-version identifier used by the interactive AI tutor.
actor,**/student_ai_tutor_interaction_trajectory.csv,"Actor responsible for the event: student, AI tutor, system, or another explicitly labeled source."
ai_model_api_provider,**/student_ai_tutor_interaction_trajectory.csv,"API-provider family used for the AI tutoring condition, such as anthropic, gemini, openai, openai-responses, or openrouter."
ai_model_endpoint_id,**/student_ai_tutor_interaction_trajectory.csv,Provider-facing model or endpoint identifier used for calls; it does not encode the full experimental reasoning preset.
answer_attempt_number,**/student_ai_tutor_interaction_trajectory.csv,One-based ordinal of this submitted answer for the practice problem.
completion_tokens,**/student_ai_tutor_interaction_trajectory.csv,Completion tokens for one causally ordered tutoring interaction event.
concept_id,**/student_ai_tutor_interaction_trajectory.csv,Stable concept identifier joining to concepts[].concept_id in lesson_plan_including_practice_problems.json. A move to a new concept is represented directly by a change in the nonblank concept_id sequence; no redundant transition flag is stored. Trajectory events associated with a lesson-plan concept; blank when an event was not recorded against a concept.
content,**/student_ai_tutor_interaction_trajectory.csv,Recorded textual payload. Blank or whitespace-only student_message and ai_tutor_message rows remain in the raw trajectory but are excluded from the derived nonblank message counts in student.json. Events with textual content; blank is retained for raw event fidelity.
event_type,**/student_ai_tutor_interaction_trajectory.csv,Canonical event category describing what occurred at this trajectory or telemetry row.
finish_reason,**/student_ai_tutor_interaction_trajectory.csv,Provider-reported reason the model response ended.
is_correct,**/student_ai_tutor_interaction_trajectory.csv,Whether the recorded submitted answer matches the deployed answer key under the applicable scoring contract.
latency_ms,**/student_ai_tutor_interaction_trajectory.csv,Latency ms for one causally ordered tutoring interaction event.
lesson_plan_id,**/student_ai_tutor_interaction_trajectory.csv,Lesson plan id for one causally ordered tutoring interaction event.
message_id,**/student_ai_tutor_interaction_trajectory.csv,Recorded message identifier when a supplementary trajectory row corresponds to a stored chat message; blank on events without one. Interpret within its session.
practice_problem_id,**/student_ai_tutor_interaction_trajectory.csv,Practice problem id for one causally ordered tutoring interaction event.
problem_origin,**/student_ai_tutor_interaction_trajectory.csv,"Whether the practice problem came from the prepared lesson plan (lesson_plan) or was generated during tutoring (created_during_tutoring). Creation events may be buffered before display, so this field must be interpreted with event_type and trajectory_position rather than generation timestamps alone. Practice-problem events."
prompt_tokens,**/student_ai_tutor_interaction_trajectory.csv,Prompt tokens for one causally ordered tutoring interaction event.
reasoning_tokens,**/student_ai_tutor_interaction_trajectory.csv,"Provider-reported reasoning-token usage when exposed separately; null means unavailable, not zero."
recorded_elapsed_seconds,**/student_ai_tutor_interaction_trajectory.csv,"Event recording time, in seconds from the study-session origin. This can differ from timestamp_elapsed_seconds."
submitted_answer,**/student_ai_tutor_interaction_trajectory.csv,Student answer submitted at this individual practice-problem attempt.
timestamp_elapsed_seconds,**/student_ai_tutor_interaction_trajectory.csv,event time for one causally ordered tutoring interaction event. Measured in seconds from the shared origin described in README.
trajectory_event_id,**/student_ai_tutor_interaction_trajectory.csv,"Recorded event identifier in a supplementary tutoring trajectory. Treat it as an opaque event key, not a person identifier or a chronological position."
trajectory_position,**/student_ai_tutor_interaction_trajectory.csv,Authoritative one-based causal order of events in the tutoring interaction trajectory.
ai_model_preset_id,**/student_cost_summary.csv,"Experimental AI condition identifier combining the model family with the configured reasoning-effort or mode, for example opus-4.8-off versus opus-4.8-xhigh."
best_available_cost_method,**/student_cost_summary.csv,Best available cost method for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
best_available_total_cost_usd,**/student_cost_summary.csv,Best available total cost usd for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
completion_tokens,**/student_cost_summary.csv,Completion tokens for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
lesson_plan_calls,**/student_cost_summary.csv,Lesson plan calls for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
practice_generation_calls,**/student_cost_summary.csv,Practice generation calls for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
prompt_tokens,**/student_cost_summary.csv,Prompt tokens for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
reasoning_tokens,**/student_cost_summary.csv,"Provider-reported reasoning-token usage when exposed separately; null means unavailable, not zero."
recorded_cost_usd_sum,**/student_cost_summary.csv,Recorded cost usd sum for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
records_with_cache_accounting,**/student_cost_summary.csv,Records with cache accounting for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
records_with_recorded_cost,**/student_cost_summary.csv,Records with recorded cost for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
replay_estimated_total_cost_usd,**/student_cost_summary.csv,Replay estimated total cost usd for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
student_id,**/student_cost_summary.csv,"Globally unique anonymous released study-session identifier, including instrument and release scope."
total_cost_records,**/student_cost_summary.csv,Total cost records for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
tutor_calls,**/student_cost_summary.csv,Tutor calls for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
unmetered_practice_generation_calls,**/student_cost_summary.csv,Unmetered practice generation calls for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
unmetered_tutor_calls,**/student_cost_summary.csv,Unmetered tutor calls for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero.
ai_lesson_plan_available,**/student_data/student.json,Whether the expected primary AI lesson-plan artifact exists for an other-research AI record. Other-research AI profiles only; omitted from main because the artifact is required there.
ai_model_api_provider,**/student_data/student.json,"API-provider family used for the AI tutoring condition, such as anthropic, gemini, openai, openai-responses, or openrouter. AI profiles only; omitted for control and human arms."
ai_model_endpoint_id,**/student_data/student.json,Provider-facing model or endpoint identifier used for calls; it does not encode the full experimental reasoning preset. AI profiles only.
