Use of Norm-Referenced Tests

¶ … norm-referenced test is an assessment that produces a score (or scores) that represent an estimate of where the individual stands with respect to a predefined peer group on a particular trait, dimension, or ability (Rust & Golombok, 2014). Norm-referenced tests allow for a comparison on whether an individual performed at, above, or below expectation with respect to individuals that are similar to them. For example, traditional IQ tests yield standardized scores with a mean of 100 and a standard deviation of 15 (or 16; Sattler & Ryan, 2009).

The standardized score is a score that should be interpreted and not the raw scores. In terms of simple point estimates (single IQ scores) the researcher/clinician can compare the individual performance to the norm -- reference group with respect to the score's deviation from the mean. Comparing individual scores to norm -- reference scores in this manner allows the researcher/clinician to determine how far above/below the individual has scored compared to the typical performance for his/her peers on the test (Rust & Golombok, 2014).

The comparison of a score on a norm-referenced test such as an IQ test should always be made within the specific context of the peer group to which the individual's scores are being compared (Sattler & Ryan, 2009). For example, most IQ tests do not specifically use educational attainment as a variable and defining the particular comparison scores.

Thus, the performance of a nuclear physicist who produces a full-scale IQ score of 95 would be interpreted quite differently than a factory worker with the same IQ score, even though both scores are technically within the average range of performance for the reference group (Sattler & Ryan, 2009). Moreover, point estimates such as the Full Scale IQ score or standardized scores on IQ subtests have poorer reliability than interval estimates such as confidence intervals and summary scores have greater reliability than individual test score results (Urbina, 2014).

Whenever possible it is advised to interpret the more reliable estimate in favor of the less reliable ones (e.g., confidence intervals over point estimates of IQ scores, Full Scale IQ score over Index Scores, and Index Scores of single test scores; Sattler & Ryan, 2009).

The reason why raw scores are transformed into standardized scores is to allow researchers/clinicians to make meaningful comparisons between the individual score and the performance by the individual's peer group and to make comparisons between performances on different tests that measure the same or even different abilities, traits, etc. (Rust & Golombok, 2014; Urbina, 2014). The use of standardized scores for peer-group comparisons has already been addressed above.

In the latter case, researchers and clinicians would be unable to compare raw scores on two different IQ tests or between two tests that measure totally different qualities (e.g., such as IQ and memory tests) because these tests all have different means and standard deviations. Thus, an individual's performance on a test with a raw-score mean of 75 and raw standard deviation of eight could not.