2 The Impact of Intelligence and Dispositional Valuations on Mutual Understanding


In the previous chapter, the linguistic and social mechanisms behind mutual understanding were discussed. In the current chapter, this background information is used to highlight the role of intelligence and dispositional valuations in the MU process. In Section 2.1, the intelligence and dispositional valuations, which are the independent variables of the current research, are described and relevant empirical evidence is reviewed. Section 2.2 covers main effects of intelligence and dispositional valuations, which are independent on the personality of the interaction partner. Section 2.3 reviews research regarding assortative patterns in relationship formation (i.e., the tendency for people to establish relationships with similar individuals). The existence of such patterns is interpreted as indirect evidence for dyadic effects of intelligence. Section 2.4 discusses dyadic effects, which are dependent on the interaction between the personalities of both interaction partners. Finally, the combined effects of main and dyadic influences and their implications for intellectually gifted individuals are covered in Section 2.5.

2.1  Review of Independent Variables

In studying the effect of intelligence and dispositional valuations on MU, the current dissertation focuses on three broad construct categories. Two out of three categories belong to the overarching category of intelligence. Although the exact definition of intelligence is still contested (Sternberg, 2000), most researchers would agree with David Wechsler’s (1958, p. 7) global description that it is involved in “the global capacity of individuals to act purposefully, to think rationally, and to deal effectively with their environment.” As stated above, intelligence is further divided into two subcategories. Following many intelligence researchers (e.g., Ackerman, 1996; Baltes, 1997; Cattell, 1963), a distinction is made between fluid intelligence and crystallized intelligence. Fluid intelligenceas a context-independent resource needed to adapt to the environment, whereas crystallized intelligence represents the outcome of investment of fluid intelligence in specific domains, such as knowledge or expertise. The third broad category studied here is labeled dispositional valuations, which are defined as a set of constructs that involve the differential value attached to actions and end goals. In the current framework, traits in this category include openness to experience, interests, and values. In the following, relevant research pertaining to intelligence and dispositional valuations is reviewed.

2.1.1  Fluid Intelligence

Belsky (1990, p. 125) defined fluid intellectual ability as „on-the-spot reasoning ability, a skill not basically dependant on our experience.” Similarly, Cattell (1971, p. 99) defined it as “an expression of the level of complexity of relationships that an individual can perceive and act upon when he does not have recourse to answers to such complex issues already stored in memory”. Both definitions stress the fact that fluid reasoning is involved in problems that are new to the person in question. Tests of fluid intelligence usually involve nonverbal materials that are not often encountered in everyday life. For example, the widely-used Raven’s (1960) Progressive Matrices requires subjects to learn rule-based regularities in the design of unfinished abstract figures and to apply these rules to complement the picture (see Figure 4).


Figure 4. Sample Item of a Matrices: Test of Fluid Intelligence

Note. Item 175 of the IST-2001 (Amthauer, Brocke, Liepmann, & Beauducel, 2001)

Fluid intelligence can be compared to Hebb’s (1949) Intelligence A, which is hypothesized to be rooted in physiological processes. Consistent with this equivalence, various studies have demonstrated that age-dependent physiological declines are associated with a sharp drop in fluid intelligence, beginning in young adulthood (e.g., Baltes, Staudinger, & Lindenberger, 1999).

The relation between fluid intelligence and other intelligence factors is still unclear. For example, Gustafsson (1988) argues that fluid intelligence is identical to general intelligence. Carroll’s (1993) meta-analysis on the structure of intelligence, the most extensive contribution to this issue thus far, identified eight main intelligence factors that are related to a higher-order general factor. His results show that fluid intelligence is the factor closest to general intelligence but he did not argue that the constructs are identical (for similar findings, see Bickley, Keith, & Wolfle, 1995).


Unsatisfied with the supposedly „detached” nature of psychometric, fluid intelligence,a number of researchers have introduced the constructs of „emotional intelligence” and/or „social intelligence” to account for individual differences in relational skills. However, performance tests of social and emotional intelligence have often failed to demonstrate substantial independence from psychometric intelligence tests (for emotional intelligence, see Brody, 2004; Davies, Stankov, & Roberts, 1998; Roberts, Zeidner, & Matthews, 2001; for social intelligence, see Keating, 1978; Shanley, Walker, & Foley, 1971). Because of this, social and emotional skills are discussed here as particular instances of fluid intelligence applied in social contexts.9

The current study includes a sample of gifted individuals because, for reasons elaborated below, they are expected to be especially affected by intelligence differences between persons. Intellectual giftedness is defined here as an IQ of at least 130, so that the most intelligent 2% of the population can be described as intellectually gifted. Of course, the cutoff of 130 is somewhat arbitrary, yet it most closely reflects the consensus in the scientific community (Rost, 2000). Some alternative models conceptualize giftedness as multidimensional. For example, Renzulli (1986) defined giftedness as the simultaneous presence of not just an above-average IQ but also a high level of creativity and task commitment (for other multidimensional models of giftedness see Gagné, 1991; Heller, 2001). These models have been criticized for a lack of conceptual and empirical rigor (e.g., Rost, 2000). Since the present study is theoretically interested in the entire distributional range of intelligence, it is not necessary to dive into this discussion, since the assumption made in the current dissertation is that the mechanisms that are responsible for the effect of intelligence on MU are also at work at more moderate intelligence levels.