ai_model_preset_id,**/student_data/student.json,"Experimental AI condition identifier combining the model family with the configured reasoning-effort or mode, for example opus-4.8-off versus opus-4.8-xhigh. AI profiles only; this is the correct model-arm grouping variable."
ai_tutor_messages_sent_to_student,**/student_data/student.json,"Number of nonblank AI-tutor messages sent to the student in the canonical interaction trajectory; blank or whitespace-only rows are excluded. Present in profiles with structured AI interactions, including human-arm records with recorded AI exposure; omitted otherwise."
assessment_form_order,**/student_data/student.json,Randomized PQ or QP assessment-form counterbalancing assignment. All profiles; values are PQ or QP.
assessment_max_score_points,**/student_data/student.json,Maximum raw score for one released GRE section assessment. All profiles; observed value is 27.
assigned_intervention_ended_elapsed_seconds,**/student_data/student.json,Recorded end marker for the assigned intervention phase. It does not necessarily mark the end of active tutoring engagement. Present when the source recorded an intervention end. Measured in seconds from the shared origin described in README.
assigned_intervention_started_elapsed_seconds,**/student_data/student.json,"Recorded start marker for the assigned AI, control, or human intervention phase. Present when the source recorded an intervention start. Measured in seconds from the shared origin described in README."
assigned_intervention_wall_clock_seconds,**/student_data/student.json,"Elapsed wall-clock time between intervention phase markers; this is not active tutoring time, video-watch time, or on-task engagement. Present when a stored duration or both markers are available. Human values can include long between-session delays."
control_condition,**/student_data/student.json,Control-arm protocol variant with exact enum reviewed_pretest or did_not_review_pretest. Control profiles only. All 190 main controls are reviewed_pretest; other research contains 37 reviewed_pretest and 250 did_not_review_pretest.
control_pretest_to_video_start_elapsed_seconds,**/student_data/student.json,Elapsed seconds from the first server-recorded pretest submission to the first control-video session start. Control profiles when both events or a fallback measurement exist.
effort_pledge_recorded_elapsed_seconds,**/student_data/student.json,Server event time at which the student accepted the no-outside-help and best-effort integrity pledge. All staged profiles. Measured in seconds from the shared origin described in README.
experiment_run_id,**/student_data/student.json,Operational experiment-run identifier recorded at enrollment. All profiles; interpret jointly with instrument and arm.
human_lesson_plan_available,**/student_data/student.json,"Whether a participant-linked human lesson plan is available; false is an explicit source-availability state, not an imputation. Human profiles."
human_tutor_id,**/student_data/student.json,"Opaque anonymous identifier for the assigned human tutor, used for within-release grouping without exposing identity. Human profiles."
human_tutoring_transcript_available,**/student_data/student.json,Whether at least one retained human-tutoring transcript source is available. Human profiles.
human_tutoring_transcript_file_count,**/student_data/student.json,Number of retained session transcript files represented by human transcript metadata. Human profiles.
instrument_id,**/student_data/student.json,Assessment instrument identifier: gre_quant or gre_verbal. All profiles.
irt_theta_lift,**/student_data/student.json,Posttest minus pretest empirical-Bayes 2PL IRT ability on the within-instrument scale. Profiles with released IRT estimates. Quant and Verbal theta values are not cross-instrument comparable.
lesson_plan_prompt_version_id,**/student_data/student.json,Exact prompt-version identifier used to generate the AI lesson plan. AI profiles only; kept distinct from the interactive tutor prompt.
posttest_ended_elapsed_seconds,**/student_data/student.json,Server event time marking the end of the posttest phase. Profiles with a recorded posttest end. Measured in seconds from the shared origin described in README.
posttest_inter_submission_interval_count,**/student_data/student.json,Number of posttest inter-submission intervals underlying the corresponding median and p90 fields. A restart/resume task uses the sitting with the most distinct first submissions; the canonical posttest start breaks ties. All profiles; zero means no usable posttest first-submission intervals were recorded in the defined timing window.
posttest_irt_theta,**/student_data/student.json,Posttest empirical-Bayes 2PL IRT ability scored using fixed pretest item parameters within the instrument. Profiles with released IRT estimates; not comparable between Quant and Verbal.
posttest_median_inter_submission_seconds,**/student_data/student.json,Median seconds between consecutive first answer submissions to distinct posttest questions. It is not response-level time_spent_seconds and does not span the tutoring intervention. Profiles with at least one usable posttest inter-submission interval; omitted when the interval count is zero.
posttest_p90_inter_submission_seconds,**/student_data/student.json,90th-percentile seconds between consecutive first answer submissions to distinct posttest questions. The value is descriptive even when fewer than 13 intervals exist; protocol exclusion flags require at least 13. Profiles with at least one usable posttest inter-submission interval; omitted when the interval count is zero.
posttest_score_points,**/student_data/student.json,Number of posttest response rows scored correct under the deployed scoring contract. Profiles with a released posttest score.
posttest_started_elapsed_seconds,**/student_data/student.json,Server event time marking the start of the posttest phase. Profiles with a recorded posttest start. Measured in seconds from the shared origin described in README.
posttest_time_on_items_seconds,**/student_data/student.json,Sum of released response-level client-accumulated time_spent_seconds values for the posttest; distinct from wall-clock elapsed time. Profiles with a usable posttest response table or legacy item-time measurement.
posttest_wall_clock_seconds,**/student_data/student.json,Elapsed wall-clock seconds from posttest start marker to posttest end marker; distinct from summed item-view time. Profiles with a stored duration or both posttest markers.
posttutoring_survey_lesson_plan_helpfulness_rating,**/student_data/student.json,How helpful was the lesson plan (concepts covered) for your learning? Integer rating 1–5: 1 = Not at all helpful; 5 = Extremely helpful. Collected after tutoring and before the posttest. Event histories may differ from the final profile; neither source is overwritten.
posttutoring_survey_practice_problem_helpfulness_rating,**/student_data/student.json,How helpful were the practice problems the tutor created for your learning? Integer rating 1–5: 1 = Not at all helpful; 5 = Extremely helpful. Collected after tutoring and before the posttest. Event histories may differ from the final profile; neither source is overwritten.
posttutoring_survey_tutor_helpfulness_rating,**/student_data/student.json,Student rating from 1 to 5 for how helpful the tutor conversation was for learning. AI tutoring survey respondents; collected after tutoring and before the posttest.
practice_answer_submissions_count,**/student_data/student.json,"Total number of submitted practice answers, including repeated attempts on the same practice problem. Profiles with canonical AI practice outcomes."
practice_final_state_accuracy,**/student_data/student.json,Fraction of answered practice problems whose final recorded state is correct; not first-attempt accuracy. Profiles with at least one answered practice problem.
practice_problem_definitions_count,**/student_data/student.json,"Total defined practice problems available in the retained primary lesson plan plus definitions created during tutoring, regardless of whether shown. Profiles with retained AI lesson-plan or generated-definition artifacts."
practice_problems_answered_count,**/student_data/student.json,Number of distinct practice problems with was_answered=true. Profiles with canonical AI practice outcomes.
practice_problems_final_correct_count,**/student_data/student.json,Number of answered practice problems whose final recorded correctness is true. Profiles with canonical AI practice outcomes.
practice_problems_shown_count,**/student_data/student.json,"Number of distinct practice problems with was_shown=true; deliberately distinct from definitions, answered problems, and answer submissions. Profiles with canonical AI practice outcomes."
pretest_and_posttest_focus_lost_event_count,**/student_data/student.json,"Count of recorded test_focus_lost events accumulated across both timed assessments; it is a focus-loss telemetry count, not proof of a literal browser-tab switch. All profiles."
pretest_ended_elapsed_seconds,**/student_data/student.json,Server event time marking the end of the pretest phase. Profiles with a recorded pretest end. Measured in seconds from the shared origin described in README.
pretest_inter_submission_interval_count,**/student_data/student.json,Number of pretest inter-submission intervals underlying the corresponding median and p90 fields. A restart/resume task uses the sitting with the most distinct first submissions; the canonical pretest start breaks ties. All profiles; zero means no usable pretest first-submission intervals were recorded in the defined timing window.
pretest_irt_theta,**/student_data/student.json,Pretest empirical-Bayes ability from the instrument-specific 2PL IRT calibration. Profiles with released IRT estimates; not comparable between Quant and Verbal.
pretest_median_inter_submission_seconds,**/student_data/student.json,Median seconds between consecutive first answer submissions to distinct pretest questions. It is not response-level time_spent_seconds. Profiles with at least one usable pretest inter-submission interval; omitted when the interval count is zero.
pretest_p90_inter_submission_seconds,**/student_data/student.json,90th-percentile seconds between consecutive first answer submissions to distinct pretest questions. The value is descriptive even when fewer than 13 intervals exist; protocol exclusion flags require at least 13. Profiles with at least one usable pretest inter-submission interval; omitted when the interval count is zero.
pretest_score_points,**/student_data/student.json,Number of pretest response rows scored correct under the deployed scoring contract. Profiles with a released pretest score.
pretest_started_elapsed_seconds,**/student_data/student.json,Server event time marking the start of the pretest phase. All staged profiles. Measured in seconds from the shared origin described in README.
pretest_time_on_items_seconds,**/student_data/student.json,Sum of released response-level client-accumulated time_spent_seconds values for the pretest; distinct from wall-clock elapsed time. Profiles with a usable pretest response table or legacy item-time measurement.
pretest_wall_clock_seconds,**/student_data/student.json,Elapsed wall-clock seconds from pretest start marker to pretest end marker; distinct from summed item-view time. Profiles with a stored duration or both pretest markers.
primary_main_leaderboard_exclusion_reason,**/student_data/student.json,Plain-language primary reason the record is outside the final main leaderboard. All other-research profiles; forbidden in main profiles.
prior_same_instrument_study_exposure,**/student_data/student.json,Whether the account had a prior start on the same instrument under the repeat-participation audit. Other-research records where the source recorded this audit signal; omitted from physically pruned main profiles.
protocol_adherence_exclusion,**/student_data/student.json,Structured migration evidence for a record excluded under the final protocol-adherence rules. The 219 outcome-blind protocol-adherence exclusions in other research.
protocol_adherence_exclusion.all_reasons[],**/student_data/student.json,All reasons recorded for this observation; interpret it within the file and parent-record context shown by the example path.
protocol_adherence_exclusion.candidate_exclusions_sha256,**/student_data/student.json,SHA-256 provenance hash for candidate exclusions used to pin the exact source bytes.
protocol_adherence_exclusion.migration_version,**/student_data/student.json,Migration version recorded for this observation; interpret it within the file and parent-record context shown by the example path.
protocol_adherence_exclusion.outcome_or_lift_used,**/student_data/student.json,Outcome or lift used recorded for this observation; interpret it within the file and parent-record context shown by the example path.
protocol_adherence_exclusion.primary_reason,**/student_data/student.json,Primary reason recorded for this observation; interpret it within the file and parent-record context shown by the example path.
research_consent_recorded_elapsed_seconds,**/student_data/student.json,Server event time stored with the research-consent record. Present only where the source profile retained consent_ts; consent itself is separately recovered for all profiles. Measured in seconds from the shared origin described in README.
score_lift_points,**/student_data/student.json,Raw change in correct-answer points from pretest to posttest. Profiles with both scores.
student_agreed_to_no_outside_help_effort_pledge,**/student_data/student.json,"Whether the student accepted the integrity pledge to do their best and not use an LLM, AI tool, or outside help. All staged profiles; observed value is true."
student_consented_to_research,**/student_data/student.json,Whether the student gave research-participation consent. All staged public profiles; observed value is true.
student_id,**/student_data/student.json,"Globally unique anonymous released study-session identifier, including instrument and release scope. All profiles. The local task_id/anon_id aliases were removed; the folder name remains the local task key."
student_messages_sent_to_tutor,**/student_data/student.json,"Number of nonblank free-form messages sent by the student to the tutor; blank or whitespace-only submissions and practice-answer events are excluded. Profiles with released structured AI interaction data, including four adjudicated human-arm records with accidental AI exposure."
student_practice_answer_key_dispute_count,**/student_data/student.json,Number of student_disagreed events indicating the student clicked that they disagreed with a practice answer key. Profiles with AI practice data.
student_survey_education_level,**/student_data/student.json,"Student's selected current education-level category, including secondary-school, undergraduate, graduate, and not-in-school categories. Profiles with an intake response."
