The precise definition of personality has been a point of discussion amongst many different
theorists within many different disciplines since the beginning of civilisation. Personality can be
defined as "the distinctive and characteristic patterns of thought, emotion, and behaviour that define
an individual's personal style and influence his or her interactions with the environment" (Atkinson,
Atkinson, Smith & Bem, 1993: 525). It can be proposed that personality psychology has two
different tasks. "The first involves specifying the variables on which individuals differ from one
another. The second involves synthesising the psychological processes of human functioning into an
integrated account of the total person" (Atkinson et al., 1993: 532). There are many different theories
of personality and many different theorists. The purpose of this essay is to examine the trait approach,
specifically the five-factor model. Both the development and limitations of the Five-Factor model of
personality shall be discussed.

    Trait theory is based on several assumptions. The first assumption is that any difference
between people that is seen as significant will have a name. Secondly, these names, known as traits,
are conceived of as continuous dimensions. In general, trait theories assume that people vary
simultaneously on a number of personality factors. These traits are of both the conjunctive and
disjunctive form. Therefore, to understand a trait, it is necessary to understand what a particular trait
is and what type of behaviour is evidence of that trait. (Atkinson et al., 1993). Five factor theorists are
one set of trait theorists. The claim of five factor theorists is that behaviour can be best predicted and
explained by measurement of five dominant personality factors. The five factor theory is a fairly
recent proposal and has its basis in earlier work, which shall be discussed.

    One of the statistical techniques most commonly used in the study of personality is that of
factor analysis:
    By identifying groups of highly intercorrelated variables, factor analysis enables us to
    determine how many underlying factors are measured by a set of p original variables. In other
    words, factor analysis is used to uncover the factor structure of a set of variables. (Diekhoff,
    1992: 333)
A factor analysis will generally show that a smaller number of factors represents the same information
as the original number of variables. Once the variables making up the factors have been identified,
some of the redundant variables may be removed (Diekhoff, 1992). As such, a large number of traits
may be reduced to a number of personality factors. The procedure of factor analysis was a significant
part of both the development and criticism of the five personality factor theory, as well as the theories
on which it is based.

    An experiment conducted by Allport and Oddbert (1936, cited in Goldberg, 1990) was based
on the assumption that a dictionary contains a list of every possible trait name. Oddbert and Allport
took every word from a dictionary that related to personality descriptors. This list was then revised to
remove synonyms and unclear or doubtful words. Another researcher, Raymond Cattell (1945, cited
in Atkinson et al, 1993) further revised the Allport-Oddbert list to 171 words. A study was then
conducted by Cattell on a group of subjects who were asked to rate people they knew on the 171
traits. The results were factor analysed and 12 personality factors were found. However, 4 additional
factors were found by analysing self-ratings. Cattell concluded that, in the adult human, 16
personality factors were dominant.

    Eyesenck, (1953, cited in Atkinson et al, 1993) was another major theorist to use factor
analysis. Although using the same basic approach as Cattell, Eyesenck used a more discriminatory
factor analysis which resulted in far less than 16 factors. Eyesencks' major factors are introversion-
extroversion and neuroticism. These are believed to be ordinal factors and as such, scores on each
dimension are independent of one another. The majority of future studies concluded that the actual
number of personality factors, for which there is significant evidence, is between Eyesencks' two and
Cattells' 16.   

    Since Cattells' study, many researchers have conducted similar studies, or re-analysis of
Cattells' original data. Most of the researchers, such as Norman (1967, cited in Merenda, 1993)
found support for far less than 16 personality factors. At most, it was generally concluded that there
are between three and seven factors of personality. As a compromise, many researchers agree that
there are five personality factors, as suggested by Norman's original work (1963, cited in Goldberg,
1990). Support for the Five-Factor model comes from current researchers such as McCrae and Costa
(1985) and Goldberg and Saucier (1995). Opposition to the theory is also abundant, such as the work
of Jack Block (1995).

    All trait theorists agree that there is a finite number of traits on which people have a "score".
The exact number of traits is still currently a point of contention amongst theorists. However, "today
we believe it is more fruitful to adopt the working hypothesis that the five-factor model of personality
is essentially correct." (McCrae & John, 1992: 175). There is also still "disagreement among analysts
as to factor titles" (John, 1990: 96). Many writers have adopted the names used by Norman (1963,
cited in Goldberg, 1990) which are; extroversion, agreeableness, conscientiousness, emotional
stability and culture. For simplicity, this is the version of the five factor model that shall be adopted
for this essay.

    The best known limitations of the five factor model of personality relate to the problems of
trait theory in general. Trait approaches are directed primarily at specifying the variables of
personality. There is little dealing with the dynamic processes of personality functioning. Traits are
static entities and more complete theories of personality, such as those of Eyesenck, come from a
combination of trait theory with another psychological theory. For example, Eyesenck adopted a
learning theory to combine with trait theory. As such, trait theory, and therefore the five factor model,
do not deal with a large aspect of personality: change.

    Mischel (1968, cited in Atkinson et al. 1993) is perhaps the best known critic of the trait
theorists. Basically Mischel states that the underlying assumption of the approach may be untrue:
people may have such dynamic personalities that they do not possess trait-like characteristics. Mischel
also claims that there should be a high correlation between scores on a trait measure for a subject and
performance in a situation where that trait is evoked. However, according to Mischel, the correlation
is extremely low. Mischel further argues that knowing a persons' "traits" does not help predict their
behaviour and measures of the same trait do not correlate highly with one another. Although this
criticism seems almost perfect, there is still a large number of trait theorists. Their responses to
Mischel's criticism shall be evaluated.