2.1.2 Crystallized Intelligence


In contrast to fluid intelligence, which is concerned with individual differences in reasoning ability in the face of novel stimuli, crystallized intelligence pertains to the products of the investment of fluid ability in specific environmental domains. These products are acquired over the life course instead of passively influenced by biological or environmental forces. Probably as a result, individual differences in crystallized intelligence are more resistant to age-related declines in neurophysiology (Baltes et al., 1999).

An important facet of crystallized intelligence is vocabulary, which refers to agreed-upon conventions regarding the meaning of words. Whereas some people only know a selected number of words, other people use highly differentiated terms and have detailed knowledge of synonyms, antonyms, and proverbs. Research has found that individual differences in vocabulary are an important source of differences in crystallized intelligence, yet tests of vocabulary also load high on the general factor of intelligence (Carroll, 1993; Ullstadius, Gustafsson, & Carlstedt, 2002). Consistent with the notion that crystallized intelligence may still increase while more fluid resources are already on the decline, a recent meta-analysis by Verhaeghen (2003) has demonstrated that vocabulary scores of older adults are higher than those of younger adults, with an average effect size of d = 0.80.

2.1.3 Dispositional Valuations

As stated previously, dispositional valuationsinvolve the differential value attached to actions and end goals. In the current dissertation, the most important constructs in this category in terms of their impact on the MU process are openness to experience, interests, and values. As a dispositional valuation, openness to experience is conceptualized as influencing the value that people attach to structural features of cognition. Interests concern the differential valuation of certain activities. Finally, values involve the valuation of broad end goals. In the following subsections, a description as well as a brief summary of relevant evidence regarding these variables is provided.  Openness to Experience


Openness to experience is the fifth factor of the „Big Five” model of personality description (John & Srivastava, 1999). It is usually treated as an intrapsychic dimension, describing individual differences in the structure and functioning of the mind. For example, McCrae and Costa (1997) link openness to differences in the „breadth, depth and permeability of consciousness and in the recurrent need to enlarge and examine experience” (p. 826).10

Open individuals seek out and reflect upon new experiences. This feature has some resemblance to Need for Cognitive Closure (NCC), which has been defined as the preference for a definite answer on some topic and an avoidance of confusion and ambiguity (Kruglanski, 1990). In fact, it is quite difficult to think of a highly open person that is intolerant of complexity and ambiguity, and vice versa. Consistent with this argument, a study by Webster and Kruglanski (1994) found significantly negative correlations between NCC and dogmatism and authoritarianism, which are aspects of low openness (rs ≈ .30). Interests and Values

Asendorpf (2003) conceptualized interests as tendencies to attribute pleasure and curiosity to some activities but not to others. Interests have been most extensively studied within the context of vocational aspirations. Numerous interest taxonomies have been proposed, such as Holland’s (1959) hexagonal model, which proposes six basic interests. Realistic interests involve concrete objects and things; investigative interests involve intellectual pursuits; artistic interests are concerned with art and creativity; social interests concentrate on working with people; enterprising interests involve projects and commercial enterprise; and conventional interests are focused on clerical and computational tasks.


Rokeach (1973) defined a value as „an enduring belief that a specific mode of conduct or end-state of existence is personally or socially preferable to an opposite or converse mode of conduct or end-state of existence” (p. 5). In the current framework, values can be conceptualized as the value attached to behavior (mode of conduct) and goals (end-state of existence). Because they can be important motivators for behaviors and have great emotional significance, values potentially affect MU.

2.2 Main Effects of Intelligence and Dispositional Valuations

Now that the most important categories of intelligence and dispositional valuations have been reviewed, the following section focuses on the main effects of these variables on the MU process. Such effects are at work when an individual’s personality has a direct influence on the MU process, regardless of the personality of the interaction partner.

2.2.1  Fluid Intelligence

According to Chandler (1994), signs have both denotative and connotative meaning. Whereas denotation refers the literal and commonsense meaning of a sign (i.e., the meaning of a word in a dictionary), connotation refers to its socio-cultural and idiosyncratic associations. The ability to decode denotative meaning is more closely related to vocabulary (i.e., crystallized intelligence). In contrast, decoding personal connotative meaning requires the use of contextual cues. Because fluid intelligence is involved in the ability to integrate new information and make inferences, more intelligent people should be more adept in decoding the connotative meaning of an utterance.


As stated previously, fluid intelligence is closely related to social intelligence. Because of the relatively undeveloped state of research regarding emotional intelligence (Zeidner, Roberts, & Matthews, 2004), the following discussion mainly focuses on social intelligence. Social intelligence was defined by O’Sulivan, Guilford, and deMille (1965) as „ability to judge people” (p. 5) with respect to „feelings, motives, thoughts, intentions, attitudes, or other psychological dispositions which might affect an individual’s social behavior” (p. 4). O’Sullivan et al. (1965) constructed tests for the measurement of social intelligence, starting from the assumption that „expressive behavior, more particularly facial expressions, vocal inflections, postures, and gestures, are the cues from which intentional states are inferred” (p. 6).

Expressive cues can be used to decode the connotation of verbal messages. Imagine a man and a woman who are driving in their car towards some destination. The woman is behind the steering wheel and has stopped at a red traffic light. After a while, the man comments that „The light has turned green.” When the man is very cynical in his tone, the sentence might be interpreted to mean something as „You are so stupid that I have to tell you what to do all of the time”. Equally possible, the sentence might communicate a more collaborative attitude of wanting to help the partner in attending to the traffic. In this example, a person with high (social/emotional) intelligence would be able to correctly identify the underlying message.

