Why I don't believe timed IQ tests measure g

I'm not a psychologist, but you know what they say about assholes and opinions... So I wanted to divulge my opinions about timing IQ tests because I feel particularly insecure about being relatively slow today. IQ tests are often timed for 2 reasons:

1. Convenience

2. Because smarter people "get" things faster

Well, I agree with 2... that smart people get things faster.

But there still exists a problem timing IQ tests. When you time IQ tests, you introduce several extraneous variables:

- One such variable is a "thinking style." Some people rush to find quick answers and miss details. Now, for every timed IQ test, there will probably be an optimal level of urgency, or, in other words, an optimum depth of analysis. Analyze the question too superficially and you'll pick the red herrings and miss all the tricks in a trick question. Analyze the question too deeply and you run out of time while looking for the fifth different pattern in a simple number sequence (which has an "obvious" solution to everyone else and 3 obvious solutions to you), just to be sure you found the right answer. So, besides g, you're also measuring the testee's subjective feeling of being rushed (and his inclination to rush).

- Another variable is resistance to time stress. My experience has been that everyone's brain shuts down to different degrees when they realize they have only 10 seconds to solve a problem. Instead of inspecting the figure series people to different degrees paralyze themselves by thinking "oh shit, I only have 5 seconds left!"

- Reading speed. This may be related to inclination to rush. People read differently. Some people read very quickly and get the gist while missing the details. As I've said before, my feeling is that for every IQ test, there is an optimum degree to which you should rush (and read fast) that would give the highest IQ score. Reading speed is different because it is more dependent on a person's natural inclination to either read for detail or to get the big picture. A detail oriented person might take longer to inspect a figure sequence but will be more likely to discover the important details of the sequence.

- Expectation of item difficulty: In a figure sequence test, the order in which your brain registers certain details of the figures in the sequence should almost be random. There is a probability that the brain registers the "right" details during a given time. But nevertheless, if a person has the expectation that the problem will be difficult, he'll probably try to find a complicated pattern in details that are irrelevant to the "correct" pattern without noticing other details of the pattern that would lead him to finding the correct pattern. By not timing the test, you eliminate this concern because the testee will then feel free to take his time and inspect all the details of the figures before deducing a pattern.

- Speed of darkening ovals on a scantron sheet.

- Ability to focus. Figuring out patterns is fun to me. But doing this under timed conditions stresses me out and exhausts me. I often find my mind starts to wander and I have to reign it in.

My overall opinion on this matter is that you should give people plenty of time to finish a test (enough time that 95 % of people have at least chosen an answer to every item). Such a test would produce a much more reliable measure of g. If you insist on including a measure of mental speed, I would say the way to do that is by giving a person a series of simple tasks and measuring the amount of time it takes the person to do these simple tasks (barring obvious distractions, etc.), not by restricting the amount of time the person has to complete a series of tasks (which introduces unnecessary time stress).

So Jensen commented that the time it takes a person to solve a problem is related to his/her intelligence and to the problems difficulty. It makes sense that if you gave a smarter person an untimed test he would not rush through items but also naturally "get" the patterns faster than a dumber person. Thinking style would only minimally interfere because when the test is not timed, everyone is more inclined to be careful in inspecting figures, etc. However, as soon as you inform the person that the test is timed, the extraneous variables will start to appear (every person will start feeling time pressure and start analyzing things more/less superficially dependent on their natural inclination to analyze in depth or shallowly, etc.). This is why this latent response time is actually a big red herring.

ALSO: Just so we're absolutely clear, I don't think I do badly on strictly timed IQ tests (12 questions, 2 minutes type tests), just not nearly as well as I do on less strictly timed IQ tests (tests timed like the GRE, etc.)