So You Want to Try Mednick's Remote Associates Test
I've been administering and analyzing RAT scores for about twelve years across a few different research labs. The test itself is simple enough that you could probably skip the explanation and just do it, but there are enough people misunderstanding what it actually measures that I should clarify before anyone gets too excited about "unlocking hidden potential." Here is how the test works, what the numbers mean, and where the whole thing falls apart if you take it too literally. The Remote Associates Test was created by S. Mednick in 1962 as a way to measure associative thinking, which is basically how well your brain connects seemingly unrelated concepts. You get three words, like cube, cottage, and Swiss, and you have to find the one word that links them all. In that example the answer is cheese. You see it immediately if your associations are working right, or you stare at it for twenty seconds and feel slightly stupid, which is normal. The standard version has 60 items. Most people score between 12 and 18 correct answers. A score above 20 is considered high, and scoring in the bottom quartile doesn't mean much of anything beyond that your associative network might not be primed at that moment. This is not an IQ test, and Mednick himself warned against treating it as a catch-all intelligence measure. It measures one specific cognitive skill, which is the ability to retrieve distant semantic associates under time pressure.
I should mention a practical issue that comes up constantly when people try this online. There are dozens of poorly constructed copies on the internet with wrong answers, incomplete items, or questions where the intended answer is genuinely ambiguous. The original 60-item form is public domain and widely available through academic channels, but most free versions floating around the web have been corrupted by people who thought they knew better. If you are using a version where slice, partner, and butter has bread as the answer instead of the intended peanut, your score is already garbage because the test relies on having a single unambiguous solution for every item. Here is what I wish more people understood about how the RAT actually works under the hood. The test is not measuring creativity directly, despite what every self-help blog says about it. It is measuring divergent thinking efficiency in a very narrow context. Your brain has to suppress the most obvious association and reach for a less common one. When you see mine, salt, and lover, the first word that pops into your head might be something obvious and wrong. The correct answer is dead, as in dead mine, dead salt, dead lover. You have to override your initial response, and that cognitive inhibition component is what separates people who score well from people who don't. Another thing beginners miss: the test is heavily loaded with verbal and cultural knowledge. If English is your second language, or if you grew up in a culture where certain common English collocations are unfamiliar, your score will be artificially low and that does not reflect your actual associative ability. I once had a participant who scored 4 out of 60 on a commercial RAT and seemed genuinely distressed about it. He was a brilliant engineer from Nigeria. Five of his misses were items that depended on American idioms he had never encountered. We ran a version adapted for his background and his score jumped to 41. The test is not neutral.
How to actually take the test properly. Find a clean copy of the original 60-item Mednick form, preferably from a psychology test repository or the journal where it was originally published. Set a timer for 15 minutes. Do not look up answers. Write down your best guess for each triplet and move on. The time pressure is part of the design, not something you can skip. If you give yourself unlimited time, you are no longer measuring rapid associative retrieval, you are measuring something completely different. I also recommend taking it in a quiet environment where you are not distracted. Background noise, phone notifications, or even just a mildly stressful mental state can tank your score by 3 to 5 points regardless of your actual ability. This is not a high-stakes exam, but it is sensitive to situational factors in a way most people do not expect. I have seen people who consistently score in the 16-to-20 range drop to 10 or below simply because they were tired or testing in a noisy lab. If you want to look for a version online, search for "Mednick Remote Associates Test 60 items pdf" or check the Psychonomic Society test archives. Academic databases like PsycTESTS also carry validated versions. Avoid any site that asks for payment to reveal your score, because the answers are easily available and the test itself is free to use.
Get the Full Details
There are legitimate downsides to relying on the RAT as a standalone assessment, and I should be blunt about them. The test has moderate reliability at best, with test-retest correlations typically ranging from 0.50 to 0.65. That means your score can vary significantly depending on when you take it and how you feel. It also suffers from ceiling and floor effects, particularly with highly educated samples where many people cluster near the top of the scale, making it hard to differentiate between strong performers. The broader problem is that associative thinking measured by the RAT is only loosely correlated with real-world creative output. Yes, there is a relationship, but it is far from deterministic. Some of the most creative scientists and artists I know scored mediocrely on the RAT, and some people who ace it have never produced anything notable outside of taking tests. The test is a decent snapshot of one cognitive process, nothing more. If you are looking for a more robust assessment of creative potential, consider pairing the RAT with something like the Torrance Tests of Creative Thinking, which measures divergent thinking across multiple dimensions including fluency, flexibility, and originality. Or use the Alternative Uses Task, which asks you to generate novel uses for common objects and has better predictive validity for actual creative performance than the RAT does. Neither is perfect, but together they give you a much clearer picture than a single 60-question multiple-choice test ever could.
For people who want to practice improving their RAT performance, the honest answer is that you can get better at the test itself through repetition, but that is largely practice effect rather than genuine cognitive improvement. I usually tell people to expect about a 2-to-4 point gain on a second attempt simply from familiarity with the format and some of the more common triplets. That does not mean your brain has become more creative. It means you have learned the test. The original Mednick form and supplementary materials are available through the American Psychological Association's test supplier directory if you need a formally administered version for research purposes. Otherwise, the free academic versions I mentioned earlier are sufficient for casual use. Just make sure you are using a verified copy and not some botched rewrite that will invalidate your results.
