Nomothetic and idiothetic measurement in personality

Abstract

Lamiell (1981Lamiell ( , 1982a has recently advanced an idiographic approach to the measurement ofpersonality that he has labeled "idiothetic " In contrast to classical nomothetic strategies, the idiothetic method of scaling personality traits discounts normative transformations of attnbute scores and, according to Lamiell, provides the basis for a useful index of cross-situational behavior consistency An analysis of Lamiell's procedures indicates several problems with the idiothetic approach to personality assessment and with the proposed measure of behavior consistency Notable i !> the fact that the method does not account for base rates of behavior m situations and, therefore, may suggest anomalous interpretations of personahty charactenstics and of behavior consistencies It is argued that conventional nomothetic methods are sufficient for the goals of explaining and predicting behavior A resurgence of interest in the idiographic study of personality has stimulated several innovative proposals for empirical research The Journal qf Personaltty recently pubhshed a special issue (West, 1983) devoted in part to a discussion of current developments in the apphcation of idiographic pnnciples fbr personality measurement and for the prediction of behavior Because of its emphasis on the unique and personal determinants of individual behaviors, it is generally thought that an ldiographic analysis ofpersonality will allow for greater accuracy m the prediction and explanation of behavior compared to traditional nomothetic techniques (e g ,

Comparing Idiothetic, Ipsative, and Normative Indices of Consistency

“…The current study did partially replicate Paunonen and Jackson's (1986a) observation that an individual's attnbutes identified as consistent across time using the idiothetic model will often be the same attnbutes identified by the normative method as being consistent across time for that individual There was a moderately high degree of withm-subject correlation between the vanance scores of the attnbutes (over the three occasions) as defined by the idiothetic method with the vanance scores as assessed by the normative model (Table 3) However, the correlations were substantially lower than those obtained by Paunonen and Jackson (1986a) Only 64% of the vanance was shared by the idiothetic and normative scores in the current study, compared to 88% in Paunonen and Jackson's study The results of the current study provided a more realistic estimate of vanation in scores across time because the occasions were actually across different time penods rather than being simulated, and additional vanation in the normative and idiothetic indices was allowed to occur…”

“…In addition, and most important, the substantial correlation of 94 obtained by Paunonen and Jackson (1986a) was due in part to how the data were analyzed Vanation in the idiothetic and normative transformations of the raw data was severely limited by the formulas they employed For example, when determining subjects' idiothetic scores, they used the same constants in the idiothetic formula for each attnbute, for each subject, and for each occasion The minimum and maximum scores for each subject, at each occasion, and for each attnbute were always the same That IS, the maximum and minimum values were always 2 and 18 given the weights ( 2, 4, 6, and 8) and the 1 to 9 scale (see Formula 1, Footnote 1) As a result, each idiothetic score simply involved the subtraction of a constant from the raw score, and the difference divided by another constant Thus, a correlation between the idiothetic indices and their corresponding raw scores would have to be 1 0 Similarly, Paunonen and Jackson did not calculate new norms for each occasion (l e , calculate a new mean and standard deviation) They instead used as norms for the second, third, and fourth occasions, the norms obtained from the means and standard deviations of the first occasion As a result, each normative score once again simply involved the subtraction of a constant from the raw score, and the difference divided by another constant The only change in the calculation for either the ldiothetic or the normative index involved a change in the formula for the normative score across the attributes The normative constants subtracted and divided from each raw score were different for each attribute (since there was a different mean and standard deviation for each attnbute) It may then be apparent that a high correlation between the idiothetic and normative indices was a likely outcome The only vanation that was allowed to occur involved the change m the normative score across attnbutes By mixing attnbutes, one is no longer adjusting the normative scores by a constant, for the means and standard deviations are different for different dimensions If this single vanation was not allowed, the correlation would have had to have been 1 0 The vanation that was allowed was not substantial, hence the correlation of 94 No vanation in the calculation of the idiothetic index was possible across occasions, across attributes, or across subjects No vanation m the normative index calculations across subjects is sensible (because the norm should be the same for all subjects), but the other constraints are not necessary and may have contnbuted to a high correlation by not allowing any unique vanation in information to occur…”

Self‐Ratings of Personality: A Naturalistic Comparison of Normative, Ipsative, and Idiothetic Standards

“…Across all three studies, we repeatedly found that the nonnative ratings were less extreme than the implicit, ipsative, and ldiothetic ratings and that it was the difference in response extremeness that mediated the differences m similanty To understand why the normative ratings are less extreme, consider the phenomenon refen-ed to by Heider (1958) as "egocentnc attnbution" and by Ross, Greene, and House (1977) as ttte "false consensus bias " We suggest that when responses are based on normative standards, the tendency to view oneself as being like others or "average" becomes salient, making normative self-ratmgs less extreme and closer to the average Alternatively, the greater extremeness of the ipsative and idiothetic ratings may have resulted from the explicit focus of their instructions on the extremes of the rating scale However, whether one chooses to explain the results by focusing on the less extreme normative ratings or the more extreme ipsative and ldiothetic ones, it appears that under minimal instructions subjects tend to anchor their judgments at the extremes rather than at the middle of the rating scale Similarity of the different rating instructions The existence of differences among the ratings should not overshadow that they are all very similar Because we found differences that were replicated across all three studies, we conclude that subject carelessness, fatigue, or inattention to the instructions cannot explain the overall pattem of similanty Instead, It appears that the high degree of similanty among ipsative, idiothetic, and normative sconng models, which has been well documented m the psychometnc literature (e g , Block, 1957, Paunonen & Jackson, 1986, IS also reflected in individuals' subjective judgments For those who wish to improve personality assessment, this result should be disappointing because it suggests that changing the standards of companson provided to the raters will not have much impact on their ratings Moreover, for those interested m understandmg the implicit standards that underlie personality descnptions, this result suggests that such standards are difficult to distinguish m naturalistic ratings Instead, in order to untangle their influence on self-evaluations, normative, ipsative, and ldiothetic standards will have to be manipulated so that their mdependent effects can be assessed (Goolsby, 1985) Individual differences in evaluative styles Perhaps our most provocative finding IS that there appear to be individual differences in the use of the evaluative standards studied here Although not addressed in this research, such differences may contribute to the discrepancies that are often (^served when the same person is rated by different mdividuals More generally, we hypothesize that evaluational styles are central to people's self-concepts, their affective reactions to the self and to others, and then* subsequent interpersonal and achievement behaviors (Campbell, Fairey, & Fehr, 1986) Therefore, the analysis of these individual differences will not only be important for personality assessment, but may also contnbute to our understanding of general social processes…”

“…This was demonstrated for ipsative and normative measurement some time ago (Block, 1957, Bun, 1937, Cattell, 1952, Eysenck, 1954, Heilbnin, 1963, Kogan & Fordyce, 1962) