1.2 Psychometric Scoring Metrics: Raw Scores, USS, SAS, Stanines & Norms

Read the score type before interpreting the number

A CogAT report contains several representations of performance. Raw Score, Universal Scale Score (USS), Standard Age Score (SAS), percentile rank, and stanine are not interchangeable. Reading the label first prevents a number such as 90 from being mistaken for a percentage correct, a percentile, or a standard score.

For example, a score of 90 labeled SAS is below the SAS mean of 100. A percentile rank of 90 describes a high position in a comparison group. These two “90” values communicate different information. Neither one independently tells you how many questions the student answered correctly.

A useful reading sequence is to confirm the form, level, date, and norm edition; read each battery separately; then inspect any composite. Ask the school to explain unfamiliar report fields rather than estimating missing values from a practice percentage.

Raw score: count correct answers

Riverside defines the raw score as the number answered correctly. Wrong answers do not subtract from that count, and omissions do not add points. A battery with 52 items cannot have a raw score above 52. The complete high-school battery maxima are 64 Verbal, 52 Quantitative, and 60 Nonverbal.

Consider an independent 18-question practice set with 13 correct, three wrong, and two omitted. The raw count is 13, and the fraction correct is $13/18$, approximately 72.2%. That is descriptive practice information, not an official SAS or percentile. A student's percentile also depends on the test's calibrated difficulty and the applicable comparison group.

Keep the number attempted separate from the number correct. Someone answering 12 of 12 attempted items on a 20-item set has 100% accuracy on attempted items but only 12 correct out of 20 available. Someone answering 16 correctly out of 20 has more correct answers despite lower attempted-item accuracy. Recording both quantities helps you distinguish difficulty from incomplete coverage.

USS: the bridge to normative scores

The USS is a developmental scale extending from kindergarten through grade 12. It provides an entry into age and grade norm conversions. A raw count on one level cannot simply be compared with the same count on another level: the tasks differ in difficulty.

Suppose two original practice sets each contain 20 items, one straightforward and one challenging. Getting 15 correct on both is not enough to establish equivalent developed reasoning. An official scale accounts for the administered form and level; our practice bank does not contain the calibration tables required to do this.

Use the reported USS for its intended interpretation. Do not invent an item-weighting formula, a “typical high-school USS range,” or a fixed amount of growth per year. Changes across reports deserve review of age, administration conditions, and scoring rules as well as the numerical difference.

SAS: an age-based standard score

The SAS has a mean of 100 and standard deviation of 16. Riverside's Forms 7 and 8 report guide gives a range of 50–160. The distance from the mean can be expressed as a standardized difference:

$$z = \frac{\text{SAS} - 100}{16}$$

For SAS 116, $z=(116-100)/16=1$. For SAS 132, $z=2$. These are checked calculations about the scale. They do not establish a universal gifted cutoff. Exact reported percentile conversions should come from the applicable owner norms, rather than a generic normal-distribution calculator.

A common error is importing a standard deviation of 15 from another assessment. On the CogAT scale, two standard deviations above 100 is 132, not 130. Likewise, a particular SAS does not prove a fixed IQ, predict a course grade, or determine a career.

Percentiles: identify the comparison group

An Age Percentile Rank (APR) compares performance with same-age students in the national norm group. A Grade Percentile Rank (GPR) compares performance with same-grade students. Local norms compare with a specified local school or district group. These are different reference groups, so their ranks may differ without an error in the report.

An APR of 90 describes relative standing, not 90% correct. Percentile distances also are not equal achievement increments: the difference between ranks 50 and 55 cannot be interpreted as the same amount of reasoning change as between 94 and 99.

Stanines and composites

A stanine condenses a normative result into one of nine broad bands; the scale's average is 5. Broad bands help communicate results without treating every small score difference as a distinct ability category. A stanine is not a count of questions missed or a course mark out of nine.

Read the actual composite label: VQ, VN, QN, or VQN identifies which batteries were included. Composites use the owner's USS-based procedures and norm conversions. Averaging three SAS values or adding three percentiles does not reproduce the official composite. Nor is a median stanine the composite SAS; it serves a different purpose in the Ability Profile.

Current norms and uncertainty

Riverside's August 2026 resource guidance confirms that 2024 norms are available in DataManager. Check which norm edition was used on a particular report. Do not mix a historical conversion table with a newly normed report or assume that a changed percentile necessarily means the student lost reasoning skill.

Score measurement has uncertainty. Request the relevant report guidance or confidence interval if a small difference matters to a decision. This guide does not assign every battery a universal standard error, declare nearby scores statistically identical, or round a borderline result up to a program cutoff.

Official sources checked October 10, 2026: score definitions, report guide, and current DataManager resources.