Study of the Relationship Between Achievement Test Results and the Socionic Personality Profile

Part 1 – Tests of Predominantly Logical and Predominantly Emotional Intelligence

V. L. Talanov

Saint Petersburg, January 2019

Introduction

In socionics, a relationship between a person’s socionic profile and an objective profile of particular abilities measured directly by achievement tests had always been presumed, but until 2018 no one had tested it experimentally. Moreover, in mid-2018 a discussion even arose on the Socioforum, during which some fanatical “Neveevites” questioned the very possibility that such a relationship could exist. According to them, abilities (including logical abilities) were one thing, while socionics – supposedly stood on its own.

In October-November 2018, Andrey Parfenov, with the participation of the author of the present article, conducted a pilot experiment on the Socioforum and in his VK community “Cognitive Socionics” to test relationships between socionic personality profiles measured with V. L. Talanov’s 350-item sociodiagnostic questionnaire (form 2018-350-8) and indicators of the Amthauer intelligence test (IST). Although the sample was small (about 40 people), statistically significant correlations were demonstrated between overall success on the test (the IQ indicator), as well as the results of its individual subtests, especially the arithmetic ones, and the magnitude of logical functions (especially Ti) in the participants’ socionic functional profiles (see https://vk.com/soc_experiment?w=wall-162420791_61%2Fall).

Independently of this pilot study, in November-December 2018 V. L. Talanov conducted work to test the relationships between the socionic profile measured by the questionnaire and performance on the Amthauer IST and on the RMET test for recognizing other people’s emotional states from their gaze. A substantially larger sample was used, and comparison of the results of three different methods at once – the socionic questionnaire, the Amthauer test (oriented predominantly toward measuring logical intelligence), and the RMET test (oriented predominantly toward measuring emotional intelligence) made it possible to conduct a much deeper analysis of the results, including separating the influence on them of general intelligence (unrelated to either logic or ethics) from specific logical, emotional, sensory, and intuitive abilities arising from the corresponding characteristics of a person’s temperament but not directly related to the person’s intelligence. The results of this experimental work are presented in this article.

Methodology

Measurement Instruments

To obtain socionic profiles, V. L. Talanov’s 350-diagnostic-item questionnaire “form 2018-350-8” was used, employing the response-processing and socionic-profile derivation procedure standard for all of V. L. Talanov’s sociodiagnostic questionnaires. The participant’s raw responses are first corrected for the individual level of self-assessment, the tendency to answer yes or no more often, and the extremity or caution of ratings (that is, their greater or lesser dispersion from the mean). After this, the responses to each questionnaire item are finally normalized to a mean of zero and a unit sample standard deviation. The initial type profile, whose 16 numbers reflect the degree of proximity of the respondent to each of the 16 standard sociotypes, is obtained as a set of 16 linear correlations between the participant’s finally normalized responses and 16 vectors of previously known averaged typical responses to this sequence of questions characteristic of each of the 16 standard sociotypes. So that subsequent operations can be performed on the resulting correlation values as on quantities distributed according to a normal distribution, these correlation values are additionally replaced by their Fisher transforms. The Fisher transforms of the 16 linear correlations constitute the primary (raw) type profile of each participant. The type profile is converted into a 15-number trait profile, and the trait profile, in turn, is easily recalculated into an extended 12-number functional profile (all conversion operators can be seen, for example, here: http://sociotoday.narod.ru/val_funk.docx - see the “Methodology” section).

During the study, the questionnaire was posted online on a website from which it could be downloaded for completion. Anonymous respondents (who wished to receive their detailed socionic summary) sent their completed questionnaire for processing from an anonymous email address created specifically for this purpose, and on that same day they received at the same address, in response (as a reward for their work), their detailed socionic summary. As a condition of typing, the respondent was also required to complete two additional tests (IST and RMET) on third-party websites (with which the author of the study had no connection), while also entering their results on these tests into the questionnaire. The absence of falsification in the reported results for the Amthauer IST (the most labor-intensive test to complete) was controlled by the fact that on the test website every result is stored under a particular name and number and can therefore be checked; respondents entered in the questionnaire the URL link containing the number of their result and, as their anonymous nickname when taking the IST, also entered – for additional identification - the anonymous email address created specifically for correspondence concerning the socionic questionnaire. Thus, while preserving complete respondent anonymity, the possibility of falsification or errors (for example, if a participant indicated an incorrect age range when taking the IST, or skipped some of the “troublesome” subtests) was excluded in the IST experiment. The RMET is much shorter, less tiring, and more interesting than the Amthauer test; in this case the motive for falsifying one’s result is minimal, so the author did not additionally recheck the RMET results reported by respondents after they had taken the test on a third-party resource.

