The Utility of Traits in Explaining Personality
Personality can be conceptualized in a number of ways. We think of personality in terms of an individual's nonanatomical characteristics (e.g., relational patterns, emotionality, self-presentation), unconscious motives, unconscious childhood residuals, heritage, brain physiology, and traits. Probably the most prevalent conceptualization of personality today is in terms of traits, which may or may not embody aspects of other conceptualizations or aspects of personality such as heritability or brain physiology. The purpose of this paper is to discuss the advantages and disadvantages of using the trait conceptualization to understand personality.
Traits have been defined as "characteristics on which everyone can be judged" (Briggs, 1999). Briggs presents "talkativeness" as an example of a trait, and distinguishes a trait from a "personal disposition", which characterizes a person more individually and specifically. (He uses "fastidiousness" as an example of a personal disposition.) One problem with this particular distinction, and a problem with conceptualizing personality in terms of traits, is what the level of description or analysis is. From a linguistic point of view, what is the difference between fastidiousness and talkativeness (per Briggs, a personal disposition and a trait, respectively)? It seems as though anyone could be placed along a trait continuum of either fastidiousness or talkativeness, yet Briggs and Allport (whom Briggs cites) consider talkativeness to be a broader descriptive personality term than fastidiousness.
Another word for personal disposition, and the one that some trait theorists (e.g., Costa and McCrae) use is "facet". Costa and McCrae (1992), the authors of the NEO-PI-R, a popular and validated personality measure based on the five-factor model, organize personality constructs into factors and facets. Factor seems to be synonymous with the above definition of trait. Yet another synonym, dimension, is also generally understood as synonymous with factor (i.e., it is used to describe personality as subsuming facets), but it is also sometimes used as another, higher level of analysis. However, each of these levels can be thought of as a set of traits. The point here is that one of the biggest disadvantages of using "traits" to describe personality is that one study or theory may use it to describe one level of personality (e.g., facet) and another study or theory will use it to describe another (e.g., dimension). The word is used to describe any number of analysis levels; it is merely a label or taxonomic unit, and so for those purposes, it is useful. The problem with trait as a label is that it lacks depth. Clearly, personality is more complex than a set of labels.
Many seminal personality theories use more than one level of analysis to reflect the depth and complexity of personality. However, hierarchies or taxonomies differ between theorists (e.g., Cattell and his 16 traits; Eysenck and his three dimensions; Costa and McCrae and their five factors and 30 facets). The problem may be one of semantics. Personality research has its roots in lexical studies. Can personality be distilled into simple linguistic descriptions? Allport (1936), Goldberg (1981), and McCrae and Costa (1997) have all relied heavily on language to classify personality. Empirical evidence, even across languages and cultures, has shown the five-factor structure to be robust (McCrae and Costa, 1997), and this model is based on language.
However, McCrae and Costa make a distinction between "personality language" and "personality structure" (p. 510, 1997). It is the five-factor personality structure that is maintained across cultures and languages, not the language. In their 1997 study, they report that several translations of their NEO-PI-R yielded either a different structure (e.g., only four factors in Hungarian; De Raad & Szirmk, 1994) or different constructs within a five-factor structure (e.g., Chinese; Yang & Bond, 1990). Natural language seems to limit the ability to classify and explain personality as a universal concept.
The key to understanding personality in terms of traits is to present it as a structure of traits, as McCrae and Costa and others have done. Unfortunately, even though the five factors have repeatedly shown structural validity, the facets within these factors have not (Goldberg, 2001; Costa & McCrae, 1991). It is difficult to determine whether this is a language problem, a measurement problem, or a structural problem.
On the surface, it seems to be a language problem. For instance, the NEO-PI-R's Agreeableness factor is comprised of the following facets: trust, straightforwardness, altruism, compliance, modesty, and tender-mindedness. It is easy to imagine a straightforward individual who is not necessarily agreeable. Conversely, it is imaginable that a person who relates with others agreeably does not necessarily trust others. Disagreeableness presumably lies at the opposite end of the Agreeableness spectrum. But does assertiveness? Is it not possible to be agreeable and assertive at the same time?
The facets of Extraversion can be similarly analyzed. Warmth is one facet of Extraversion. Clearly it is possible to relate to others warmly without being extraverted. It is also possible for an introvert to be assertive and active.
Based on the above logic, establishing structural validity for the facets might be a semantic problem. Even the linguistic labels of the NEO-PI-R factors may be problematic. Other theorists have different names for the five factors, for instance Emotional Stability for Neuroticism and Surgency for Extraversion. Regardless, the essence of the five factors has remained stable. What remains unstable is the ability to capture and measure aspects of personality and its structure. It is difficult to measure personality (or any other psychological construct, for that matter) in ways other than self- and other-reports of affects, behaviors, and cognitions (in short, traits). It is difficult but not impossible. Other methods for measurement include the use of physiological (e.g., EEGs) and observational instruments (e.g., videotapes).
Of course, some of the structural problems with lower-level traits of personality might be dealt with statistically. The use of structural equation modeling may help theorists to detect more precise patterns underlying each factor. On the other hand, perhaps the inability to validate the lower level of personality structure is a function of the uniqueness and individuality that characterize each of us.