Because of the generative nature of language, words have to be combined in novel, creative ways to communicate meaning. Intelligence is associated with a higher level of word fluency, so intelligent individuals can also be expected to be more skilled at encoding their own thoughts and feelings into linguistic utterances. Indeed, a study by Quay, Hough, Mathews, and Jarrett (1981) demonstrated that general cognitive ability is positively associated with communicative encoding. Furthermore, a meta-analysis by Davis and Kraus (1997) reported that cognitive ability is associated with superior empathic skills. The hypothesized superior ability of intelligent individuals to encode and decode verbal utterances gives rise to the first main effect hypothesis:


Main Effect Hypothesis 1: Fluid intelligence is positively related to MU. [Hm-1]

2.2.2 Crystallized Intelligence

The relation between vocabulary and the ability to encode and decode verbal material is close and obvious. To understand an utterance, it is necessary to understand the meaning of its constituting words. Consistent with this, a meta-analysis of 52 studies by Stahl and Fairbanks (1986) showed that teaching children the meaning of words improves their general reading comprehension, even of texts that do not contain the words that were taught (d = .30). Although the relative contribution of fluid intelligence and vocabulary is still debated, some empirical evidence exists that vocabulary has a direct effect on language comprehension (Cain, Oakhill, & Lemmon, 2004).


When a person wants to express an idea or thought, the first step in this process is the selection of the word(s) that are most appropriate captures its meaning (Cleland & Pickering, 2003). Richer vocabularies are associated with a larger pool of words and concepts to express ideas (Lohman, 2000). Consistent with this, research on cognitive differentiation has found that the number of constructs a person uses to describe events is positively associated with communication effectiveness (for a review see O’Keefe & Sypher, 1981). In addition, a study by Applegate, Kline, and Delia (1991) found that cognitively differentiated individuals communicate in a more person centered manner, which has been shown to promote interpersonal comfort (Jones & Guerrero, 2001). As a result, these factors should be related to increased levels of MU. This gives rise to the following hypothesis:

Main Effect Hypothesis 2: Crystallized intelligence is positively related to MU. [Hm-2]

2.2.3 Dispositional Valuations

Because interests and values are not expected to exert a main effect on MU, the following section on main effects of dispositional valuations focuses only on openness to experience. Openness to Experience


When people adjust communicative message to the background knowledge of their conversation partners, they have to be sensitive to the other person’s perspective instead of assuming shared knowledge. Because people who are low in openness are thought to dislike unstructured, ambiguous situations, they may be less effective during the process of audience design than people high in openness (for a similar argument, see Gagne & Lydon, 2004, p. 333).

There is some evidence that this is indeed the case. Richter and Kruglanski (1999) let 99 college students provide descriptions of various abstract figures. Prior to the task, participants were told that either they themselves or a previously unknown participant were to use these descriptions in a referential communication task (see Section, for a description) scheduled several weeks later. Results showed that descriptions written by participants high in need for closure (i.e., low in openness) were shorter and contained more idiosyncratic references. Consistent with the idea that such messages contain a lower amount of socially shared information needed for effective communication, they were less likely to be successfully identified by others.

The following considerations give rise to the following hypothesis:


Main Effect Hypothesis 3: Openness to experience is positively related to MU. [Hm-3]

2.3 Between-Person Differences in Intelligence and Dispositional Valuations

A central organizing hypothesis of the current dissertation is that between-person differences in intelligence and dispositional valuations are negatively related to MU. When between-person differences are detrimental in social relationships, then people should be motivated to seek out peers with similar personality structures (because of the fundamental human drive for intimate, close relationships; Baumeister & Leary, 1995; McAdams, 1989; Reis, 1990). Accordingly, assortative patterns are an indirect argument for the detrimental influence of between-person differences in intelligence and dispositional valuations (the hypothesized mechanisms behind this association are discussed in Section 2.4). In the following section, evidence for niche picking with regards to general intelligence, crystallized intelligence, and dispositional valuations is reviewed. Subsequently, some interpretations for these findings are discussed.

2.3.1  General Intelligence

A number of empirical studies have focused on the similarity between marriage partners (assortative mating) in general intelligence. Table 2 summarizes the results of some representative studies. As can be seen, an average spousal correlation of .34 (after Fisher r-to-z transformation and back-transformation) is obtained across studies. This figure is almost identical to the weighted spousal correlation of .33 as reported by Bouchard and McGue (1981) in a review of 16 studies. Thus, it can be concluded that there is a moderate degree of spousal similarity in terms of general intelligence levels.


There is little research on assortative similarity in friends. Only some indirect evidence points in this direction. For example, it has often been documented that gifted children prefer older friends (Janos & Robinson, 1985). It could be speculated that this represents an effort to affiliate with persons of a comparable mental age. However, this interpretation needs to be backed up by more future research.

2.3.2 Crystallized Intelligence

Compared to the findings on assortative mating for general intelligence, less is known about spousal similarities in crystallized intelligence. Some of the studies reviewed in Table 2 report separate spousal correlations for verbal IQ (which is related to crystallized intelligence and vocabulary) or knowledge. For example, Tambs et al. (1993) found a correlation of .34 for the Information subtest of the WAIS, which measures general knowledge. Nagoshi et al. (1987) found a similarity correlation of .10 for verbal IQ, whereas Willoughby (1927, 1928; cited in Vandenberg, 1971) found a correlation of .44. When these coefficients are averaged, a similarity coefficient of .30 is found, which is very similar to the findings for general intelligence.

