Complete Psychological Portrait of the Sensory-Logical Introvert (SLI) - Results of an Experimental Study

V. L. Talanov

St. Petersburg, December 2011

Introduction

This article is part of a large new series of our works devoted to the detailed experimental study and description of the empirical psychological portraits of the 16 basic socionic psychotypes, the 8 socionic functions (in their various positions), and the poles of socionic traits. The results of all works in the series (including the results presented in the tables of this article) were derived from mathematical-statistical analysis of respondents’ answers to psychological questionnaire items (more than 5,500 distinct questionnaire items in total). These responses were obtained from more than 5,000 different respondents with preliminarily determined socionic psychotypes.

Psychological types (essentially, independent, non-overlapping sectors of multidimensional psychological space) gradually became the primary, basic concept in empirical socionics. In socionics, they are most often diagnosed as wholes. In American type theory, which is very close to socionics in the resulting structuring of psychological space, this is not at all the case: the basis is invariably taken to be not the psychotypes but, on the contrary, four basic traits. These traits are extraversion-introversion, rationality-irrationality, logic-ethics, and intuition-sensing. Each trait constitutes a hyperplane in psychological space; the four hyperplanes ultimately divide the space into 16 sectors, each of which received the name of one of the 16 psychotypes. This approach was also used both in the works of Carl Gustav Jung and in the early theoretical constructions of the founder of socionics, Aušra Augustinavičiūtė. Subsequently, however, socionics in Russia, Ukraine, and a number of other European countries came to rely in the diagnosis of psychological types not on four-factor diagnostic questionnaires but on live expert diagnosis of types in their “holistic” expression. Despite all its obvious disadvantages from the standpoint of diagnostic standardization, this tradition produced undeniable scientific advantages. The “advantages” in this case consisted in the fact that, within the emerging socionic culture, it became easier to note the most diverse characteristic properties of psychotypes arising at the intersection of the four dichotomies. These properties were studied rather than ignored, although they were often, in a sense, newly formed properties in relation to the semantic content of the four basic diagnostic traits. Thus, for example, it was found that the psychotypes of the second and third quadras differ from those of the so-called “peripheral” quadras by a new property, decisiveness-judiciousness, which cannot be reduced to the four basic traits; accordingly, the same sensing can behave differently in “decisive” and “judicious” psychotypes - in one case as aggressive “extraverted sensing,” and in the other as “introverted sensing,” which differs very substantially in semantic content. Differences were similarly identified between manifestations of extraverted and introverted intuition, extraverted and introverted ethics, and extraverted and introverted logic. The theoretical models of socionics developed accordingly, incorporating new concepts. This development did not occur in type theory, where rigid constraints at the diagnostic “input” prevented fundamentally new conclusions from being found at the “output.” Unlike type theory, socionics for a long time proceeded by empirically “feeling out” the most objective boundaries between psychological types; that is, in constant living contact with natural truth, it formed optimal boundaries between sectors of psychological space, without hesitating to take into account, when drawing these boundaries, newly discovered properties of types that could not be directly reduced to their differences in extraversion-introversion, rationality-irrationality, logic-ethics, and intuition-sensing. Was it good that this was so? Yes, very good. The point is that the substantive content of the four basic traits originally used in diagnosis was never “God-given,” but represented a preliminary result of the same empirical observations by the founding researchers of the diversity of human manifestations. There is, for example, no reason why intuition should manifest identically in all types of the so-called “intuitive” pole (today we know that these manifestations are highly distinctive in each of the eight types); however, the founders’ initial “rough hypothesis,” especially beginning with Briggs and Myers, prescribed that, within this initial “rough hypothesis,” intuition should manifest in the same way in different types, and Briggs-Myers diagnosis was likewise aimed at detecting exclusively the traits common to types belonging to the same dichotomous pole. Accordingly, type theory distorted the boundaries between types, shifting them away from the locations of their natural attraction. Another example is the rationality-irrationality axis, which, likewise being unambiguously fixed in the original Briggs-Myers diagnostic questionnaires, prescribed identical behavioral manifestations of rationality or irrationality for all types - although today we already know well that irrationality manifests somewhat differently in intuitives than in sensors, and somewhat differently in extraverts than in introverts (for example, if improvisation is the key feature of extraverted irrationality, then passivity and laziness are the key features for introverts, although in the broader sense both manifestations arise from a deficit of judging functions and the resulting deficiency of planning). Nevertheless, such nuances were ignored in type theory, leading to incorrect placement of the boundaries between types on the basis of purely voluntaristically selected “diagnostic” properties arbitrarily chosen and fixed in questionnaires. As a result, any potentially new properties of psychotypes and their groups that could not be reduced to the diagnostic questions originally embedded in the Myers-Briggs questionnaire were blurred and went undetected in research.

