Test Scoring and Analysis Using SAS

Chapter 7: Assessing Test Reliability

Introduction

Computing Split-Half Reliability

Computing Kuder-Richardson Formula 20 (KR-20)

Computing Cronbach’s Alpha

Demonstrating the Effect of Item Discrimination on Test Reliability

Demonstrating the Effect of Test Length on Test Reliability

Conclusion

Introduction

Test reliability is a measure of repeatability. A test would not be very useful if, upon repeat administrations, a person’s score varied widely. Intuitively, test reliability is related to test length. You would not expect a test of only a few items to be highly reliable because a person could make a few accurate guesses and obtain a high score or make a few bad guesses and obtain a low score.

One of the classic methods of assessing test reliability ...