For the Amthauer test (Form A of this test, one of two parallel forms), respondents were directed to the third-party internet resource https://psytest24.ru/amthauer/index.php?iter=1, which allows reported results to be verified, and for the RMET (recognition of a person’s emotional states from their gaze) – to the third-party internet resource http://www.aspergers.ru/node/232#.

Sample Size

At the end of the experiment, 142 respondents had completed all three diagnostic methods (questionnaire, IST, RMET); an additional 65 people had completed only two methods (questionnaire, IST), and another 84 people had completed the other pair of methods (questionnaire, RMET). In total, 291 participants took part in the experiment, completing the questionnaire and at least one of the two additional tests.

Indicators Analyzed

Fourteen indicators from the two additional tests were compared with the socionic profiles – type, trait, and functional profiles (by calculating linear correlations with the numbers of these profiles). What are these indicators? First, there is the RMET result itself (an integer from 0 to 36; in the actual sample, in the range from 15 to 34, indicating the number of emotional states correctly selected from a list when evaluating 36 photographs of actors’ eyes). Of the remaining 13 indicators, eleven are the ten direct results of the Amthauer IST (the overall IQ score and separate scores for each of the 9 subtests), plus one more indicator – the sum of the scores on subtests 1, 3, 5, and 6 (designated in this work as indicator “А”). Subtests nos. 1, 3, 5, and 6 had been selected by the author for the additional Amthauer indicator “А” even before the start of the large experiment with it, on the basis of analysis of the content of its subtests and the results of Andrey Parfenov’s preliminary pilot experiment, as those most likely to be associated with a positive logic-ethics balance in respondents’ socionic profiles. Two additional indicators used in the study are integral indicators reflecting the interaction between IST and RMET results. One of them, equal to the difference between the normalized values of indicator А and the RMET indicator, according to the experimental hypothesis should characterize the purely logical component of a person’s abilities, unrelated to his general intelligence, which is independent of the logic-ethics balance (because when the two specified indicators are subtracted, their logical component is added while the intellectual component is canceled out). The second indicator, conversely, equals the sum of the normalized values of indicator А and the RMET indicator – it reflects, conversely, a predominantly purely intellectual component, which according to the preliminary experimental hypothesis is unrelated to respondents’ socionic logic-ethics balance (because the role of the personality’s logical-emotional balance in the combined results of tests of logical and emotional intelligence should be canceled out to a substantial degree).

All respondents’ socionic profiles were used in calculating correlations with the intelligence-test indicators in three different versions of their normalization, individualized for each respondent.

First – raw socionic profiles were used (the type profile of each participant, consisting of Fisher transforms of the correlations of that participant’s responses with the 16 vectors of standard type responses, and the trait and functional profiles derived from it).

Second, correlations were also calculated for profiles that had undergone individual normalization for each participant up to a unit standard deviation of each type profile (to obtain it, the values of each participant’s raw type profile are divided by the standard deviation of the set of 16 values of that raw profile; the trait and functional profiles are obtained from the normalized type profile by the usual procedure).

Third, correlations were also calculated for profiles that had undergone optimized (incomplete) normalization, namely division not by the standard deviation (sigma) of the raw profile, but by sigma raised to the power “а”, where 0<a<1 (specifically, а=0,21). The point is that the sigma of the raw profile depends simultaneously on several factors. First, there is a group of subfactors affecting the thoughtfulness and reliability of the respondent’s answers to questionnaire items. The greater the respondent’s diligence, intelligence, life experience, as well as logic and rationality, the higher, as a statistical tendency, the sigma of the resulting raw type profile. All these subfactors combine into one jointly acting factor representing some degree of interference (noise in the results), and ideally its influence on the height of the type profile should be eliminated completely. But there is also a second factor, reflecting the genuinely different degree of a person’s functional heterogeneity across different people. Some people differ very markedly from the average level on a number of their socionic indicators (the sigma of their raw socionic profile is objectively large), but there are objectively also people in whom almost all functions of the psyche genuinely differ very little from the average level - that is, the personality characteristics of such people are objectively weakly expressed, and accordingly their type, trait, and functional profiles should objectively be smoothed, that is, have a small sigma.