Socionics avoided these shortcomings in the course of its development. As a result, among all existing psychological classifications, the boundaries between sectors of multidimensional psychological space that were gradually felt out by its expert “public opinion” today apparently correspond best to the real physiological factors of brain activity that objectively underlie the division of psychological diversity into psychotypes. The adequacy of the socionic classification is also emphasized by the fact that it is precisely within the boundaries between types accepted in socionics that new properties are detected that are generated nonlinearly at the intersection of several “basic” traits (from physiological considerations, precisely this kind of generation should be expected from “correct” models of the organization of psychological space).

Socionics often criticizes itself for the low agreement of expert type diagnoses among different experts. It should be noted, however, that this is not a critical deficiency that calls into question the achievements of socionics as a science. When reading numerous descriptions of psychotypes by different socionic authors, it is not difficult to see that, for each type, authors mostly note the same characteristic manifestations. This knowledge and these observations are indisputable scientific truths; they coincide across different authors and arise from their observation of numerous strongly expressed cases that are “obvious” from the standpoint of type diagnosis. But this knowledge indeed cannot always be translated in the reverse direction, from the general to the particular, when it is necessary to diagnose the type of a subject whose type is not so strongly expressed. It should be noted that agreement of expert diagnoses is low even among representatives of the same socionic school. If we set aside a number of obvious quasi-socionic “reinventors of the wheel” and fraudulent amateurs, inventors of their own socionics who present themselves as experts, then the spread of diagnoses is due not so much to differences among schools in their views of type boundaries as to the fundamental limitation of any expert’s ability to take into account and weight simultaneously the diagnostic contribution of the many manifestations of a subject that are important for type diagnosis. Our studies, discussed below, directly show that any property “characteristic” of a psychotype, function, or pole of a socionic trait is only probabilistic in nature. There are no properties or manifestations that can be called necessary or sufficient for diagnosing any type. One can only say that, within a type, a particular property occurs among its representatives with a higher probability than in the population as a whole. Reliable type diagnosis can therefore be based only on the simultaneous consideration of many specific properties, and a reliable socionic diagnosis is possible only by relying on the statistical law of large numbers. Only formalized procedures (questionnaires or rigidly formalized and sufficiently long expert interviews) can reliably rely on the statistical law of large numbers; human thinking is incapable of this, however experienced or socionically “advanced” an expert may be. Socionic expert type diagnosis is especially unreliable in cases where the person being typed, in terms of psychological characteristics, is objectively located in a problematic zone near the boundary between sectors of psychological space (and there are many such cases). In our experience-based opinion, the reliability of expert type diagnosis cannot in principle exceed 60%, and only with special selection of the most strongly and contrastively expressed persons being typed can it approach 90%. Formalized procedures (for example, well-constructed large questionnaires) do not have these accuracy limitations, but for this purpose such questionnaires must necessarily take into account the full diversity of psychotype properties described in socionics (that is, ideally, diagnose psychotypes holistically rather than by several basic traits) and must not repeat other errors of questionnaires constructed according to the outdated recipes of Briggs-Myers inventories.

By the present time, socionics has already exhausted all the advantages that it could derive from the “non-formalized” development of its scientific culture. The characteristic properties of psychotypes and psychic functions have been described; further refinement of them within expert diagnosis is impossible. The time has come to fix and formalize the current achievements of socionics in the language of more exact sciences. Only by means of the language of mathematical statistics, which is precise and sensitive even to minor psychological nuances, will it be possible in the future to build interdisciplinary bridges from socionics to brain physiology. Only by formalizing in standard questionnaires the present level of socionic knowledge about the characteristic properties of types and functions (and this present level exceeds by an order of magnitude the level of knowledge available to the creators of the first Briggs-Myers type-theory questionnaire) can current socionic type diagnosis be standardized and its accuracy sharply increased. If applied type diagnosis can still tolerate (and sometimes is forced to tolerate) expert typing “by eye,” reliable and modern diagnostic questionnaires are simply indispensable for scientific research. Without them, socionics researchers have already spent a decade in a scientific dead end, engaging at worst in openly unscientific myth-making and at best in chewing over old theoretical cud. The present series of our scientific works, based on mathematical-statistical methods for analyzing a very large body of experimental material, is devoted to solving these corresponding current problems of socionics.

Our study is the first and unique study of this kind with respect to the structuring of psychological space accepted within the socionic paradigm (or, more broadly, neo-Jungian personality typology, which divides the diversity of psychological space into 16 independent and comparatively symmetrically arranged sectors called psychotypes). The properties of functions and traits are derived from the properties of psychotypes, because the properties of particular functions in different positions and of the poles of socionic traits are ultimately calculated by averaging the properties of particular groups of psychological types (for example, to identify the properties of program extraverted sensing, the properties of the SLE and SEE psychotypes are averaged, and so on). Therefore, the key potential question for studies of this kind is: how are the psychotypes whose properties are subsequently analyzed diagnosed?

In our studies, psychotypes were determined using questionnaires by correlating respondents’ response vectors with the “reference” (averaged) responses to the same questions given by “average representatives” of all 16 sociotypes. Thus, to determine the type of a particular person, 16 empirically obtained correlations were compared with one another. The maximum correlation indicated the diagnosis of the leading psychotype (sociotype). All 16 correlations together constituted the respondent’s complete individual psychological profile, characterizing their individual distinctiveness as fully as possible. Thus, if the maximum correlation assigned a person to the LII type, the second-highest correlation could assign that person to one of several other types (EII, LSI, LIE, ILE, even SLI or IEI). Accordingly, the height of the other profile peaks, apart from the highest peak, indicates the individual character of an accentuation additional to the basic type. The reliability of determining respondents’ leading type by this procedure in our study, as calculations showed, averaged no less than 90% correct diagnoses (the percentage was verified by convergence of diagnoses in a retest using a questionnaire with completely different diagnostic questions). At the first stages of developing the system of diagnostic coefficients (identical to ideal reference responses), the “reference responses” of the types (the basis of diagnosis) were understood as the averaged responses of respondents in the training sample who had stated their type with sufficient competence on the basis of self-typing or repeated external typing (predominantly by experts from professional socionic schools in Moscow and Kyiv). The competence of a stated diagnosis was assessed by the number of discrepancies in the basic dichotomies between the stated type and the initial preliminary questionnaire-based type diagnosis. Stated diagnoses with three or more dichotomy mismatches were rejected from the training sample as clearly false or erroneous. Other tools were also used to control the quality of incoming questionnaires and stated diagnoses; these were unrelated to their socionic content and therefore could not, through partial rejection of material, negatively affect the results of the analysis. Ultimately, within each type in the training sample, all honestly given responses of respondents who had previously been typed by different experts with partially divergent views on the boundaries between types were averaged. This achieved a reflection of the view of type boundaries averaged across the general “socionic culture” of Russia and Ukraine. Additional correction of the types’ reference responses, increasing the accuracy and reliability of the diagnostic coefficients, was introduced using three special mathematical procedures. The first of these coefficient-refining procedures was stepwise recurrent refinement of respondents’ diagnoses. The second refining procedure was the so-called symmetrization of the training-sample data array (the procedure is based on the fact that the reference response of any type to a given question can be predicted with fairly high accuracy, using decomposition by Reinin traits, from the reference responses given to that question by the other 15 types). The third and most important refining procedure was symmetrization of the diagnostic coefficients in an array of two thousand aggregated psychological properties that described, sufficiently evenly, the entire diversity of possible manifestations of a hypothetical complete psychological space. The latter procedure rests on the assumption that, in psychological space, the difference (that is, the mathematical distance in property space) between the LIE and LII psychotypes should be the same as between EIE and EII, SLE and SLI, and so on, while the difference between ILI and SLI should be approximately the same as between LSI and LII, and so on and so forth. If the stated equality of distances is not achieved, the inequality is partially corrected by additionally rotating the required axes in multidimensional space, which produces a small correction of the diagnostic coefficients originally obtained empirically for each questionnaire item and each of the 16 psychological types. At the first stages of obtaining and refining the diagnostic coefficients of questionnaire items, two additional refining procedures were also used. One controlled for the presence of negative kurtosis at the boundary between poles in probability-density graphs describing the distribution of the magnitude of the four basic socionic traits; the other controlled the maximization of the substantive content generated by the three quadra traits while varying the rotation of the four basic axes under the condition that their mutual orthogonality be preserved. At the final stages of refining the diagnostic coefficients, the stated psychotypes were no longer used, because all the information they could provide had already been used. Diagnoses in this case were made exclusively on the basis of the questionnaire itself, using the diagnostic coefficients from the previous approximation, while further refinements were introduced only by the procedure of symmetrizing the array of diagnostic coefficients (also taking into account the requirement of the maximum possible orthogonalization of all socionic axes). The same truncated procedure was also performed when the database was expanded with new, additional questionnaire items.

Thus, initially (at the very first stage of developing the system of diagnostic coefficients for questionnaire items), a “training sample” was formed for each TIM from people who had self-typed, provided that they had sufficient socionic experience, or who had been typed by DIFFERENT experts from Moscow, St. Petersburg, and Ukraine and had stated that TIM. It is clear that no more than 45-55% of the stated diagnoses in it were correct, but this is sufficient to obtain a sufficiently reliable mean value of quantitatively graded responses for each questionnaire item (when the mean is calculated, errors in opposite directions cancel one another in most cases - the statistical law of large numbers). At the next stage, statistical procedures were carried out to control negative kurtosis, maximize the content of the quadra traits under the condition of their maximum orthogonality to the basic traits, and, finally, symmetrize the results (TIMs in multidimensional psychological space should be located at equal distances from one another, forming a symmetrical structure). As a rule, these procedures introduced only minor corrections, but they helped eliminate systematic errors (caused, for example, by the fact that some socionic schools at one time typed too many subjects as “Hamlets,” and also by the fact that in self-typing some TIMs turn out to be preferred over others). At the final stages, the diagnostic coefficients obtained using the training samples were already used for direct questionnaire-based diagnosis, ignoring the previously stated TIMs. Diagnostic coefficients were recalculated on the expanded sample, and checks were made of the symmetry of TIM placement in psychological space, the correspondence of the most characteristic properties identified for the TIMs to their “canonical” descriptions by the majority of authors, and the properties obtained for the 8 functions and the poles of the basic traits - again, whether the resulting findings corresponded to the majority of descriptions. All checks were passed. In individual cases where apparent roughness was detected, additional source material was collected for refinement. As a result, after many successive approximations, using more than 5,000 different subjects in total, an almost ideal system of diagnostic coefficients was obtained for more than 5,500 questionnaire statements, making it possible subsequently to assemble any shorter diagnostic questionnaires from them with very high reliability of type determination. The validity of the developed diagnostic procedures was confirmed by a very high correlation between the diagnostic results of pairs of questionnaires containing completely different questions and repeatedly administered to the same subjects (above 0.93). Content validity is also confirmed by the maximal correspondence of the experimentally obtained “psychological portraits” of psychotypes and functions to descriptions of their characteristic properties by the majority of authors.

Obtaining valid diagnostic coefficients for more than 5,500 original questionnaire items characterizing more than 2,000 different psychic properties makes it possible to assemble from them any number of different diagnostic questionnaires, each of which is immediately ready for use and requires no additional adaptation. When selecting questions from the database for any new questionnaire, it is necessary only to satisfy the condition that diagnostic symmetry of the coefficients be preserved (roughly speaking, the sums of squares of the diagnostic coefficients for all 16 psychotypes across the selected questions must be equal, and the systems of diagnostic coefficients obtained from them for diagnosing the 15 traits must be mutually orthogonal, which is equivalent to their scalar products being zero). Selecting questions for a questionnaire under these conditions is readily accomplished using simple mathematical control procedures.

Thus, the socionic questionnaire-based type-diagnostic method we developed, subsequently used in a large series of studies to obtain detailed “psychological portraits” of psychotypes and psychic functions, is distinguished by high accuracy and reliability of type diagnosis (above 90%); it is more accurate than any expert typing and is free of experts’ subjective systematic errors. The method also differs fundamentally for the better from previously known socionic diagnostic questionnaires. It is characterized by:

1. complete absence of authorial subjectivity in selecting questions and assigning particular diagnostic values to them;
2. use of questions with maximum discriminative ability, mathematically confirmed;
3. absence of distortions - for example, whereas some other questionnaires previously diagnosed irrationality using manifestations of irrationality characteristic mainly only of intuitives;
4. no reliance on traits at all - TIMs are diagnosed holistically.