Because educational systems are designed to transfer knowledge, a person’s educational level may be taken as a proxy for the sophistication of his or her knowledge structures. As can be seen in Table 2, the correlation between spouses’ educational status is higher than found for fluid or crystallized intelligence. For example, Phillips et al. (1987) found an average spousal correlation of .41, Nagoshi et al. (1987) of .47, and Reynolds et al. (2000) of .54 (latent path correlation). Note, however, that these high correlations may partly be a result of similarities in fluid intelligence or a function of social stratification (see discussion below).


Table 2 Spousal Correlations for General Intelligence, Crystallized Intelligence, and Education






Phillips et al. (1987)




Tambs et al. (1993)




Nagoshi et al. (1987)





Mascie-Taylor & Vandenberg (1988)



Willoughby (1927, 1928) ab




Jones (1928) b



Burks (1928) b



Reynolds et al. (2000)








Note. G = General intelligence, Gc = Crystallized intelligence, Edu = Education.
a nonverbal intelligence
b cited in Vandenberg (1971)

2.3.3 Dispositional Valuations

There exist indications for assortative mating in terms of dispositional valuations, especially for openness to experience. McCrae (1996, his Table 5) summarized spousal similarities for 14 openness-related traits and found an average cross-spouse (assortative) correlation of .41 (based on 19 coefficients; range .19-.74). Lykken and Tellegen (1993) reported that spousal personality correlations are mainly restricted to traits such as religiosity, conservatism, authoritarianism, and the endorsement of traditional values (rs .33 - .57; see also Feng & Baker, 1994), which are all facets of low openness.

Results from a number of studies suggest the degree of spousal similarity is higher for openness than for other personality traits. For example, McCrae (1996) reported spousal correlations for 103 couples and the strongest effect for openness (r = .33), whereas the only other significant effect was found for conscientiousness (r = .21). Botwin, Buss, and Shackelford (1997) found significant assortment effects for agreeableness and conscientiousness (rs = .22 to -.33), but more so for openness (rs = .38-.51). Finally, Waller (1999) studied personality similarity in 149 spouse pairs and found only one significant correlation for conventionality (r = .41), which is related to (low) openness.


Between-person differences in dispositional valuations might affect the formation of friendships. Consistent with this notion, Cheng, Bond, and Chan (1995) asked 434 Chinese adolescents (aged 17-20) about their own and an „ideal friend’s” personality and found a high similarity correlation (r = .56) for openness to experience. Johnson (1989) sampled 56 pairs of close friends and 50 pairs of acquaintances from a residential, white, U.S. mid-western city (average age 37) and asked them what attracted them to the other person. An average of 76% of the close friends and 66% of the acquaintances named the similarity of values and interests, which made it the single most important self-rated predictor. These findings suggest that people prefer friends that are similar in their dispositional valuations.

2.3.4 Interpretation of Assortative Patterns Regarding Intelligence  Empirical Evidence

It is often assumed that assortative patterns regarding intelligence are the results of active phenotypic assortment. That is, people are hypothesized to actively seek out cognitive peers. In the current study, it is hypothesized that they do this because cognitive peers have less difficulty in communicating personally relevant thoughts and feelings. As a result, feelings of MU in particular and relationship satisfaction in general should be maximal when between-person intelligence differences are small.


To the best of the present author’s knowledge, there exists only one empirical study that directly analyzed the relation between relationship quality and between-person intelligence differences. Lewak, Wakefield, and Briggs (1985) studied a sample of couples from the general population (N = 81), half of which underwent marital therapy. Correlations between marital satisfaction and spousal IQ differences (assessed with the WAIS-R) were non-significant for simple difference scores (rs ranging between -.04 and .10) and small but negative for squared difference scores. Only in the case of verbal intelligence, a somewhat larger correlation of -.21 was found, but this association did not reach statistical significance.

Lewak et al. (1985) concluded from their findings that marital satisfaction is independent from similarity in intelligence. However, the validity of these conclusions is threatened by methodological flaws. Most importantly, the community and marital therapy sample differed markedly in the level and range of marital satisfaction (d = .91, p < .01), but not in the level of intelligence. Because it is not known whether the two samples also differed in their level of assortative mating on intelligence, marital satisfaction differences between samples may have canceled out differences between partners within the same sample. Thus, it is premature to take the Lewak et al. (1985) study as evidence against dyadic effects of fluid intelligence. Alternative Explanations

As stated above, findings on spousal and friendship similarity suggest that people actively seek out cognitive peers as their friends or romantic partners. This might be the case because intelligence differences impair MU and thus conflict with the fundamental human need to establish close and intimate relationships. However, as is the case for all conclusions based on correlational data, the causal factors behind this assortment are very difficult to study empirically.


One alternative explanation for assortative correlations such as the ones reviewed above is that modern societies are stratified according to intelligence levels. As a result, individuals will have more contact with peers of comparable intelligence (social homogamy). Therefore, assortative correlations regarding intelligence or any other education-related trait may be caused by passive availability instead of active selection (for empirical evidence, see Nagoshi et al., 1987; Reynolds et al., 2000; Tambs et al., 1993). Illusory Similarity

When personality similarity is assessed by taking only the perspective of a single individual into account, it may be influenced by a tendency to „project” one’s own personality traits into the other person. In an exemplary study, Murray et al. (2002) asked 105 married/cohabiting couples and 86 dating couples about their own and their partner’s personality. Illusory similarity was operationalized as the degree to which individuals rated their partners as more similar than they really were. Results showed that illusory similarity was positively related to relationship satisfaction in the married sample. In addition, illusory similarity in women predicted a higher stability of relationships in the dating sample. These results suggest that people in satisfying and stable relationships perceive similarities in their partners that are not evident in reality.11

Murray et al.’s (2002) findings have important implications for the current dissertation, as Study 1, 2, and 3 use the same informants to rate a) their own personality, b) the personality of their relationship partners, and c) the quality of the relationships with these partners. As a result, associations between perceived similarity and relationship quality are potentially confounded by an ego-centric tendency to project one’s own personality on partners in satisfying relationships. This should be taken into account when interpreting the corresponding results.

2.4 Dyadic Effects of Intelligence and Dispositional Valuations


In the previous section, the literature regarding between-person differences in intelligence and dispositional valuations was reviewed. It was concluded that there is a tendency for people to affiliate with cognitive peers. One of the main hypotheses of the current dissertation is that between-person differences in intelligence and dispositional valuations are detrimental to establishing a sense of MU in social relationships. However, the data on which this conclusion is based are indirect, and there are several possible alternative interpretations. In the current section, more direct evidence is presented.

2.4.1  Fluid Intelligence

In the following, dyadic effects of fluid intelligence are discussed. This discussion is divided in two subsections. First, intelligence plays a role in shaping differences in the kinds experiences people make in life, which may be related to MU. Second, differences in the generation of insight and ideas related to intelligence and their impact on the associative phase of the MU process are discussed.  Shared Experiences and Evaluations

Because of its profound social consequences, intelligence can be an important determinant of the kind of experiences people make (Gordon, 1997). One of the principal reasons for this lies in the educational and vocational segregation according to intelligence. For example, university students are exposed to a very specific intellectual climate (consisting of reading books, discussing theories, etc.) that differs from that of their peers who directly enter the labor market after high school (Arnett, 2000). Accordingly, people with different intelligence levels are more likely to experience nonshared associations during a conversation, which may hamper MU. Insight and Ideas


As reviewed in Section 2.1.1., fluid intelligence manifests itself in reasoning about novel phenomena. Moreover, more intelligent individuals have been found to be more creative in solving divergent problems (Harris, 2004). Therefore, they may be more likely to produce insightful and creative associations during a conversation. For example, while discussing the long waiting lines in front of a museum, the idea might come up to send text messages to potential visitors when the queue is shorter than usual (as was done during the Museum of Modern Art exhibition in Berlin, 2004).

The generation of new insightful and creative associations poses a challenge to the MU process, since such associations are not likely to be shared between interaction partners. First, the creative person is faced with the task of establishing a link between the interaction context and the creative or insightful idea. For example, the person in question might remark: „Looking at the long queue, I was wondering whether it would be possible to reduce the waiting time with some technical device.”

Second, when the context of the new idea has been made transparent, the interaction partner needs to be convinced of its validity (e.g., would sending SMS text messages work?). In case of marked individual differences in intelligence levels, less intelligent people will not necessarily be able to understand the logic behind the novel ideas of very intelligent people. Especially when the person in question is very excited about the idea and wants to share it with other persons, MU will be hampered. This leads to the following prediction:


Dyadic Effect Hypothesis 1: Between-person differences in fluid intelligence are negatively related to MU. [Hd-1]

2.4.2 Crystallized Intelligence

Interpersonal differences in vocabulary size should be related to dyadic effects in the MU process because persons with a very large vocabulary size are more likely to use words that are not shared by their interaction partners. When this is the case, subsequent decoding of the utterance by their interaction partners will be hampered, leading to interruptions in the MU process. Such „complicated” use of language is a potential hindrance to the sharing of meaning because it makes the discovery of shared experiences (Duck, 1994) more difficult. For this reason, between-person differences in vocabulary might even be consciously maximized to create interpersonal distance (e.g., when a person uses difficult „upper class” words).


The above considerations give rise to the following hypothesis:

Dyadic Effect Hypothesis 2: Between-person differences in crystallized intelligence are negatively related to MU. [Hd-2]

2.4.3 Dispositional Valuations

According to Duck’s (1994) theory, individuals with more shared experiences have a larger potential to develop a deep and meaningful relationship. A crucial prerequisite in his four-phase model is that both individuals agree in the evaluation of these experiences (see Figure 3). Also, recall that a crucial step in the formation of intimacy is that individuals feel valued by their interaction partners. Because dispositional valuations are directly involved in evaluations of other persons (e.g., Heaven & Oxman, 1999), they are hypothesized to be an important source of dyadic effects on the MU process. In the following, the effect of openness to experience, interests, and values is discussed. Openness to Experience


As described above, openness to experience is closely related to interpersonal differences in tolerance for ambiguity. According to McCrae (1996), between-person differences in the thinking style of open vs. closed individuals can be a serious interaction problem and lead to mutual avoidance. For example, a study by Kirton (1976) classified managers as either „innovators” (high openness) or „adaptors” (low openness). It was found that adaptors regarded innovators as neurotic and insensitive to others, whereas innovators saw adaptors as dogmatic, inflexible, and conservative (for related evidence, see de Dreu, Koole, & Oldersma, 1999). Consistent with this, a large-scale survey by Gurtman (1995) showed that individuals low on openness complain more about being too easily swayed by others12, which might be the underlying reason for their „defensive” and rigid thinking style. Indeed, according to McCrae (1996, p. 331):

Open people are bored by the predictable and intellectually undemanding amusements of closed people; closed people are bored by what they perceive to be the difficult and pretentious culture of the open.


This gives rise to the following hypothesis:

Dyadic Effect Hypothesis 3: Between-person differences in openness to experience are negatively related to MU. [Hd-3]

↓44 Interests and Values

Differing valuations of activities and goals are another potential source of nonshared evaluations of experience. For example, in countries with a military draft, all healthy men of a certain age are called to arms for a period of one to three years. Although they share a mutual experience, they might disagree on the meaning of this experience. For example, some persons enjoy military service as a time of male bonding and serving their country, whereas other individuals develop an aversion to the strict discipline and unquestioning patriotism (Goldstein, 1943).

Research in the so-called attraction paradigm (Byrne, 1971) has established that individuals are attracted to others with similar attitudes. Most studies carried out in this paradigm have relied on the so-called bogus stranger method. In this method, individuals are given a description of an imaginary person, who is displayed as either similar or dissimilar to the self. Results from a large number of studies using this paradigm are consistent with the notion that people are attracted to others with similar attitudes (Byrne, 1997).


The findings carried out within the attraction paradigm are easily reconciled with Duck’s (1994) notion that MU is dependent on the exchange of similarly evaluated experiences. There have been concerns, however, that the bogus stranger methodology may not be ecologically valid. Indeed, in the absence of any other information about a person, it makes sense for individuals to base their evaluations entirely on attitude similarity. Employing more naturalistic designs, however, Sunnafrank (1983; 1984; 1992) demonstrated that the influence of perceived similarity on attraction fades away when subjects are allow to interact with each other. Because these results are only based on a few experiments and some research has provided contradicting evidence (Cappella & Palmer, 1990), the more conservative prediction is that similarity in interests and values is positively related to MU:

Dyadic Effect Hypothesis 4: Between-person differences in interests and values are negatively related to MU. [Hd-4]

2.5 Combining Main and Dyadic Effects

In the following section, some implications of the combined influence of main and dyadic effects are discussed. As will be seen, most research in this domain has focused on the social effects of (fluid) intelligence. The following section focuses on two domains. First, Simonton’s (1985) model of the impact of intelligence on group influence is discussed. After this, research on the social adjustment of intellectually gifted individuals is reviewed, much of which assumes a dyadic effect of intelligence on social relationships (though the theory behind this assumption often remains implicit). This section finishes with an hypothesis regarding the combined impact of main and dyadic effects in cognitively gifted individuals.

2.5.1  Simonton’s (1985) Model of Intelligence and Group Influence Description of the Model


As stated above, intelligence is hypothesized to exert a positively main effect on MU, yet at the same time, large dyadic differences in intelligence may be detrimental to effective communication. Simonton (1985) used a similar logic to predict the optimal intelligence level for influence in groups. On the one hand, more intelligent subjects are assumed to generate higher quality contributions (consistent with Main Effect Hypothesis 1). This can be an important advantage because the contribution of each group member must „survive” the criticism of more intelligent members in order to become accepted. When only this Criticism factor would be at work, the most intelligent group member should eventually win over the group and become its leader.

This prediction might sound plausible at first, but it does not coincide with everyday experience, where it is often the case that political and economic leaders are notthe most intelligent members of the population. For example, it is contended that John F. Kennedy had an IQ of 119, which would place him „only” one SD above the mean.

To explain this observation, Simonton hypothesized that highly intelligent individuals’ ideas are too complicated to be communicated to most other people (Comprehension factor). Because less intelligent individuals are not able to comprehend more intelligent persons’ remarks, they are not likely to be influenced. Formalized in mathematical terms, Simonton assumed that an intelligence difference of more than one SD is enough to hamper interpersonal communication. When only the Comprehension factor is at work, people with an intelligence level of 108 should have the largest potential to influence others.


Figure 5. Model of the Relation Between Intelligence and Group Influence According to Simonton (1985).

Note. According to Simonton (1985), the effect of intelligence on group influence is dependent on the additive effect of two factors. First, more intelligent people face less criticism from other group members (depicted by the asymptotically decreasing curve). Second, however, interpersonal comprehension is highest at more moderate levels of intelligence (depicted by the bell-shaped curve). When the Comprehension and Criticism factors are combined (via subtraction), a function of the predicted association between intelligence and group influence is obtained that peaks around an optimal IQ level of 119.

As stated above, extremely intelligent people are more likely to produce effective solutions to problems. Because most members of normal social groups have more limited cognitive abilities, however, these ideas may be too complicated to be able to influence others. When the Comprehension and Criticism factors are combined (by means of subtraction), a curvilinear (inverted U) relation between IQ and interpersonal influence is obtained that peaks around an IQ of 11913 (see Figure 5). This new optimum lies in between the extremely high IQ level predicted by the Criticism factor and the moderate IQ level predicted by the Comprehension factor.

Although this prediction of a curvilinear, inverted-U relation between intelligence and group influence is relatively clear-cut, the model and the exact form of the intelligence x group influence function depend on a number of assumptions:

↓48 Empirical evidence

In spite of the plausibility of its theoretical predictions, Simonton’s model is not yet well-established and direct empirical evidence is mostly lacking. Only some indirect evidence is consistent with its predictions. Specifically, there are scattered findings that group leaders are indeed moderately above-average in intelligence (i.e., about 1.2 SDs above the population mean). For example, Gibb (1947) found that military officer candidates were 1.2 to 1.5 SDs more intelligent than the group they were to lead. In addition, Ghiselli (1963) tested the success rate of middle managers and found that optimal levels were achieved by those scoring between 1.2 and 1.5 SDs above the mean. Because direct support for the model is mostly lacking, the current study tries to test the Simonton model in the context of close interpersonal relationships.