In processing Talanov’s socionic questionnaires, there is, in principle, a special technique that makes it possible to separate the influence of these two factors and selectively (by means of a special correction individually sized for each respondent) minimize the influence of only the first one (the artifactual factor) on the resulting raw type profile. However, this technique requires the introduction of a special control scale into the questionnaires and is not used in questionnaires of the latest series, including the socionic questionnaire used in the present study. Nevertheless, a partial reduction of the parasitic influence of the first (artifactual) factor on the height of the type profile, while retaining the principal contribution of the second factor, is also achieved without individual corrections - by means of a general group correction, namely by dividing each raw profile by its sigma raised to the power “a”, where а=0,21 (see http://sociotoday.narod.ru/vysota.docx). This transformation of raw socionic profiles maximizes their correlations with external tests, which in this case should be higher than the correlations both for a raw type profile not corrected for sigma at all and for one fully corrected for sigma, up to bringing the profiles of all respondents to the same unit sigma (that is, unit standard deviation).

Thus, dividing the original profiles by their sigma raised to the power 0,21 yields the third type of profiles (optimized to maximize correlations with external tests), for which correlations with the block of external intellectual achievement tests were also calculated in the present work.

Equalization of the Experimental Sample for Equal Representation of Sociotypes (Mathematical Pseudo-Equalization Procedure)

Correlations between the block of intelligence-test indicators (including RMET) and the block of socionic indicators were calculated in two ways. In the first case, no corrections were made for heterogeneity of the experimental sample in the proportions of sociotypes represented in it, and correlations were calculated directly between the sample indicators. Heterogeneity of the sample in the proportions of sociotypes can introduce some distortions (not very large) into the resulting correlations.

In the second case, before correlations were calculated, the subsample of 142 participants who completed all three tests was equalized by means of a special mathematical procedure to obtain an “as if” equal representation of all 16 psychotypes. The corresponding mathematical “pseudo-equalization” procedure is fully justified mathematically and consists of several stages. First, the proportion of each sociotype in the sample is calculated, with a respondent of each sociotype contributing to the determination of this proportion with a weight equal to the height of the corresponding maximum peak in the respondent’s raw type profile. Next, the square root is taken of the resulting proportion of each sociotype in the sample; finally, for each respondent, all normalized indicators (both intellectual and socionic) intended for subsequent calculation of intercorrelations are divided by the square root of the sample proportion of the sociotype to which that respondent belongs, judging from the maximum peak of the type profile. Applying this mathematical procedure to the data (before calculating correlations) brings all subsequent intercorrelations among them to the numerical form that would have been obtained if all sociotypes had been represented completely equally in the experimental sample.

Results and Discussion

Table 1 presents, for each of the 14 indicators of the “intellectual block,” the sum of the squares of the linear correlation coefficients between that indicator and all 350 items of the sociodiagnostic questionnaire (for the subsample of 142 participants, without pseudo-equalization for equal representation of sociotypes). The proportions of sociotypes in the subsample differ by no more than a factor of 4; with this not very large heterogeneity, the correlations in the sample differ little from the correlations in an ideal sample with equal representation of sociotypes.

The sums of squared correlations in Table 1 are proportional to the variance commonality of each of the 14 indicators of the “intellectual block” with the socionic (temperament-psychological) block. It is noteworthy that among the Amthauer subtests, subtests nos. 1, 3, 5, and 6 show the greatest variance commonality with the socionic questionnaire. Consequently, the additional indicator “А,” equal to the sum of the results of these four subtests, was selected quite justifiably as the most indicative marker of the influence of the socionic profile on achievement-test results. It is also noteworthy that the variance commonality of the composite indicator А with the socionic questionnaire (equal to 4,36) is greater than 3,62 - the variance commonality with the questionnaire for the final IQ indicator measured in the Amthauer test by averaging the results of all 9 subtests. The lowest variance commonality with the questionnaire is shown by the 8th Amthauer subtest (tasks involving mental rotation of cubes – supposedly measuring spatial relations). Ambiguity in interpretation when performing this task evidently makes it of low validity. Commonality with the questionnaire (that is, with the psychological characteristics of personality) is also low for subtests no. 7, no. 2 (an excessively easy task), and no. 4 (poorly constructed answer keys for determining correct answers).

