As a child who was tracked into gifted classes since elementary school, I vividly remember whole lessons dedicated to strategically taking standardized tests.

An example of a tricky question we learned to answer correctly: "1. Big is to bigger as little is to ____" is "A. tiny," not "B. big", because the question is looking at changes in intensity, not just increasing size.

We were also taught general test-taking strategies like budgeting time for an entire test, making an educated guess on, or skipping & returning to, tricky questions, and so forth. Getting explicitly taught these skills made me perform better on standardized tests. And families like mine, with the socio-economic resources to get their kids taught these skills, will do better on standardized tests--including IQ tests--as a result.

I remember during tests I'd get stuck at the hardest problems and stubbornly solve them only to run out of time on the easy questions and haphazardly answer them until the "pen up" was announced. At some point a teacher taught us how to maximize our scores resembling the strategy you mention and yes, it did help. We weren't smarter by any means, but had a better strategy to taking tests.

Easiest way to "game" it is to take it multiple times or practice/cram for it.

At least around here it seems people are allowed to take the Mensa test multiple times.

I never took the official one but I understand a lot of effort goes into constructing test sets that doesn't benefit from cramming but (except for getting used to the format, which you can do online) only on raw "processing power".

I'm not a scientists or a psychiatrist so take my word for just that: my understanding.

Also I note that from what I hear on HN it seems to be that in USA IQ tests contain both spelling and history questions which - in my ears sounds like something very different from a raw IQ test.

There are lots of tests out there that all correlate with general mental ability. Some of these, like mental rotation tests, benefit a lot from practice. Some of them, like Raven's Progressive Matrices, don't. The tests used in modern IQ tests are the ones where practice don't give much benefit.

Everyone does better at IQ tests if they repeatedly practice IQ tests. A lot of those rely on you being fast at identifying a repeating pattern but there's only a finite amount of them. You can learn a few of the complex ones that are harder for you and bruteforce a sequence into fitting it, and anyone can do that.

OK, I'm sure people will have better results on their 10th test than the first.

So I admit you can "game" the test somewhat, but I expect the effect to be minor, and largest for the more intelligent.

If there is real data on this I'd enjoy seeing it.

> You can learn a few of the complex ones that are harder for you and bruteforce a sequence into fitting it, and anyone can do that.

Anyone smart can do that :)

Here's an interesting randomized controlled study that shows nominal improvements when retaking the Wechsler Adult Intelligence Scale-IV after 3 months and 6 months. https://sci-hub.tw/10.1080/13854046.2012.659219 There's a long discussion of how significant, if at all, those improvements were, which I only quickly glanced over. Apparently practice effects are well established, at least in principle, and there are some well known analytical and examination methods (albeit contended--part of the discussion compares and contrasts them) that attempt to correct for them.

People in this thread talk alot about gaming multiple-choice tests, but when I was given the Wechsler Intelligence Scale for Children at age 11 I mostly remember alot of pictorial and physical work (puzzles, blocks), mental arithmetic (e.g. 2, 3, 4+ digit sums), verbal memorization, and other visual and oral interaction, not much written material. And it was administered by a developmental psychologist in a 1:1 setting. I have no doubt such an examination might be susceptible to practice effects, but I'd be surprised if it were as significant as for common written examinations. In the county (state?) I went to school the gifted programs required a formal Wechsler or Stanford-Binet exam, which I gather was a far more common requirement than today. Perhaps that's too expensive of a gateway when you have legions of parents knocking down the door to get their child tested for specialized programs, or when concerned about bias--test every kid, rather than relying on teachers identifying a small number of children for additional screening and examination.

The simplest way to 'game' multiple choice based forms of IQ test is to guess the multiple choice options when you don't actually know the answer. This isn't likely to allow you to fake genius, but will boost your score relative to a majority who don't attempt all questions unless there's a very well-designed negative marking scheme.

Willingness to guess answers to ensure the whole test is completed probably is associated with many of the success outcomes associated with higher IQ scores, but it isn't exactly what most people would consider to be intelligence.

You learn to look after patterns.

No different that "grinding" leetcode problems, IMO.

If you do enough, you start to see a lot of similar patterns, over and over.

From there, even if you don't truly know what pattern is right, you can at least do some exclusion - like with multiple choice.