Test error & score confidence (1)

In this lesson, you will learn about the concept of the standard error of measurement, how it relates to confidence intervals, and why understanding reliability is key to interpreting test scores. Dr Graham Tyler explains how we can estimate how closely a candidate’s test score reflects their true ability, and why this matters for the validity of psychometric tests. The unit also introduces the importance of using tests with good reliability and offers practical advice for interpreting results in real assessment situations.

After this talk, you will:

• Understand why this module may seem more challenging than previous ones, and that this is normal for many learners.

• Recognise the importance of confidence intervals in psychometric testing.

• Know the difference between observed score, true score, and error in test results.

• Be able to explain the concept of standard error of measurement (SEM) and why it matters.

• Understand how to calculate the standard error of measurement.

• Realise how test-retest reliability affects your confidence in a candidate's observed score.

• Know how to use SEM to determine the confidence interval for an individual’s test score, and what 68%, 96%, and 99% confidence intervals mean.

• Be aware that if test reliability is low, the confidence interval becomes wider, making the score less meaningful.

• Remember that observed scores are never a perfect reflection of the true score—there is always a margin of error.

• Understand that the standard error of measurement is often about half a standard deviation when reliability is at least 0.7.

• Be prepared to compare candidate scores confidently using these statistical concepts, leading into understanding the standard error of difference in future sections.