The RMET (determining emotional state from other people’s gaze) has practically the same high variance commonality with the psychological side of personality as the composite indicator А from four Amthauer subtests. Thus, the integral indicator equal to the difference between the normalized А and RMET indicators, that is, the difference in success on logical and emotional achievement tests, is of considerable interest because it may claim to reflect only the psychological (socionic) component, with the component of a person’s general intelligence and general life experience canceled out.

Table 1

Tables 2, 3, and 4 show correlations between socionic-profile indicators from the questionnaire and the main indicators of the achievement tests intended to measure components of logical and emotional intelligence (with general intelligence also participating in each of these cases, that is, in particular, a person’s general development, life experience, and the speed and efficiency of information processes, without dividing them into predominantly logical and ethical ones).

All correlations with Amthauer test indicators (N=207) whose absolute values exceed 0,14 are reliable in their sign with a confidence coefficient of 1-2а=0,95, and those exceeding 0,18 - with a confidence coefficient of 0,99.

Correlations with combined indicators of the Amthauer and RMET tests (N=142) whose absolute values exceed 0,175 are reliable in their sign with a confidence coefficient of 1-2а=0,95, and those exceeding 0,225 - with a confidence coefficient of 0,99.

Table 2 reflects correlations for indicators of raw socionic profiles, Table 3 – profiles with full normalization to a standard equal height in all participants, and Table 4 – profiles optimized for maximum correlation with external tests. Indeed, for the profiles in Table 4, correlations with tests of the intellectual block are the largest (among the three specified types of profiles, that is, completely unnormalized, fully normalized, and partially optimally normalized).

Table 2

Correlations of respondents’ socionic-profile indicators with results of Amthauer subtests (intelligence test battery) and the RMET test (recognition of the emotional expression of 36 faces)

Table 3

Correlations of respondents’ socionic-profile indicators with results of Amthauer subtests (intelligence test battery) and the RMET test (recognition of the emotional expression of 36 faces)

Table 4-1

Correlations of respondents’ socionic-profile indicators with results of Amthauer subtests (intelligence test battery) and the RMET test (recognition of the emotional expression of 36 faces)

Table 4-2

After artificial mathematical equalization of the sample by representation of TIMs (division of all correlated indicators by the square root of the frequency weight of the respondent’s TIM in the sample)

Correlations of respondents’ socionic-profile indicators with results of Amthauer subtests (intelligence test battery) and the RMET test (recognition of the emotional expression of 36 faces)

The last table presented, Table 4-2, is the most informative for considering correlations between socionic indicators and achievement-test indicators.

In particular, it shows that the highest scores on the Amthauer test (and especially on its four most valid subtests) are obtained by the four Scientist types, while the lowest are obtained by the four Social types. Overall, 38% of success in performing the Amthauer test is determined by a person’s socionic profile, while as much as 54% of performance on the four most valid subtests, nos. 1, 3, 5, and 6, is determined by it. Only the remaining share of success is associated with intelligence itself, the person’s experience, and the operation of various random situational factors while taking the test. The leading contributions to success are made by the trait poles of Democrats, Logicians, Result types, Merry types, Intuitives, and Statics (in precisely this descending order), while in the language of functions, Ti, Te, Ne, Qi, and De contribute to successful test performance, whereas Fi and Fe hinder it.

And if the difference between the normalized results of the two groups of tests, indicator “А” of the Amthauer test and the overall RMET result, is considered, then almost 76% of the variance of this quantity proves to be programmed into the person’s socionic profile.

The test-retest reliability coefficient of the Amthauer IST (retest interval — 1 year) — 0,83-0,91. The reliability coefficients of parallel forms — 0,95 (source: Amthauer Intelligence Structure Test. Material from Psylab.info - encyclopedia of psychodiagnostics http://psylab.info/%D0%A2%D0%B5%D1%81%D1%82_%D1%81%D1%82%D1%80%D1%83%D0%BA%D1%82%D1%83%D1%80%D1%8B_%D0%B8%D0%BD%D1%82%D0%B5%D0%BB%D0%BB%D0%B5%D0%BA%D1%82%D0%B0_%D0%90%D0%BC%D1%82%D1%85%D0%B0%D1%83%D1%8D%D1%80%D0%B0).

Consequently, the reproducible variance associated with the test constitutes from 0,69 to 0,83 of the test’s total empirical variance.