    The main defence of the trait approach comes in two forms. Firstly a conceptual form in
which Mischel's understanding of what makes up a trait is questioned. The second form of defence
comes from a methodological perspective, where the measurement of "trait" behaviour is examined.
To be able to appropriately comment on trait theory, it is important to understand exactly what a trait
is. McCrae and Costa (1995) suggest that not every person has every trait. Therefore it is possible to
confuse descriptors of behaviour with traits. There needs to be consistencies of behaviour to evidence
a trait. Also traits can be of either a conjunctive or disjunctive type. It has been suggested that the
evidence suggested by Mischel is invalid because aggression was seen as conjunctive when it is
actually disjunctive. Correcting this mistake could significantly increase the correlation between
different measures of the same trait. As such, one criticism of Mischel may be answered.

    The second defence of trait theory examines the research method used by Mischel. It is
proposed that it is necessary to have many more than one observation of behaviour, before comparing
behaviour to trait scores. The reasoning behind this argument is that each trait test has at least 20 to
40 items. As such, there should be at least half as many observations. A single question test would be
unacceptable and therefore a single observation of behaviour should also be unacceptable. Another
possible experimental error may have occurred due to moderator variables. Moderator variables such
as sex of subject may change the correlation between behaviour and trait scores. If these variables are
controlled for, the correlation may significantly increase and Mischels' criticism may need to be re-
evaluated.

    Cattell's 16pf, the predecessor of the five factor model, also had a significant limitation. The
16 pf had a low predictive power of performance of a subject on a given test, when used alone.
However, the personality profiles which can be created using the 16pf are reasonably effective in an
applied situation in predicting adjustment of an individual entering a particular group. Also, the
performance predicting power of the 16 pf can be improved by giving the 16pf and correlating it to
some measure of the person's performance. Multiple regression can then be used to weight each of
the 16pf factors so that correlation between the 16pf score and performance is at maximum. This
gives a more satisfactory prediction of performance using the 16pf, yet it's predictive power is still
quite low. The 16pf is still used in many applied situations because no other psychological tool is
available with better predictive power. Since the five factor model is based on the 16pf, this limitation
is also applicable to the five factor model.

    It is possible to suggest that the limitations pertaining to the trait approach and 16pf are
insignificant or not applicable to the big five model of personality. However, there are limitations that
specifically relate to this model. Jack Block (1995) and Dan McAdams (1992) are the main theorists
to evaluate the five factor model specifically and examine it's limitations. Block's criticisms are
answered by theorists such as McCrae and Costa (1995) and Goldberg and Saucier (1995).

    The basis of Block's argument is that it is uncertain that all important trait-descriptive terms
are representatively distributed in language. For instance, collectively suppressed traits might be
unrepresented. Another major point is that the Big Five are very broad and might not differentiate
accurately enough for practical applications. For example, assigning people to high, middle and low
on each of the factors gives 243 personality types, which may be enough types but doesn't solve the
broadness problem. Block suggests a few changes to procedure should be adopted but admits "my
suggestions are mild, obvious and entail scientific sobriety coupled with slow, hard work aiming to
educe order from the present jumbled empiricism characterising personality psychology". (Block,
1995: 209).

    Both Costa and McCrae (1995) and Goldberg and Saucier (1995) suggest that Block has lost
sight of why the five factor model was developed. Block criticises the model for not being applicable
to practical situations when it's purpose is to describe the full range of personality traits. Block's
criticism also "does not distinguish between the Big Five model ... from alternative models of the
causal underpinnings of personality differences" (Goldberg & Saucier, 1995: 221). A large amount of
crucial evidence supporting the Big Five model is also left out of the criticism. Each reply also
suggests that Block's closing suggestions provide few specific proposals of alternative models.

    McAdams' (1992) critical appraisal of the five-factor model outlines several major limitations.
McAdams views the five-factor model as "essentially a 'psychology of the stranger', providing
information about persons that one would need to know when one knows nothing about them. It is
argues that because of inherent limitations, the Big Five may be viewed as one important model in
personality studies but not the integrative model of personality". Some of the limitations described are
those applicable to all trait theories and one applies to the 16pf and any theories based on the 16pf.
However, two limitations specific to the five factor model are discussed.

    The main limitation specific to the five factor model of personality are firstly a failure to offer
a program for studying personality organisation and integration and secondly a reliance on statements
about individuals by other individuals. The extent to which the five-factor model is a major advance
in personality study therefore depends on what is hoped to be gained in the field. If personality study
is interested in the study of observer's trait ratings, the big five model is extremely useful. If the
purpose of the field is also to investigate observers' attributions about individual differences the five-
factor model is less significant. If the study of personality aims to emphasise the whole person and the
dynamic nature of personality, the model seems to be only of minor concern. As such, from the view
of "multifaceted personology, the five-factor model is one model in personality... not the model of
personality" (McAdams, 1992: 355).

    In conclusion, the support and criticisms of the five factor model are not as black and white as
would be hoped. Each argument has logical reasoning and can provide evidence to support itself.
Each view also has a large number of supporters. Neither one is necessarily correct, as it is possible
for the model to be applicable at some stages, and not applicable at others. As a result, it is probable
and acceptable to conclude that the five factor theory may or may not be an appropriate model of
personality. Perhaps a comparison of how much supporting literature there is for each argument is a
useful method for deciding which theory an individual may choose to support.
--------------------------------------------------------------