2.5.2 Application of the Simonton (1985) Model to the Adjustment and Social Relationships of Gifted Individuals


The hypothesized negative dyadic effect of intelligence discrepancies on MU has clear implications for gifted individuals. Because the percentage of individuals at both extreme ends of the normal distribution is very small, the Simonton model predicts that gifted individuals, who are more than two SDs above the mean, can only communicate with about 16% of the population (i.e., applying the 1 SD criterion). By comparison, a perfectly average individual with an IQ of 100 could communicate with about 68% of the population. For extremely gifted individuals, the Simonton (1985) hypothesis predicts even greater problems. For example, the pool of communication partners of a person with an IQ of 145 is predicted to be restricted to only 2% of the population. As a result, these individuals may have more difficulty in finding friends or romantic partners with whom they can communicate at a satisfactory intellectual level. This should be associated with reduced feelings of being understood.

There exist few well-designed studies on the adjustment of gifted individuals. This is true despite an increased interest in the topic of giftedness in recent years (Rost, 2000, p. 7). Many studies use self-selected or clinical samples (e.g., participants of a summer camp for the gifted or psychiatric patients), which are not representative of the broader population. As a result, only findings from a small number of selected studies are reviewed here. As will be seen, empirical findings are mixed, with some studies showing superior adjustment, whereas others suggest adjustment problems, depending on the type of outcome and the level of giftedness that is considered. Studies Showing Superior Adjustment of Gifted Individuals

Terman (1925; Terman & Oden, 1959) compared teachers ratings of 532 gifted subjects (aged 7-14 years; mean IQ = 151) and 533 classmates (aged 10-14 years; mean IQ unknown) with regard to a number of social characteristics. This classic study has provided a wealth of information regarding the social adjustment of gifted children. In contrast to the stereotype of the sickly, socially awkward gifted child, his results demonstrated that 70% of these children were judged to outperform their peers in terms of leadership, whereas ratings of popularity ratings were slightly above-average (56%).


Terman’s results suggest an above-average adjustment of gifted children. Note, however, that his gifted sample was identified through teacher nominations. Because teachers (often implicitly) associate giftedness with well-adjusted behavior, it cannot be ruled out that the higher reported adjustment level of gifted children was partly caused by a so-called halo effect: the tendency of positive evaluations in one domain to affect more generalized impressions of a person. That is, it is possible that the Terman teachers were particularly fond of well-behaving and adjusted pupils and generalized these impressions to the cognitive abilities of these children (Rost, 2000). Accordingly, such children may have been more likely to be included in Terman’s gifted sample.

Janos and Robinson (1985) reviewed several studies on giftedness and adjustment and found that in the majority of studies, moderate levels of giftedness were related to better adjustment. Their conclusion corroborates Terman’s results that gifted individuals have better social reasoning and perspective taking skills. It is also consistent with a meta-analysis of 20 studies by Hoge and Renzulli (1993), who found that gifted children have higher levels of general self-esteem (average effect size d = .20; see Roznowski, Reith, & Hong, 2000, for similar findings). Studies Showing Some Adjustment Problems of Gifted Individuals

The studies reviewed above point to superior adjustment for gifted individuals. In the following, evidence for adjustment problems is discussed.


In Germany, Rost (2000) followed 107 gifted children and adolescents (mean age 15 years) drawn from a community sample (mean IQ = 136) and 118 highly achieving age mates (mean IQ = 102) and compared them to a control group of comparable size. Although both gifted and the highly achieving individuals had higher levels of academic self-esteem and their classroom behavior was rated more positively by their teachers, they had more negative self-perceptions of peer popularity. The highly achieving group also had a somewhat more negative self-concept of relationships with peers of the other sex, whereas the male subjects of this sample also perceived same-sex relationships as more problematic. Finally, both gifted and highly achieving individuals reported a lower frequency of meeting friends than controls.

Hollingworth’s (1942) classic study followed a sample of 12 extremely gifted children (IQ = 180, initial age = 12) from New York until they were in their early twenties. According to Hollingworth, these children experienced great difficulties in relating satisfyingly to their normal intelligence peers. In her words, they were „too intelligent to be understood by the general run of persons with whom they make contact”, leading to a state of „loneliness and personal isolation from their contemporaries” (p. 264). She saw the reason for this lack of understanding in the fact that „other children [did] not share their interests [and] their vocabulary” (p. 262).14

From her observations, Hollingworth concluded that extreme levels of giftedness are related to serious communication problems. Note, however, that her results are limited by several methodological weaknesses. First, her extremely small sample makes it difficult to generalize to broader populations. Second, it can be questioned whether her subjects really had an IQ above 180, since most intelligence tests do not reliably differentiate at such a high level. Third, Hollingworth herself contented that „as persons become adult, they naturally seek and find on their own initiative groups who are like-minded, such as learned societies” (p. 264). Thus, it could be that the adjustment difficulties she found in children can be compensated for in older age (but see Janos and Robinson, 1985, who speculated that adjustment problems might increase with age).