For the difference between IST and RMET indicators, the proportion of reproducible variance cannot be higher, because the test-retest reliability of the RMET (consisting of only 36 tasks), according to the laws of mathematical statistics, lies in an even lower interval.

Thus, the sum of squared correlations equal to 0,76 (that is, accounting for about 76% of the total variance), measured between the 15 quasi-orthogonal socionic traits and the difference between the normalized IST and RMET values, is very close to a complete 100% reflection of all the variance of the IST and RMET tests that is not associated with general intelligence and random situational factors in taking the tests, but is instead associated with a person’s abilities conditioned by innate characteristics of temperament.

From this it should be concluded that a person’s abilities associated with innate temperament (but not with life experience), whether logical, emotional, or other, that in any way actually influence performance on the IST and RMET achievement tests are fully (practically without exception) reflected in the person’s socionic profile obtained using Talanov’s questionnaire (350 diagnostic items).

Appendix. Items of the 2018-350-8 Sociodiagnostic Questionnaire Most Closely Correlated with Individual Indicators of the IST and RMET Intellectual Achievement Tests

Table 5-1. Correlates of the Overall Amthauer Test (IST) Score

Table 5-2. Correlates of Scores on the IST Subtest “Sentence Completion”

Table 5-3. Correlates of Scores on the IST Subtest “Exclusion of the Odd Word”

Table 5-4. Correlates of Scores on the IST Subtest “Analogies”

Table 5-5. Correlates of Scores on the IST Subtest “Generalization”

Table 5-6. Correlates of Scores on the IST Subtest “Memory, Mnemonic Abilities”

Table 5-7. Correlates of Scores on the IST Subtest “Arithmetic Problems”

Table 5-8. Correlates of Scores on the IST Subtest “Number Series”

Table 5-9. Correlates of Scores on the IST Subtest “Spatial Imagination on the Plane”

Table 5-10. Correlates of Scores on the IST Subtest “Spatial Relations”

Table 5-11. Correlates of Scores for Indicator “А” of the IST (Sum of the Scores of the Four Most Valid Subtests)

Table 5-12. Correlates of RMET Scores (Achievement Test for Recognition of Emotional Facial Expressions in Photographs)

Table 5-13. Correlates of the Difference Between Normalized IST and RMET Achievement-Test Scores

Balance of logical and emotional-ethical abilities

Table 5-14. Correlates of the Sum of Normalized IST and RMET Achievement-Test Scores

General intelligence (equally weighted for logic and ethics)

References

1. Andrey Parfenov’s pilot experiment testing correlations between indicators of the functional socionic profile and subtests of the Amthauer IST: https://vk.com/soc_experiment?w=wall-162420791_61%2Fall
2. Amthauer test (IST), Form A: https://psytest24.ru/amthauer/index.php?iter=1
3. RMET (recognition of a person’s emotional states from their gaze): http://www.aspergers.ru/node/232#
4. Amthauer Intelligence Structure Test. Material from Psylab.info - encyclopedia of psychodiagnostics: http://psylab.info/%D0%A2%D0%B5%D1%81%D1%82_%D1%81%D1%82%D1%80%D1%83%D0%BA%D1%82%D1%83%D1%80%D1%8B_%D0%B8%D0%BD%D1%82%D0%B5%D0%BB%D0%BB%D0%B5%D0%BA%D1%82%D0%B0_%D0%90%D0%BC%D1%82%D1%85%D0%B0%D1%83%D1%8D%D1%80%D0%B0
5. V. L. Talanov. Socionic and Non-Socionic Components in the Height of the Type Profile Empirically Obtained from Sociodiagnostic Questionnaires (July 2017): http://sociotoday.narod.ru/vysota.docx (download)
6. V. L. Talanov. Validity of V. L. Talanov’s Psychodiagnostic Questionnaires in Light of the Empirically Identified Semantic Content of Socionic Functions of the Psyche (2017): http://sociotoday.narod.ru/val_funk.docx (download)
7. List of V. L. Talanov’s recent articles: http://sociotoday.narod.ru/index4.html, http://sociotoday.narod.ru/index3.html

© V. L. Talanov 2019

Free reproduction is permitted provided that a URL link is given to one of the two original sources of the author’s article, http://www.sociotoday.narod.ru/socionic_IST_RMET.docx or http://otkroysebya.ru/socionic_IST_RMET.docx.