To examine the group of most active high-range test-takers, and in particular the question of their intraindividual score development, all of those having taken 25 or more I.Q. tests scored by I.Q. Tests for the High Range are considered in this report. Personality tests are not counted toward the cut-off of 25, so this count corresponds to that which is currently given in test score reports. Since the "when taken" information is needed to study intraindividual score development across tests and over time, and since "when taken" information is stored at subtest level in the case of compound tests, the below statistics use the subtest scores of possible compound tests, not the compound scores. Compound tests are tests that comprise subtests that have statistical reports of their own, and the "when taken" information resides at subtest level because subtests are sometimes taken at different times (even though one does not receive score reports for the subtests).
There are several other reports on intraindividual score development (not specifically about prolific test-takers) to be found on the Statistics index page listed near the present report.
26 candidates have taken at least 25 tests as defined above, one female and the rest male. About 20 to 21 of them can be described as stable and generally unproblematic. Of the rest (all male) two are clearly unstable and/or unreliable, unable or unwilling to communicate, unable to understand and retain information communicated to them, and unable to keep track of which tests they have already taken and therefore often ordering tests already taken (but the last is true for quite a few "stable" candidates too).
The "rest" also includes about 3 candidates regarding whom there is a mild suspicion (without proof) of practices like also having taken tests under someone else's name, or having used answers found on the Internet (without reporting this to the scorer) in the case of very old, meanwhile discontinued tests that people were discussing on a public forum back then. Furthermore, the "rest" includes one chronic insulter; as a test scorer, one may put up with insulters because they bring in usable data, thus sacrificing oneself for the sake of science so to speak.
To answer the question whether people's scores get systematically higher or lower when taking many tests over much time, for each candidate the correlation is computed between scores and months when the tests were taken, when months are numbered-through such that January 1995 is 1, and so on. The average of these intraindividual correlations is .115, meaning that there is as good as no correlation between score and "when taken" over the 767 tests taken by 26 candidates. In other words, when one takes many tests over a long period, there is as good as no increase in score overall. For information, other reports about intraindividual score development have shown that there actually is an increase in score over the first several tests and the first few years of test-taking.
Now there is a belief among some that if you keep taking tests long enough, you will eventually get a very high score. And of course, the probability of a high (or low) outlier score increases with the number of tests taken. But all scores considered, there is virtually no upward trend. In practice though, people do not honestly show you all of their scores, but like to impress with a single high outlier while withholding the dozens of much lower scores they also have. This is the main reason why score reports by I.Q. Tests for the High Range nowadays contain a qualified average I.Q. based on all of the tests taken. This prevents people from misleading the public with a somehow obtained high outlier score. That kind of megalomanic behaviour is what gives high-range psychometrics a bad name.
On average, 29.5 tests have been taken per person (possible compound tests only being counted at subtest level). For transparency it should be explained that the threshold for inclusion was 25 tests by the normal way of counting, which includes compound tests also counted at total score level; when counting only at subtest level in the case of compound tests, this leads to lower numbers, in actuality ranging from 19 to 43.
On average, the time span from the candidates' first to latest test submission is 145 months. That is, 12 years and one month.
Experimentation in earlier reports on intraindividual score development has shown that the standard deviation is the best indicator of intraindividual spread of scores. For this group, the intraindividual standard deviation is on average 54 protonorm points, and that is somewhat larger than it is over the total group of high-range candidates (reported elsewhere). So, when taking very many tests over much time, one's scores end up somewhat wider apart.
The average of the intraindividual median scores is protonorm 385, or I.Q. 137. This is about 2 I.Q. points below the median of high-range candidates in general. Prolific candidates are less intelligent than the average high-range candidate, and that seems to be caused by a small over-representation of unstable or otherwise problematic cases among them. All of them are above I.Q. 120 though.
On average, candidates took 36 months to reach their personal highest score. Only about 3 or 4 of them took much longer and reached their top score late in their test-taking career. And 4 candidates had their peak in the first month.
On average, candidates reached their highest score with the 7th test. Since possible compound tests are counted at subtest level, from the viewpoint of a candidate that has taken one or more compound tests this may be only the 6th to 3rd test, because candidates tend to see a compound test as one test.