An analysis of the extremely gifted individuals from the Terman sample is consistent with the notion that extreme levels of giftedness are associated with problems. When his sample was 41 years old, 71% of the men (n = 551) and 67% of the women (n = 453) were rated as well-adjusted. The mean IQ score in this well-adjusted group was 136 and 131 for the men and women, respectively (Terman and Oden, 1959). In contrast, the combined groups of subjects with „some maladjustment” and „serious maladjustment” had an average IQ of 149 and 139. The differences between these two adjustment groups correspond to effect sizes of .42 and .28 for men and women, respectively (pooled SDs = 26 and 28). Conclusion

The review of the literature on the social adjustment of gifted individuals gives rise to some mixed conclusions. On the one hand, the classic Terman study and the literature reviews by Janos and Robinson (1985) and Hoge and Renzulli (1993) suggest superior generalized adjustment for gifted individuals. Somewhat in contrast, the findings of Rost (2000), Hollingworth (1942), and the re-analysis of the Terman sample suggest that gifted children could face some difficulties in their social relationships, especially at extremely high IQ levels.15

The notion that only extreme levels of giftedness might be associated with problems in social relationships is consistent with the predictions of the Simonton model. As stated before, individuals with extremely high intelligence levels should indeed face the greatest trouble in communicating their thoughts and feelings, and these difficulties should translate into adjustment problems in the social domain. Even in modern-day educational and vocational systems that are stratified according to intelligence, extremely gifted individuals might face problems in finding cognitive peers.


Further support for the notion that extreme levels of giftedness cancel out the positive social effects of intelligence comes from a large-scale study by Schneider, Clegg, Byrne, Ledingham, and Crombie (1989). These authors tested 150 gifted children (Grades 5, 8 and 10, mean IQ ≥ 129) who were educated in special classes, 204 integrated gifted individuals (not enrolled in special gifted classes), and 193 controls (mean IQ ≈ 112). They found the correlations between IQ and social competence (measured by self-nominations) were mostly positive in the control group (r = .52 in Grade 5, r = .20 in Grade 8), non-significant in the self-contained (special education) gifted sample, and negative in the integrated gifted sample (r = -.22 in Grade 5, r = -.23 in Grade 8). In addition, the intelligence difference between integrated gifted individuals and the average of the control children in the same class was negatively related to its peer acceptance in Grade 5 (r = -.33).

The above considerations give rise to the following hypothesis:

Extreme Group hypothesis: Intellectually gifted individuals experience a lower level of MU in their social relationships.

2.6 Summary of Main and Dyadic Hypotheses


To summarize, the current study addresses a number of hypotheses about main and emergent effects of intelligence and dispositional valuations. Factors that exert a main effect influence the MU process independent of the personality of the relationship partner. The following three main effect hypotheses are addressed:

Main Effect Hypothesis 1: Fluid intelligence is positively related to MU.

Main Effect Hypothesis 2: Crystallized intelligence is positively related to MU.


Main Effect Hypothesis 3: Openness to experience is positively related to MU.

Besides focusing on main effects, the current study also focuses on dyadic effects that result from the dynamic interaction of both communication partners. A total of four dyadic hypotheses are addressed:

Dyadic Effect Hypothesis 1: Between-person differences in fluid intelligence are negatively related to MU.


Dyadic Effect Hypothesis 2: Between-person differences in crystallized intelligence are negatively related to MU.

Dyadic Effect Hypothesis 3: Between-person differences in openness to experience are negatively related to MU.

Dyadic Effect Hypothesis 4: Between-person differences in interests and values are negatively related to MU.


Combining main and dyadic effects, the following hypothesis regarding the quality of social relationships of gifted individuals is tested:

Extreme Group Hypothesis: Intellectually gifted individuals experience a lower level of MU in their social relationships.

Footnotes and Endnotes

9  Some researchers have proposed that tests of emotional/social skills are influenced by participants’ knowledge structures (e.g., Zeidner, Matthews, & Roberts, 2001), which are a form of crystallized intelligence. Because research on this issue is mostly lacking, social and emotional skills are discussed together with fluid intelligence.

10  McCrae and Costa (1997, p. 838) also speculated that open individuals have access to more thoughts, feelings, and impulses in awareness and are able to maintain more of these mental elements simultaneously in consciousness. These features of openness, however, correspond more to the categories of fluid intelligence (working memory) and crystallized intelligence (knowledge), respectively.

11  In spite of strong evidence for bias, empathic accuracy has been shown to increase with relationship length, at least under some circumstances (Neff & Karney, 2002; Thomas & Fletcher, 2003). Social judgment in close relationships is driven by both accuracy and bias, depending on the goals and motivation of the perceiver (Gagné & Lydon, 2004).

12  Of course, the fact that close-minded people are more easily persuaded by the opinions of other people makes the label „openness” somewhat problematic.

13  This corresponds to a deviation of 1.2 SDs above the mean; note that this is exactly the score Kennedy is supposed to have reached.

14  The social difficulties of Hollingworth’s gifted children may be reminiscent of the Asperger syndrome, which is characterized by impairments in social interaction and repetitive patterns of behavior in absence of any delays in cognitive development. However, individuals with Asperger have been shown to have substantial language problems (Koning & Magill-Evans, 2001), whereas Hollingworth’s subjects had superior language skills.

15  Social difficulties may co-exist with average or even superior levels of „generalized” adjustment. First, it is possible that some gifted individuals compensate problems in the social domain with the self-esteem gained by superior academic achievements. Second, the reviews of Janos and Robinson (1985) and Hoge and Renzulli (1993) are based on generalized adjustment measures that might be only weakly related to adjustment in specific social settings.

© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.
DiML DTD Version 4.0Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML generated: