Help & Documentation

Answer Key CSV

Your answer key file must include the following columns:

• item_index - Item number (1, 2, 3, ...)
• correct_option - Correct answer (A, B, C, D, E, etc.)
• max_points (optional) - Maximum points for the item (default: 1)

Example:

item_index,correct_option,max_points
1,A,1
2,C,1
3,B,1
4,D,1
5,A,1

---

Student Responses CSV

Your responses file must include:

• student_id - Unique student identifier
• item_1, item_2, item_3, ... - Student responses for each item
• group (optional) - Student group for cohort analysis

Example:

student_id,item_1,item_2,item_3,item_4,item_5
student_001,A,C,B,D,A
student_002,B,C,B,D,A
student_003,A,A,B,D,B
student_004,A,C,C,D,A

Item Difficulty (p)

The proportion of students who answered the item correctly:

p = (Number of correct responses) / (Total number of responses)

• p > 0.9: Too easy
• 0.3 ≤ p ≤ 0.9: Acceptable
• p < 0.2: Too hard

Point-Biserial Correlation (r_pb)

Measures how well an item discriminates between high and low performers:

r_pb = ((M1 - M0) / SD_total) × √(p × q)

Where:

• M1 = Mean total score of students who got the item correct
• M0 = Mean total score of students who got the item incorrect
• SD_total = Standard deviation of total scores
• p = Proportion correct, q = 1 - p

• r_pb ≥ 0.3: Good discrimination
• 0.2 ≤ r_pb < 0.3: Acceptable
• r_pb < 0.2: Poor discrimination (flagged)

KR-20 Reliability

Internal consistency reliability for dichotomously scored items:

KR-20 = (k / (k - 1)) × (1 - (Σ(p × q) / Var_total))

Where:

• k = Number of items
• p = Proportion correct for each item
• q = 1 - p
• Var_total = Variance of total scores

• KR-20 ≥ 0.8: Excellent
• 0.7 ≤ KR-20 < 0.8: Good
• 0.6 ≤ KR-20 < 0.7: Acceptable
• KR-20 < 0.6: Poor (needs improvement)

Standard Error of Measurement (SEM)

SEM = SD_total × √(1 - reliability)

Estimates the standard deviation of measurement errors.