Does practicing Ravens Progressive Matrices tests inflate your IQ score?

Question

I'm 16 years old and a year ago I got obsessed about IQ. I took a Raven IQ test and scored 115. Then during the autumn and winter of last year I took some more IQ tests using RPM and scored 126 on almost all of them. I took the Raven IQ test again in January of this year. I scored 126 on it. I wasn't convinced. So I researched the most accurate online IQ test and found the Norwegian Mensa Practice test took it and scored 128. A couple months after that I took the Mensa Denmark practice test formerly known as iqtest.dk. I scored IQ 121 on it.

Is it because I took so many Raven IQ tests my score is so high? Does that mean that my IQ isn't in the 120's? Where would you put my Non-Verbal/Spatial IQ? Also I'm planning to take the WAIS next year and don't know what to do. I want my scores to be 100% true and not at all inflated. Would this affect it? And if so how can I change that?

Edit: Just wanted to add that I didn't get feedback on the correct answers.

Answer 1

Yes, practice even as simple retest improves scores in matrices-like tests:

Raven’s APM scores increase significantly on repeated testing without any targeted training (e.g., Bors & Vigneau, 2003; Bors & Forrin, 1995; Denney & Heidrich, 1990).

• Bors DA, Vigneau F. The effect of practice on Raven’s advanced progressive matrices. Learning and Individual Differences. 2003;13(4):291–312.

• Bors DA, Forrin B. Age, speed of information processing, recall, and fluid intelligence. Intelligence. 1995;20(3):229–248.

• Denney NW, Heidrich SM, Training effects on Raven's progressive matrices in young, middle-aged, and elderly adults. Psychol Aging. 1990 Mar; 5(1):144-5.

I'm not surprised by this finding because the number of conceptually different problems in such matrices tests is limited. They questions are all basically variations on a few base patterns. The latter part is claim from memory, but I've read it somewhere besides my own observation; let me find where. (Also, I should try to quantify the improvement retest gives.)

I couldn't find the original place where I've read about the structure of Raven's, but I found newer research that found improvement in restest in particular problem types:

The (small) sample here was of 13-year olds... so this should be pretty relevant to your age. On the other hand, I'm not sure how much the finding extends (quantitatively) to adults.

Some researchers went as far as explaining the Flynn effect by this kind of better rule-based inference underlying retest improvement in Raven's:

We present a new model of the Flynn effect. It is proposed that Flynn effect gains are partly a function of the degree to which a test is dependent on rules or heuristics. This means that testees can become better at solving ‘rule-dependent’ problems over time in response to changing environments, which lead to the improvement of lower-order cognitive processes (such as implicit learning and aspects of working memory). These in turn lead to apparent IQ gains that are partially independent of general intelligence. We argue that the Flynn effect is directly analogous to IQ gains via retesting, noting that Raven's Progressive Matrices is particularly sensitive to both the effects of retesting and the Flynn effect.

And a bit more from the paper itself:

The biggest Flynn effect gains have been recorded on tests of fluid intelligence, in particular the Raven's Progressive Matrices (five-seven points per decade; Flynn, 1987, 2007, 2009). The smallest gains typically occur on measures of crystallized knowledge (Jensen, 1998a). It must be noted that not all measures of IQ show long-term gains. For example, purely psychophysical elementary cognitive tasks (ECTs), such as inspection and reaction times, show no Flynn effect (Nettlebeck & Wilson, 2004; Silverman, 2010). [...]

Studies by Verguts et al. (1999) and Verguts and De Boeck (2002) [...] have found that in experiments, participants do repeatedly reapply a small number of rules when attempting to solve RPM-type items. The rule induction process seems to be characterized by the sequential sampling of rules, each of which have a certain probably of being resampled (Verguts et al., 1999). Furthermore, the speed with which individuals are able to sample and induce rules is strongly related to prior task exposure (Verguts & De Boeck, 2002). These studies indicate that narrower abilities matter in solving RPM items, and that furthermore, reliance upon these narrow abilities increases with increasing test experience. This is consistent with the phenomenology of both retesting and Flynn effects [...]

• Verguts, T., De Boeck, P, & Maris, E. (1999). Generation speed in Raven's Progressive Matrices Test. Intelligence, 27, 329–345.
• Verguts, T., & De Boeck, P. (2002). The induction of solution rules in Raven's Progressive Matrices Test. European Journal of Cognitive Psychology, 14, 521–547.

I'm guessing I've read one the Verguts' papers, or at least something else citing their results (in the past).

Answer 2

A general principle is that skill transfer is domain specific and based on common elements.

So if you've learnt something about the Raven's items, I would expect improvement and skill acquisition transfer to be ordered from greatest to least as follows:

• doing the actual Raven's test with the exact same items again
• doing other versions of Raven's using different item sets
• doing other tests based on similar matrix-style abstract reasoning items
• doing other tests measuring abstract reasoning using other item styles

If you've only studied Raven items, then I doubt you'll get much transfer to the broad array of item-types included in other measures of intelligence.

In general, practice effects (especially, item specific practice effects) tend to decline over time. For example, if you wait six months, a year, or more, then practice effects will generally be less of a concern.

@Fizz also provides references and arguments to support the claims that Raven's may be particularly predisposed to practice effects given the inherently small number of item types and patterns. This can be contrasted, for example, with measures of verbal ability such as advanced vocabulary items which would take serious study to obtain score improvements across item sets.

Practice effects will also generally be amplified if you have actively studied the items or otherwise obtained feedback on correct answers. Perhaps this explains practice effects on the Ravens in that often people will get intrinsic feedback from the test about whether a response seems correct which allows some test-takers to infer rules about the test.

So in short, experience with the Raven's might bias subsequent Raven's test scores. But if you move away from matrix-style abstract reasoning measures, then transfer effects are unlikely to be an issue.