SELF-ESTEEM MEASUREMENT & PROBLEMS

- 48 -

the issue about whether high self-esteem causes good grades or vice versa is still being debated point, but causality probably operates in both directions (Coopersmith, 1967;

Pope et al., 1988). A student’s view of his or her academic performance will certainly affect his/her self-evaluation. Conversely, an individual’s beliefs about himself will have a strong impact on how well he performs, sometimes in spite of his actual abilities.

However, the relationship of self-esteem with academic achievement will be examined in more detail in the following section.

- 49 -

Self-esteem has been measured in almost as many ways as there are of defining it.

However, for simplicity, self-esteem measurement methods can be divided into the following two categories: a) self-report methods and b) observational methods.

2.6.2.1 Self-Report Methods

Self-report methods based on the subject’s own responses can be given in a variety of forms. These methods are economical and practical in that they can be scored and interpreted easily, and the researcher can obtain a self-description from a subject in a short period of time because the measures are structured or semi-structured. Some possible problems with these methods would be fakebility, social desirability, response styles, and acquiescence. However, some of these problems can be controlled to some extent by, for example, using equal numbers of negative and positive statements, establishing rapport with the subjects, providing a non-threatening climate, and by assuring anonymity when administering the self-report instruments.

Self-report methods can be further subdivided into the following four main categories:

a) Rating Scales, which are the most common, although these may be prone to errors in central tendency, response set, and acquiescence. Other problems are in the differential meanings and ambiguity of trait names or scale units to the respondent. b) Adjective Check Lists, which are commonly used with children because they relate more effectively to complete thoughts rather than isolated words (like names or adjectives) for describing themselves. c) Semantic Differentials, which employ categories on a continuum which separate a pair of dichotomous traits, though such a form necessitates the respondent making finer distinctions about himself which he may unable to do.

Fewer points are more likely to add only chance variance to the individual’s judgement about himself. Another problem is the ground which he can use if he does not care to respond on a particular item. The advantage of this form is that it provides both direction and intensity of response on a continuum between two terms which are

- 50 -

opposite in meaning. d) Q-Sort, where the respondent sorts out statements which he/she perceives as ranging from least characteristic to most characteristic of himself / herself in a quasi-normal distribution of piles. Although the Q technique provides a certain uniqueness in measurement, individuals may be grouped according to similarity in profiles but may be entirely different in personality structure. Cluster analysis using some type of distance function could be a possible answer to this problem. One obvious limitation of the Q sort is that the procedure is time consuming when a large number of subjects take part in a research study, because the sorting of statements is usually administered individually to each subject.

Two other forms - open-ended questionnaires and pictorial or projective techniques - are also used occasionally. However, they both present particular problems of scoring and interpretation, and will not be described here.

2.6.2.2 Observational Methods

Here, self-esteem is inferred from the individual’s behaviour. The variety of this type ranges from the structured interview to the categorising of behaviours by a clinically trained observer or the measuring of the perceptions of a third person, notably one or more of the “significant other.” Direct observations are useful for very young children, where self-report methods may be inappropriate for their age. However, the presence of the observer may produce behaviour on the part of the subject which could differ if the observer were not present.

Finally, self-esteem may be inferred from the interaction of two or more types often by assessing the congruence between self-ratings and ratings of others.

- 51 -

2.6.3 Problems in Measuring Self-Esteem

As with any personality measure, there are problems in self-esteem assessment. These may be seen as essentially those of establishing construct validity. Construct validity is necessary because a subject’s cognitions and attitudes about himself are private and beyond direct observation by the investigator. In order to measure self-esteem the researcher must use some form of self-report where the response is made by the subjects as a basis for his inferences. Self-report behaviour has usually taken the form of verbal response or a choice response, when the subject is instructed to indicate specified conscious processes. Despite their limitations these methods seem to be the only kinds which are appropriate to this type of construct, in this researcher’s opinion. One of the problems in measuring self-esteem is that of social desirability. Subjects may attribute to themselves traits which social consensus would indicate are socially desirable and acceptable while rejecting those that are socially undesirable and unacceptable. It is easy to falsify responses so that a positive or a good picture is presented on a self-report scale. For example, a subject with low self-esteem may wish to hide this fact from others. Stair (1967) suggests that an individual with low self-esteem will try to hide his feelings of inadequacy when he/she interacts with other people whom he/she feels to be important. Another problem involves “faking good” subjects who may hide their true appraisal of themselves because they are afraid of the negative evaluation they will receive if they are honest. Wylie (1961) states “no way has been worked out to determine in what cases and under what circumstances the social desirability variable distorts individual self-reports away from validity in reflecting a subject’s phenomenal field” (p. 28). Cowen & Tongas (1959) consider social desirability a very serious validity threat. In fact several researchers have found a relationship between self-esteem measures and social desirability (Cowen & Tongas, 1959; Meisels & Ford, 1969).

- 52 -

Despite considerable interest in self-esteem and self-concept there have been persistent methodological problems in measuring this construct (Wylie, 1961, 1974; Wells &

Marwell, 1976; Shavelson et al., 1976; Burns, 1981). According to Gecas (1982) self-esteem measurement is still a “serious problem” in self-concept research. Wells &

Marewell (1976) described the self-esteem literature in general as having an

“indeterminant character”. Wylie (1961, 1974) was also quite critical of research in this area arguing that there are far too many instruments used to measure self-esteem and that most are never re-evaluated for their adequacy or perceived utility. Several researchers (Wylie, 1961, 1974; Crandall, 1973; Burns, 1981) reviewed a number of self-esteem measures and they found that most of them lack validity and reliability. The reviews of Wylie (1961, 1979) examine hundreds of research reports on self-esteem / self-concept. The conclusion of the author is that most studies purporting to explore self-esteem are, in fact, not measures of self-esteem at all. According to Wells &

Marwell (1976) the most frequent form of validation used in self-esteem research is simple face validity or substitution of faith for evidence. In the case of reliability, either internal consistency, split-half, or test-retest measures is frequently missing. In the majority of the instruments reviewed, internal consistency and split-half and test-retest reliabilities have not been reported. According to Wylie (1974) test-retest reliability estimates (whether using the same or alternate forms) are even rarer in the published literature. However, more recently, researchers have developed self-esteem instruments specifically to measure particular aspects of self-esteem that are based on a theoretical model, and to use factor analysis to test the existence of these aspects.

2.6.4 Summary

There have been a great number of self-esteem / self-concept measures available which vary greatly in their reliability and validity. One of the main criticisms which is always stated in the review of self-esteem theory, methodology and research, is that the

- 53 -

constructs used in the research are vague and not well defined (Wylie, 1961, 1974;

Shavelson et al., 1976; Wells & Marwell, 1976; Burns, 1981; Beane & Lipka, 1980;

Lawrence, 1981; Harter, 1982; Hansford & Hattie, 1982; Pope et al., 1988; Abdallah, 1989a). The second important criticism is that a large number of studies lacked information about validity and reliability of the scales. For instance, Wylie (1968) reported that 90% of the 22 sets of Q-sort scales she reviewed had no information about the construct validity. Similarly, two-thirds of the Adjective Check Lists, Rating Scales and Questionnaires lacked information about reliability. Finally, one of the weaknesses of the self-esteem measures is the non-equivalence of the measures. Therefore one cannot compare different test results. It is difficult to validate a self-esteem measure by comparing it with the results of another self-esteem measure. As Abdallah (1989a) points out, “one has to be very cautions when making a comparison between the results of one’s research and those of the previous researchers”.