Raw Score vs. Scaled Score on an IQ Test

On a serious IQ test, a raw score is the behind-the-scenes count: how many items, points, or scoring credits you earned before the test converts that performance into an interpretable score. It matters, but it is not usually the number you should use to understand the result. The scaled score is the interpretation layer. It translates raw performance through the test's scoring model so the result can be compared with the right group, form, and age range. That is why the same number correct can mean different things on two different forms of a test.

A raw score answers a mechanical question: what did you earn on this exact set of items? A scaled score answers a measurement question: where does that performance sit on the test's intended scale? Think of a raw score like ingredients before cooking. The ingredients are real, but they are not the final dish. In testing, the final dish is the reported score: an IQ score, index score, percentile, or another standardized result.

Suppose two people answer 32 questions correctly. If one form has easier items and the other form has harder items, those two raw scores should not automatically be treated as equal. The scoring model has to account for item difficulty before the results can be compared fairly. Age can change the interpretation too. Many cognitive tests compare people against others in a similar age range. That is where the norm group becomes important: it tells you who the score is being compared with.

Conversion is not a cosmetic step. It is the step that makes a score readable. A good conversion process can account for item difficulty, test form, age group, and the scale the test is designed to report. That is also why raw scores are weak as public claims. If someone says they got 38 out of 50 on an online quiz, you still do not know whether the items were easy, whether the sample was credible, or whether the result was converted onto a real IQ scale. RIOT's guide to IQ score scales explains the common 100-and-15 scale in more detail.

Read the report in this order: 1. Start with the reported scale and ask whether the result is an IQ score, index score, percentile, subtest score, or just a raw count. 2. Check the comparison group, because a converted score is only as useful as the sample behind it. 3. Look for uncertainty language; if the report gives a confidence interval, treat the score as a range estimate rather than a perfectly exact point. 4. Check task coverage, because a single puzzle type cannot carry the same meaning as a broader battery. That is one reason Full Scale IQ is interpreted differently from a narrow subtest result.

Raw scores are not useless. Test developers and scorers need them to check item-level patterns, apply conversion tables, and spot unusual response profiles. They can also help explain why two reported scores differ. But the raw count should support interpretation, not replace it. If a score report stops at the raw number and calls it an IQ score, the report is missing the part that makes the number meaningful.

Use the scaled or standard score for interpretation. Use the raw score as supporting detail. The raw count tells you what happened on the items; the converted score tells you what that performance means.

Sources: APA Dictionary: standardized score; APA Dictionary: deviation IQ; APA Dictionary: raw score.