student_survey_education_major,**/student_data/student.json,Student's selected or normalized education-major category; it is not necessarily an undergraduate major. Profiles with an intake response.
student_survey_has_taken_sat,**/student_data/student.json,Recorded yes/no response to whether the student has taken the SAT. Profiles with an intake response; staged values remain strings 'yes' and 'no'.
student_survey_prior_tutoring_hours_category,**/student_data/student.json,"Student's selected category for prior tutoring exposure: none, fewer than 10 hours, 10 to 50 hours, or more than 50 hours. Profiles with an intake response; null was not converted to 'none'."
student_survey_quant_comfort_areas,**/student_data/student.json,"Student-selected Quant areas in response to “Which of these are you comfortable with?”; six-area multi-select, with none selected allowed. Collected during intake in both Quant and Verbal sessions during this release; these are Quant comfort areas, not a Verbal-skill measure. An empty selection is an observed response; an absent key means no value was retained at that location. Events may contain responses absent from the final profile or earlier different submissions. Allowed values: arithmetic, algebra, geometry, trigonometry, data_analysis, word_problems."
student_survey_quant_comfort_areas[],**/student_data/student.json,"One selected area. Student-selected Quant areas in response to “Which of these are you comfortable with?”; six-area multi-select, with none selected allowed. Collected during intake in both Quant and Verbal sessions during this release; these are Quant comfort areas, not a Verbal-skill measure. An empty selection is an observed response; an absent key means no value was retained at that location. Events may contain responses absent from the final profile or earlier different submissions. Allowed values: arithmetic, algebra, geometry, trigonometry, data_analysis, word_problems."
study_participation_attempt_number,**/student_data/student.json,Source participation-attempt ordinal for the account; it is not itself the final same-instrument repeat-adjudication result. All profiles.
study_session_completed,**/student_data/student.json,Whether the source study-session record was marked complete. Other-research profiles only; omitted from main because completion is a main-cohort invariant.
study_session_record_created_elapsed_seconds,**/student_data/student.json,"Study-session record creation, in seconds from the shared session origin; zero in sessions with a recorded creation time. This is distinct from pretest start."
supplementary_analysis_eligible,**/student_data/student.json,Whether an other-research record belongs to a specifically documented supplementary analysis cohort. Other-research profiles where the supplementary cohort defines this status; absence does not mean true.
tutor_prompt_version_id,**/student_data/student.json,Exact prompt-version identifier used by the interactive AI tutor. AI profiles only; kept distinct from the lesson-plan prompt version.
tutoring_arm,**/student_data/student.json,"Assigned tutoring modality: ai, control, or human. 'Human' refers to the tutor modality, not whether the student is human. All profiles."
unexpected_ai_lesson_plan_generated_but_not_delivered,**/student_data/student.json,Whether an AI lesson plan was generated unexpectedly for a human-arm record but was not delivered to the student. Human-arm profiles where the source explicitly records this audit flag. Omitted rather than assumed false when that source field is unavailable.
unexpected_ai_tutoring_content_delivered,**/student_data/student.json,"Whether AI tutoring content was unexpectedly delivered during a nominally human-arm study session. Human-arm profiles where the source explicitly records this audit flag. In other research, task 1120 and task 1282 are true and retain the associated accidental-AI metrics; omitted rather than assumed false when unavailable."
confusing,**/student_feedback_after_completion.json,Confusing recorded for this observation; interpret it within the file and parent-record context shown by the example path.
helpful,**/student_feedback_after_completion.json,Helpful recorded for this observation; interpret it within the file and parent-record context shown by the example path.
other,**/student_feedback_after_completion.json,Other recorded for this observation; interpret it within the file and parent-record context shown by the example path.
rating,**/student_feedback_after_completion.json,Rating recorded for this observation.
submitted_elapsed_seconds,**/student_feedback_after_completion.json,"Feedback submission time, in seconds from the student session origin. Fractional seconds are preserved; missing or invalid source times are empty."
answer_submission_count,**/student_practice_problem_attempts.csv,Number of answer submission.
concept_id,**/student_practice_problem_attempts.csv,Stable concept identifier joining the practice problem to concepts[].concept_id in lesson_plan_including_practice_problems.json. Every released practice-problem row.
final_answered_elapsed_seconds,**/student_practice_problem_attempts.csv,event time of the latest recorded answer submission for this practice problem; paired with final_student_answer and final_is_correct. Earlier submissions remain in the event/trajectory history when recorded. Measured in seconds from the shared origin described in README.
final_is_correct,**/student_practice_problem_attempts.csv,Whether the student's final recorded answer to this practice problem was correct.
final_student_answer,**/student_practice_problem_attempts.csv,Student's last recorded submitted answer for this practice problem.
first_shown_elapsed_seconds,**/student_practice_problem_attempts.csv,event time when the source application first marked this practice problem as shown. A missing event time is not zero duration or evidence that an unanswered problem was shown. Measured in seconds from the shared origin described in README.
lesson_plan_concept_label,**/student_practice_problem_attempts.csv,"Retained human-readable label associated with the practice record. In main-study sessions it is the released primary lesson plan's concept name by concept_id, or the explicit fallback 'Concept <concept_id> (source name not recorded)'. It is not guaranteed to match every additional or historical generation-context plan; supplementary records may preserve other historical labels."
lesson_plan_id,**/student_practice_problem_attempts.csv,Lesson plan id summarizing the student's recorded interaction with one practice problem.
practice_problem_id,**/student_practice_problem_attempts.csv,Practice problem id summarizing the student's recorded interaction with one practice problem.
problem_origin,**/student_practice_problem_attempts.csv,Whether the practice problem came from the prepared lesson plan or was created during live tutoring.
skipped_elapsed_seconds,**/student_practice_problem_attempts.csv,event time of the first recorded explicit skip of this practice problem; the application preserves the first skip event time. Skipping does not overwrite the final recorded answer or its correctness. Measured in seconds from the shared origin described in README.
source_problem_id,**/student_practice_problem_attempts.csv,Source problem id summarizing the student's recorded interaction with one practice problem.
was_answered,**/student_practice_problem_attempts.csv,Whether the student submitted at least one answer for this practice problem.
was_shown,**/student_practice_problem_attempts.csv,Whether the practice problem was shown to the student.
was_skipped,**/student_practice_problem_attempts.csv,Whether the source application recorded an explicit skip for this practice problem. A skipped problem may already have an incorrect answer; this is distinct from never shown or never answered.
action,**/video_events.csv,"Recorded playback event action for this control video; interpret it with video_id, playback position, cumulative watched seconds, and recorded_elapsed_seconds."
cumulative_video_watched_seconds,**/video_events.csv,"Recorded cumulative watched time for this video in seconds at the event. This is telemetry, not a direct measurement of attention."
phase,**/video_events.csv,Study phase recorded for this control-video playback event.
recorded_elapsed_seconds,**/video_events.csv,"Time this playback event was recorded, in seconds from the study-session origin."
video_id,**/video_events.csv,Identifier of the control video associated with this playback event.
video_playback_position_seconds,**/video_events.csv,"Recorded playback position in the video, in seconds, at the event; distinct from cumulative watched time."
first_video_event_elapsed_seconds,**/video_summary.csv,event time of the first recorded video event included in this summary. Measured in seconds from the shared origin described in README.
last_video_event_elapsed_seconds,**/video_summary.csv,event time of the last recorded video event included in this summary. Measured in seconds from the shared origin described in README.
maximum_cumulative_video_watched_seconds,**/video_summary.csv,"Maximum recorded cumulative watched-time counter for the video, in seconds; a player-derived counter rather than an independent measure of attention."
video_event_count,**/video_summary.csv,Recorded number of video events summarized for this video in this session.
video_id,**/video_summary.csv,Identifier of the control video associated with this playback event.
ai_model_preset_id,collection_metadata.json,"Experimental AI condition identifier combining the model family with the configured reasoning-effort or mode, for example opus-4.8-off versus opus-4.8-xhigh. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields."
cache_input_cost_multiplier,collection_metadata.json,Cache input cost multiplier for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields.
cache_multiplier_state,collection_metadata.json,Cache multiplier state for the de-identified AI usage and cost record; missing recorded cost is never interpreted as zero. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields.
comparison_task_architecture,collection_metadata.json,Comparison task architecture recorded for one completed paid human lesson-plan comparison. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields.
comparison_type,collection_metadata.json,Comparison type recorded for one completed paid human lesson-plan comparison. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields.
description,collection_metadata.json,Instructions for reconstructing original table values from this metadata.
grouped_tables,collection_metadata.json,Mapping from relative CSV path to keyed value maps. Match each CSV row on key_fields to exactly one rows entry.
instrument_id,collection_metadata.json,Assessment instrument identifier: gre_quant or gre_verbal. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields.
is_admin,collection_metadata.json,Whether admin is true for this record. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields.
is_demo,collection_metadata.json,Whether demo is true for this record. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields.
is_test,collection_metadata.json,Whether test is true for this record. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields.
key_fields,collection_metadata.json,CSV column names used jointly to select a grouped metadata row.
lesson_plan_generation_status,collection_metadata.json,Canonical field for lesson plan generation status. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields.
plan_rating_scale,collection_metadata.json,"Plan rating scale stored in the prepared lesson plan, its generation metadata, or its usage record. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields."
preference_key,collection_metadata.json,Preference key recorded for one rubric question within a completed human lesson-plan comparison. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields.
preference_scale_id,collection_metadata.json,"Preference scale id stored in the prepared lesson plan, its generation metadata, or its usage record. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields."
preference_score,collection_metadata.json,"Pairwise lesson-plan preference: positive favors plan A, negative favors plan B, and zero is a tie. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields."
pretest_maximum_score_points,collection_metadata.json,Canonical field for pretest maximum score points. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields.
research_consent_version_id,collection_metadata.json,Anonymous consent-form version identifier. Distinct source versions remain distinct; the identifier carries no calendar information. Shared values are stored in collection_metadata.json.
review_display_variant,collection_metadata.json,Review display variant recorded for one completed paid human lesson-plan comparison. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields.
rows,collection_metadata.json,Exact keyed field values shared by rows in the named CSV table; joins must resolve to one entry.
rubric_version,collection_metadata.json,Rubric version recorded for one completed paid human lesson-plan comparison. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields.
schema_version,collection_metadata.json,Version of the table-constant metadata structure.
tables,collection_metadata.json,Mapping from relative CSV path to fields constant across that table. Apply each value to every row; strings preserve original CSV encoding.
tutoring_arm,collection_metadata.json,"Assigned tutoring modality: ai, control, or human. 'Human' refers to the tutor modality, not whether the student is human. Stored once as a table constant or keyed mapping; reconstruct using the named table and key_fields."
correct_answer,question_bank_quant.csv,Canonical answer key used by the deployed scoring or grading record.
difficulty,question_bank_quant.csv,Recorded difficulty label for the assessment item or practice problem.
form,question_bank_quant.csv,Form in the immutable assessment question bank.
format,question_bank_quant.csv,Recorded assessment or practice-problem response format.
format_code,question_bank_quant.csv,Format code in the immutable assessment question bank.
options_json,question_bank_quant.csv,"JSON object mapping answer-option labels to text for choices stored at the item level. For Verbal text-completion items, choices are instead in blanks_json; {} here does not mean the item has no answer choices."
pair,question_bank_quant.csv,Pair in the immutable assessment question bank.
question,question_bank_quant.csv,"Full assessment question text, including comparison quantities and shared stimulus text when required. Historical years inside assessment stimuli are instructional content, not collection dates."
question_id,question_bank_quant.csv,"Public assessment-item identifier within this instrument, using P/Q form prefixes. Main and supplementary assessment responses join to this field. Quant and Verbal reuse the same P/Q IDs; choose the matching instrument bank."
skill,question_bank_quant.csv,Top-level GRE skill category assigned to the item.
source_question_id,question_bank_quant.csv,Historical application question identifier: M/N prefixes for Quant and Mv/Nv prefixes for Verbal. Joins expert-review context source_question_id to this bank. It is distinct from the public P/Q question_id; always select the matching instrument bank.
subskill,question_bank_quant.csv,Subskill in the immutable assessment question bank.
blanks_json,question_bank_verbal.csv,"JSON array of text-completion blanks, preserving each blank label and its answer-option label-to-text mapping. Use these choices for text-completion items even when options_json is {}."
correct_answer,question_bank_verbal.csv,Canonical answer key used by the deployed scoring or grading record.
difficulty,question_bank_verbal.csv,Recorded difficulty label for the assessment item or practice problem.
form,question_bank_verbal.csv,Form in the immutable assessment question bank.
format,question_bank_verbal.csv,Recorded assessment or practice-problem response format.
format_code,question_bank_verbal.csv,Format code in the immutable assessment question bank.
format_native,question_bank_verbal.csv,Format native in the immutable assessment question bank.
group_id,question_bank_verbal.csv,Group id in the immutable assessment question bank.
options,question_bank_verbal.csv,Answer-choice map for the question; keys vary by item and should be treated as choice labels rather than a fixed A-D schema.
options_json,question_bank_verbal.csv,"JSON object mapping answer-option labels to text for choices stored at the item level. For Verbal text-completion items, choices are instead in blanks_json; {} here does not mean the item has no answer choices."
pair,question_bank_verbal.csv,Pair in the immutable assessment question bank.
presented_order,question_bank_verbal.csv,Presented order in the immutable assessment question bank.
question,question_bank_verbal.csv,"Full assessment question text, including comparison quantities and shared stimulus text when required. Historical years inside assessment stimuli are instructional content, not collection dates."
question_id,question_bank_verbal.csv,"Public assessment-item identifier within this instrument, using P/Q form prefixes. Main and supplementary assessment responses join to this field. Quant and Verbal reuse the same P/Q IDs; choose the matching instrument bank."
skill,question_bank_verbal.csv,Top-level GRE skill category assigned to the item.
source_question_id,question_bank_verbal.csv,Historical application question identifier: M/N prefixes for Quant and Mv/Nv prefixes for Verbal. Joins expert-review context source_question_id to this bank. It is distinct from the public P/Q question_id; always select the matching instrument bank.
stimulus,question_bank_verbal.csv,Stimulus in the immutable assessment question bank.
subskill,question_bank_verbal.csv,Subskill in the immutable assessment question bank.
analysis_cohort_eligibility.applies_to[],studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for analysis cohort eligibility applies to records; this is a configuration parameter, not an observed participant result."
analysis_cohort_eligibility.apply_observed_outcome_filter,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for analysis cohort eligibility apply observed outcome filter; this is a configuration parameter, not an observed participant result."
analysis_cohort_eligibility.assessment_completion_boolean,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for analysis cohort eligibility assessment completion boolean; this is a configuration parameter, not an observed participant result."
analysis_cohort_eligibility.assessment_completion_scope,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for analysis cohort eligibility assessment completion scope; this is a configuration parameter, not an observed participant result."
analysis_cohort_eligibility.assessment_question_count,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for analysis cohort eligibility assessment question count; this is a configuration parameter, not an observed participant result."
analysis_cohort_eligibility.minimum_attempted_questions_each_test_inclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for analysis cohort eligibility minimum attempted questions each test inclusive; this is a configuration parameter, not an observed participant result."
analysis_cohort_eligibility.physically_pruned_main_leaderboard,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for analysis cohort eligibility physically pruned main leaderboard; this is a configuration parameter, not an observed participant result."
analysis_cohort_eligibility.scope,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for analysis cohort eligibility scope; this is a configuration parameter, not an observed participant result."
human_tutor_cost_sensitivity.fixed_session_hours,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for human tutor cost sensitivity fixed session hours; this is a configuration parameter, not an observed participant result."
human_tutor_cost_sensitivity.hourly_wage_usd,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for human tutor cost sensitivity hourly wage usd; this is a configuration parameter, not an observed participant result."
human_tutor_cost_sensitivity.interpretation,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for human tutor cost sensitivity interpretation; this is a configuration parameter, not an observed participant result."
protocol_adherence.ai_tutoring_activity.all_three_required,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence ai tutoring activity all three required; this is a configuration parameter, not an observed participant result."
protocol_adherence.ai_tutoring_activity.distinct_problem_definition,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence ai tutoring activity distinct problem definition; this is a configuration parameter, not an observed participant result."
protocol_adherence.ai_tutoring_activity.minimum_distinct_practice_problems_attempted_inclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence ai tutoring activity minimum distinct practice problems attempted inclusive; this is a configuration parameter, not an observed participant result."
protocol_adherence.ai_tutoring_activity.minimum_student_authored_chat_messages_inclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence ai tutoring activity minimum student authored chat messages inclusive; this is a configuration parameter, not an observed participant result."
protocol_adherence.ai_tutoring_activity.minimum_total_practice_answer_attempts_inclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence ai tutoring activity minimum total practice answer attempts inclusive; this is a configuration parameter, not an observed participant result."
protocol_adherence.ai_tutoring_activity.scope,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence ai tutoring activity scope; this is a configuration parameter, not an observed participant result."
protocol_adherence.ai_tutoring_activity.student_message_source,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence ai tutoring activity student message source; this is a configuration parameter, not an observed participant result."
protocol_adherence.ai_tutoring_activity.total_attempt_source,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence ai tutoring activity total attempt source; this is a configuration parameter, not an observed participant result."
protocol_adherence.arm_identity_exclusions[],studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence arm identity exclusions records; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.cutoff_inclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing cutoff inclusive; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.deduplication,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing deduplication; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.event_timestamp,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing event timestamp; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.final_seconds_ignored,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing final seconds ignored; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.instrument_windows.gre_quant.assessment_budget_seconds,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing instrument windows gre quant assessment budget seconds; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.instrument_windows.gre_quant.cutoff_minutes_inclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing instrument windows gre quant cutoff minutes inclusive; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.instrument_windows.gre_quant.cutoff_seconds_inclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing instrument windows gre quant cutoff seconds inclusive; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.instrument_windows.gre_verbal.assessment_budget_seconds,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing instrument windows gre verbal assessment budget seconds; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.instrument_windows.gre_verbal.cutoff_minutes_inclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing instrument windows gre verbal cutoff minutes inclusive; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.instrument_windows.gre_verbal.cutoff_seconds_inclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing instrument windows gre verbal cutoff seconds inclusive; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.insufficient_support_action,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing insufficient support action; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.interval_definition,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing interval definition; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.minimum_intervals_inclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing minimum intervals inclusive; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.percentile_method,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing percentile method; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.phase_boolean,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing phase boolean; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.phase_start_timestamp,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing phase start timestamp; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.primary_rule.label,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing primary rule label; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.primary_rule.percentile,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing primary rule percentile; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.primary_rule.rapid_interval_threshold_seconds_exclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing primary rule rapid interval threshold seconds exclusive; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.scope,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing scope; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.sensitivity_rule.label,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing sensitivity rule label; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.sensitivity_rule.percentile,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing sensitivity rule percentile; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.sensitivity_rule.rapid_interval_threshold_seconds_exclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing sensitivity rule rapid interval threshold seconds exclusive; this is a configuration parameter, not an observed participant result."
protocol_adherence.assessment_timing.submission_event,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence assessment timing submission event; this is a configuration parameter, not an observed participant result."
protocol_adherence.task_level_boolean,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for protocol adherence task level boolean; this is a configuration parameter, not an observed participant result."
release_cohort_contract.assessment_effort.maximum_tab_outs_inclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract assessment effort maximum tab outs inclusive; this is a configuration parameter, not an observed participant result."
release_cohort_contract.assessment_effort.minimum_item_time_seconds_each_test_inclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract assessment effort minimum item time seconds each test inclusive; this is a configuration parameter, not an observed participant result."
release_cohort_contract.assessment_effort.scope,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract assessment effort scope; this is a configuration parameter, not an observed participant result."
release_cohort_contract.attempt_adjudication.restricted_identity_verification_required_for_complete_check,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract attempt adjudication restricted identity verification required for complete check; this is a configuration parameter, not an observed participant result."
release_cohort_contract.attempt_adjudication.rule,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract attempt adjudication rule; this is a configuration parameter, not an observed participant result."
release_cohort_contract.attempt_adjudication.within_run_duplicate_rule,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract attempt adjudication within run duplicate rule; this is a configuration parameter, not an observed participant result."
release_cohort_contract.contract_version,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract contract version; this is a configuration parameter, not an observed participant result."
release_cohort_contract.control_pretest_review.minimum_seconds_inclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract control pretest review minimum seconds inclusive; this is a configuration parameter, not an observed participant result."
release_cohort_contract.control_pretest_review.required_condition,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract control pretest review required condition; this is a configuration parameter, not an observed participant result."
release_cohort_contract.control_pretest_review.scope,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract control pretest review scope; this is a configuration parameter, not an observed participant result."
release_cohort_contract.instrument_eligibility.gre_quant.assessment_bank_version_id,studentbench_overall_parameters.json,Immutable identifier and content fingerprint for the assessment bank used by the student.
release_cohort_contract.instrument_eligibility.gre_quant.pretest_score_maximum_inclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract instrument eligibility gre quant pretest score maximum inclusive; this is a configuration parameter, not an observed participant result."
release_cohort_contract.instrument_eligibility.gre_quant.pretest_score_minimum_inclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract instrument eligibility gre quant pretest score minimum inclusive; this is a configuration parameter, not an observed participant result."
release_cohort_contract.instrument_eligibility.gre_verbal.assessment_bank_version_id,studentbench_overall_parameters.json,Immutable identifier and content fingerprint for the assessment bank used by the student.
release_cohort_contract.instrument_eligibility.gre_verbal.pretest_score_maximum_inclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract instrument eligibility gre verbal pretest score maximum inclusive; this is a configuration parameter, not an observed participant result."
release_cohort_contract.instrument_eligibility.gre_verbal.pretest_score_minimum_inclusive,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract instrument eligibility gre verbal pretest score minimum inclusive; this is a configuration parameter, not an observed participant result."
release_cohort_contract.outcome_fields_forbidden_for_selection[],studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract outcome fields forbidden for selection records; this is a configuration parameter, not an observed participant result."
release_cohort_contract.research_provenance.main_cohort_membership_basis,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract research provenance main cohort membership basis; this is a configuration parameter, not an observed participant result."
release_cohort_contract.research_provenance.public_profile_administrative_flags[],studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract research provenance public profile administrative flags records; this is a configuration parameter, not an observed participant result."
release_cohort_contract.research_provenance.research_consent_field,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract research provenance research consent field; this is a configuration parameter, not an observed participant result."
release_cohort_contract.research_provenance.research_consent_required_value,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract research provenance research consent required value; this is a configuration parameter, not an observed participant result."
release_cohort_contract.research_provenance.source_collections[].ai_model_preset_ids[],studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract research provenance source collections records ai model preset ids records; this is a configuration parameter, not an observed participant result."
release_cohort_contract.research_provenance.source_collections[].arms[],studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract research provenance source collections records arms records; this is a configuration parameter, not an observed participant result."
release_cohort_contract.research_provenance.source_collections[].collection_id,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract research provenance source collections records collection id; this is a configuration parameter, not an observed participant result."
release_cohort_contract.research_provenance.source_collections[].experiment_run_ids[],studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract research provenance source collections records experiment run ids records; this is a configuration parameter, not an observed participant result."
release_cohort_contract.research_provenance.source_collections[].instrument,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract research provenance source collections records instrument; this is a configuration parameter, not an observed participant result."
release_cohort_contract.treatment_integrity.cross_model_lesson_plan_fallback_allowed,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract treatment integrity cross model lesson plan fallback allowed; this is a configuration parameter, not an observed participant result."
release_cohort_contract.treatment_integrity.incidental_exposure_policy,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract treatment integrity incidental exposure policy; this is a configuration parameter, not an observed participant result."
release_cohort_contract.treatment_integrity.provenance_pilot_arms_allowed_in_main,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract treatment integrity provenance pilot arms allowed in main; this is a configuration parameter, not an observed participant result."
release_cohort_contract.treatment_integrity.substantive_additional_treatment_allowed,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for release cohort contract treatment integrity substantive additional treatment allowed; this is a configuration parameter, not an observed participant result."
resampling.cost_per_gain_paired_participant_bootstrap.draws,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for resampling cost per gain paired participant bootstrap draws; this is a configuration parameter, not an observed participant result."
resampling.cost_per_gain_paired_participant_bootstrap.minimum_valid_draw_fraction,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for resampling cost per gain paired participant bootstrap minimum valid draw fraction; this is a configuration parameter, not an observed participant result."
resampling.cost_per_gain_paired_participant_bootstrap.seed,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for resampling cost per gain paired participant bootstrap seed; this is a configuration parameter, not an observed participant result."
resampling.difficulty_calibration_participant_bootstrap.draws,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for resampling difficulty calibration participant bootstrap draws; this is a configuration parameter, not an observed participant result."
resampling.difficulty_calibration_participant_bootstrap.seed,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for resampling difficulty calibration participant bootstrap seed; this is a configuration parameter, not an observed participant result."
resampling.early_mastery_cross_validation.folds,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for resampling early mastery cross validation folds; this is a configuration parameter, not an observed participant result."
resampling.early_mastery_cross_validation.seed,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for resampling early mastery cross validation seed; this is a configuration parameter, not an observed participant result."
resampling.engagement_participant_bootstrap.draws,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for resampling engagement participant bootstrap draws; this is a configuration parameter, not an observed participant result."
resampling.engagement_participant_bootstrap.seed,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for resampling engagement participant bootstrap seed; this is a configuration parameter, not an observed participant result."
resampling.gemini36_random_deletion_sensitivity.iterations,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for resampling gemini36 random deletion sensitivity iterations; this is a configuration parameter, not an observed participant result."
resampling.gemini36_random_deletion_sensitivity.seed_root,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for resampling gemini36 random deletion sensitivity seed root; this is a configuration parameter, not an observed participant result."
resampling.human_ai_section_permutation.chunk_size,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for resampling human ai section permutation chunk size; this is a configuration parameter, not an observed participant result."
resampling.human_ai_section_permutation.draws,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for resampling human ai section permutation draws; this is a configuration parameter, not an observed participant result."
resampling.human_ai_section_permutation.seed,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for resampling human ai section permutation seed; this is a configuration parameter, not an observed participant result."
resampling.problem_flag_participant_bootstrap.draws,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for resampling problem flag participant bootstrap draws; this is a configuration parameter, not an observed participant result."
resampling.problem_flag_participant_bootstrap.seed,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for resampling problem flag participant bootstrap seed; this is a configuration parameter, not an observed participant result."
schema_version,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for schema version; this is a configuration parameter, not an observed participant result."
statistical_inference.benjamini_hochberg_fdr_alpha,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for statistical inference benjamini hochberg fdr alpha; this is a configuration parameter, not an observed participant result."
statistical_inference.confidence_level,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for statistical inference confidence level; this is a configuration parameter, not an observed participant result."
statistical_inference.equivalence_margins_pooled_sd.primary,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for statistical inference equivalence margins pooled sd primary; this is a configuration parameter, not an observed participant result."
statistical_inference.equivalence_margins_pooled_sd.sensitivity,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for statistical inference equivalence margins pooled sd sensitivity; this is a configuration parameter, not an observed participant result."
statistical_inference.equivalence_tost_alpha,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for statistical inference equivalence tost alpha; this is a configuration parameter, not an observed participant result."
statistical_inference.holm_familywise_alpha,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for statistical inference holm familywise alpha; this is a configuration parameter, not an observed participant result."
statistical_inference.nominal_two_sided_alpha,studentbench_overall_parameters.json,"Prespecified protocol or analysis setting for statistical inference nominal two sided alpha; this is a configuration parameter, not an observed participant result."
ai_session_count,**/geography_equivalence_aggregates.csv,Number of AI-tutored sessions in one eligible geography-related subgroup; Quant and Verbal are combined.
ai_mean_gain_pp,**/geography_equivalence_aggregates.csv,"Mean posttest minus pretest score for the AI sessions, in percentage points on the 27-question assessment."
ai_sample_variance_gain_pp2,**/geography_equivalence_aggregates.csv,"Unbiased sample variance (ddof=1) of AI-session learning gains, in squared percentage points."
human_session_count,**/geography_equivalence_aggregates.csv,Number of human-tutored sessions in the same subgroup; Quant and Verbal are combined.
human_mean_gain_pp,**/geography_equivalence_aggregates.csv,"Mean posttest minus pretest score for the human-tutored sessions, in percentage points on the 27-question assessment."
human_sample_variance_gain_pp2,**/geography_equivalence_aggregates.csv,"Unbiased sample variance (ddof=1) of human-session learning gains, in squared percentage points."
analysis_axis,**/equivalence_robustness_grid_counts.csv,Family of subgroup definitions in the original combined AI-versus-human equivalence robustness grid.
planned_definitions,**/equivalence_robustness_grid_counts.csv,"Number of planned subgroup definitions on this analysis axis, including duplicate and ineligible definitions."
unique_groups,**/equivalence_robustness_grid_counts.csv,Number of distinct session groups first encountered on this analysis axis after deduplicating the full grid.
duplicate_definitions,**/equivalence_robustness_grid_counts.csv,Number of definitions on this axis whose session group already appeared elsewhere in the ordered grid; duplicates are not retested.
eligible_unique_groups,**/equivalence_robustness_grid_counts.csv,"Unique groups on this axis meeting the original count requirements: at least 20 AI and 20 human sessions, with at least 5 per treatment in each GRE section."
ineligible_unique_groups,**/equivalence_robustness_grid_counts.csv,Unique groups on this axis that fail one or more original session-count requirements and are not tested.
figure_path,question_bank_quant.csv,Relative path from the dataset root to the original assessment SVG. Empty when the item has no figure. Shared stimuli reference the same file.
input_usd_per_million_tokens,3_cost_data/replay_price_schedule.csv,Fixed input-token rate used for the published replay cost estimates. Measured cache multipliers in collection_metadata.json apply to this rate.
output_usd_per_million_tokens,3_cost_data/replay_price_schedule.csv,Fixed output-token rate used for the published replay cost estimates.
reasoning_tokens_additional,3_cost_data/replay_price_schedule.csv,Whether reported reasoning tokens must be added to completion tokens for output billing. False means completion tokens already include reasoning.