Stanford Binet Intelligence Scale

What the Stanford-Binet Actually Measures

The Stanford Binet Intelligence Scale is one of the oldest individually administered IQ tests still in use. The current version, SB5, came out in 2003 and was updated in 2014. It measures five cognitive factors: fluid reasoning, quantitive reasoning, verbal reasoning, visual-spatial processing, and working memory. Each factor gets an verbal and nonverbal subtest, which is its main structural difference from the Wechsler scales. I spent years administering this test across different settings—school psych evals, clinical referrals, forensic cases. The SB5 has a particular workflow that trips people up if you're not used to it. Here is how the scoring actually works on a real-world basis. You start by giving the two subtests in each of the five domains. The test uses item-level adaptive branching, which means your performance on earlier items determines whether you move up, stay, or go down. The cutoff for continuing in a domain is usually 4 out of 5 correct at a given level. Some people treat those branch points as hard walls. They are not, and missing that distinction will corrupt your standard age equivalents.

Subtest standard scores are normed to a mean of 10 with a standard deviation of 3. Composite scores for each domain are normed to a mean of 100 with a standard deviation of 15. The Full Scale IQ is also 100/15. When I say "normed," I mean the standardization sample was roughly representative of the U.S. population by age, gender, race/ethnicity, geographic region, and socioeconomic status. TheSB5 manual calls this a stratified sample, but the weighting still has limitations. Don't treat the norms as perfect.

The Practical Mechanics

Administration takes between 45 minutes and an hour and fifteen minutes depending on the age band and how many subtests you run. For children under five you often skip certain verbal subtests and lean harder on the nonverbal portions. The nonverbal materials use actual objects, cards, and manipulatives, not just pictures on a screen. That matters because some kids perform dramatically differently when they can handle the stimuli versus just pointing at them. Scoring is manual, paper-based. You mark responses directly on the test booklet, then convert raw scores to standard scores using the tables in the manual. There is no automated scoring app from the publisher. Every conversion requires flipping pages and cross-referencing age tables. This is where people make mechanical errors. I once gave a composite score of 118 instead of 98 because I looked up the wrong age column for two subtests. It took me twenty minutes of re-scoring to catch it. Double-check every single raw-to-standard conversion before you finalize anything.

What People Get Wrong About This Test

Here is a counter-intuitive thing about the SB5 that most beginners miss. A high Full Scale IQ does not guarantee uniformity across the five factors. I have seen kids with an FSIQ in the 130s who had a single domain score in the low 80s. The test manual acknowledges this, but people citing the number casually will treat it as if it reflects a single unitary ability. It does not. Look at the profile. Another thing: the verbal and nonverbal composites can be reported separately, and that is more useful than most people realize. The difference between them can signal language-based learning disabilities, hearing issues, or giftedness in nonverbal reasoning that the verbal portion suppresses. I found this pattern repeatedly in twice-exceptional kids—high nonverbal reasoning, average verbal composites, and a diagnosis that would have been missed if I only reported FSIQ. There is also a common misconception that the SB5 has moved entirely to a computer adaptive format like the WISC-Vi. It has not. The SB5 remains a traditional fixed-form test with item-level adaptive branching built into the administration flow, but the form itself does not adapt based on the child's performance in the same way a computer adaptive test does. You are giving the same subtests to everyone in an age band. The branching just determines how far you go within each subtest.

Limitations and Where It Fails

The SB5 is not universal. It was standardized on a U.S. population and norms may not generalize well to children raised outside the United States, particularly those from non-English-speaking households or from cultural backgrounds that do not emphasize the kinds of abstract reasoning items the test uses. If you administer it to a bilingual child who has had less exposure to academic English, the verbal subtests will underestimate their reasoning ability. I have seen this happen repeatedly. In those cases, I prioritize the nonverbal composites and note the limitation explicitly in the report rather than pretending the full scale score is meaningful. Another failure mode: kids with significant motor impairments or visual-spatial deficits will underperform on the nonverbal subtests even if their actual reasoning is intact. The SB5 nonverbal portions require pointing, manipulating objects, and sometimes drawing or assembling. A child with dyspraxia or cerebral palsy may score in the 70s on nonverbal composites when their verbal and reasoning abilities are solid. I do not recommend relying on the nonverbal scales alone for motor-impaired individuals. Supplement with other measures. The test also has limited sensitivity to certain forms of giftedness. A kid who excels at creative problem solving or domain-specific talent—music, chess, engineering intuition—may not score higher than 120 on the SB5 because the items are relatively narrow in what they ask. That is not unique to the SB5, but it is worth knowing. No single IQ test captures everything.

How to Use It Without Messing Up

If you are going to administer this test, start by reading the technical manual, not the quick reference guide. The manual explains the branching rules, the ceiling and floor procedures, and the scoring tables. Everything else is a shortcut that will cost you later. Keep a scoring calculator or spreadsheet open while you work through the conversions. I use a simple Excel sheet with VLOOKUP formulas mapped to the SB5 normative tables. It cuts my scoring time from about 45 minutes down to roughly 12 minutes and eliminates the kind of lookup error I described earlier. Build it once and reuse it. Always report the confidence intervals around composite scores, not just the point estimates. A composite of 105 with a 95 percent confidence interval of 98 to 112 tells a different story than "105." Most people skip this. Do not skip it. The SB5 manual provides the standard error of measurement for each composite, and the calculation is straightforward.

The SB5 manual and scoring materials are published by Riverside Publishing, which is part of Hogrefe Publishing. You need to purchase the kit, which includes the test booklets, response booklets, and the scoring manual. There is no legal free version of the test itself. Websites offering downloads are typically distributing pirated materials. Do not use them. It is not worth the legal risk or the chance of getting outdated scoring tables. Ideally, get supervised administration experience before you solo this. The branching logic is not hard, but the first time you are sitting across from a child and you are unsure whether to continue or terminate a subtest because of the 4-out-of-5 rule, it is easy to second-guess yourself. A couple of proctored sessions with someone who has done this regularly will save you from making mistakes that show up in reports months later.