The method’s diagnosis of psychotypes on the basis of their holistic characteristics, without targeted use of socionic traits during diagnosis, subsequently makes it possible to validly study manifestations of any socionic traits and test whether they have substantive content (regardless of whether the issue is extraversion-introversion and rationality-irrationality or, for example, questimity and positivism, because before this diagnostic method all traits are equal and none has an advantage initially built into it by the questionnaire’s construction). The same applies to the purity, guaranteed by the method, of constructing “psychological portraits” of psychic functions in their various positions (program, creative, and so on).

Complete Psychological Portrait of the SLI - Experimental Results

The more than five thousand questionnaire statements analyzed as a result of the long-term experiment, which respondents evaluated by “trying them on” themselves, were manually grouped by us - on the basis of their semantic content and mutual correlations - into approximately 2,300 aggregated clusters containing from one to several dozen questionnaire items similar in meaning and, as a rule, characterizing more generalized psychological properties. On the one hand, replacing the primary questionnaire statements with their generalized clusters simplifies presentation of the results by reducing the number of primary properties that must be surveyed; on the other hand, averaging across several items further increases the accuracy of the projections of psychological properties onto psychotypes. It should, however, be borne in mind that, in all the tables below, the reported factor by which the probability of detecting a property in a representative of a psychotype exceeds the probability of detecting it in the entire population or in groups of other psychotypes was calculated and is reported by us not for the averaged properties (clusters), but still for individual single members of a cluster, that is, for specific questionnaire statements. In reality, this indicator refers not to the probability that a representative of the type actually possesses a given property, but to the probability that the respondent agrees that they possess that property. When the statistical characteristics of several questionnaire statements within one cluster are averaged, the stated probability should become more contrasted as random errors in subjects’ self-assessments are leveled out. Where the probability is greater than 0.5, for a generalized cluster of several statements it should become still closer to one; where it is initially less than 0.5, it should decrease still further, approaching zero. Accordingly, the factors by which probabilities are exceeded should increase substantially for clusters consisting of several statements (increasing two- to threefold or more). This increase is ignored in the tables. The reason is that the tables are intended for practical use. When in practice we assess whether a person has some property and for this purpose rely either on the person’s self-assessment or on an outside observer’s view, we are in fact conducting an experiment on only one specific manifestation of that property, without averaging the results of several experiments checking for the presence of the property. Thus, in practical use, the reported factors by which probability is exceeded, calculated for individual specific questionnaire statements, will provide a sufficiently accurate minimum approximate estimate even when a cluster property, previously derived by averaging several questionnaire items, is tested on the basis of a single experiment.

The tables may be very useful both for obtaining a clearer understanding of the characteristic properties of the psychotype and for purposes of expert socionic type diagnosis.

All tables in the article are presented in a “newspaper version,” that is, with some omissions and abridgments. For example, certain properties whose analysis is continuing have been excluded; in addition, columns of the tables reflecting the complete system of diagnostic coefficients (reference responses) suitable for use in diagnostic questionnaires have temporarily been excluded. The indicators retained in the tables provide a sufficiently clear picture of the most characteristic properties of the psychotypes.

Table 1. Psychological and psychophysiological properties (substantive clusters combining several questions similar in meaning) exclusively characteristic (that is, most strongly and primarily characteristic) specifically of the SLI psychotype. The conditional exclusivity index is defined as the amount by which the SLI type mean exceeds the population level, plus the distance from the SLI type mean to the type mean of the psychotype ranking second in expression of the property.

Table 2. Psychological and psychophysiological properties (substantive clusters combining several questions similar in meaning) highly characteristic of the SLI psychotype, but only in second or third place after other psychotypes.

Table 3. Properties that most strongly distinguish SLI from LSE.

Table 4. Properties that most strongly distinguish SLI from SLE.

Table 5. Properties that most strongly distinguish SLI from LSI.

Table 6. Properties that most strongly distinguish SLI from SEI.

Table 7. Properties that most strongly distinguish SLI from ILI.

Table 8. Properties that most strongly distinguish SLI from ESI.