Science Assessment

Item Analysis
Discrimination Factor and Test Reliability

Discrimination Factor and Percentage Correct

Discrimination Factor is an index scaled between +1 and -1. It measures the ability of a question to correlate a student’s performance on a
question and a student’s overall performance on an assessment. The larger the positive discrimination factor (> +0.2) the greater the correlation.
In other words, students who scored well on the assessment, scored well on the question and students that didn’t score well on the assessment,
didn’t score well on the question. A negative discrimination factor indicates a reverse correlation. Students who scored well on the assessment,
scored poorly on the question and students who scored poorly on the assessment, scored well on the question. (See page 2.)

Percentage Correct is the number of questions answered correctly divided by the total number of questions on the assessment.

Target Zone
+1

Hard Question Medium Question Easy Question

Too Easy
+0.5 (Target Line)
Discrimination Factor

+0.2
Weak Question – Little to No Discrimination
0
40% 55% 70% 90%
PERCENTAGE CORRECT

Review, Rewrite or Replace Question

-1

© 2016 GravityKills.net, Created by David Carroll and Reviewed by Jerry Grizzle

Discrimination Factor – Simplistic Representation

The discrimination factor is cumbersome and time consuming to calculate without the aid of a computer spreadsheet.
Here’s the equation:

∑(SA − A)(SQ − Q)
i i
o SA is the student’s score on the assessment
Discrimination Factor = i=1 o A is the assessment average
n n o SQ is the student’s score on the question
∑(SAi − A)2 ∑(SQ − Q) i
2
o Q is the question average
i=1 i=1

A quick and easy way to visualize the discrimination factor is to plot the assessment Points
results. Select an assessment question to analyze. Plot the number of points each ABOVE
student’s assessment score is above or below the assessment average versus their Average
question response (correct or incorrect) along the dotted line. Scores that lie in the
yellow squares have a direct correlation and scores that lie in the gray squares have a Assessment
reverse correlation. The number of points plotted will equal the number of students Average
that took the assessment. Draw a line of best fit. The slope of the line of best fit
represents the discrimination factor. Positive slope = positive discrimination factor; Points
negative slope = negative discrimination factor. Repeat this process for each BELOW
assessment question to be analyzed. Average

Incorrect Correct
Test Reliability Response Response

Test reliability is a measure of the likeliness that the test will produce consistent scores.
It is scored on a scale between 0 and 1. Test reliability depends on item discrimination
factors, number of questions and diversity of subject matter tested.

0.5 Revise 0.6 Acceptable 0.7 Good Excellent 0.9 Revise 1

Revision needed unless there Assessment has
are fewer than 10 items redundancies or is very
homogeneous

© 2016 GravityKills.net, Created by David Carroll and Reviewed by Jerry Grizzle