What Is Test Reliability?

Validity asks: does this test measure what it claims to measure? Reliability asks a different question: does it measure it consistently?  Both matter. But without reliability, validity is meaningless.

What Reliability Means

If you gave the same person the same assessment twice — with a reasonable gap between testings — you should get nearly identical scores. The smaller the variance between the two results, the more reliable the instrument. Reliability is measured using a statistic called a correlation coefficient, which ranges from +1.00 (perfect positive relationship) to -1.00 (perfect negative relationship). A coefficient of 0.00 means no relationship at all.

For a pre-employment assessment, you want a correlation coefficient that approaches +1.00. That means if someone scores high in Assertiveness today, they’ll score high in Assertiveness when retested — because Assertiveness is a stable personality trait, not a mood.

Why This Matters in Practice

An unreliable assessment is actively harmful. If a person who is genuinely high in dominance scores low on your instrument half the time, you end up with data you cannot trust — and worse, data you might act on incorrectly.

Here’s the practical concern: if you want to use historical assessment data to refine your hiring benchmarks over time — to learn what scores actually correlate with success in your specific company — you need data that is consistent. Unreliable data will never yield those insights. The patterns will always seem to shift.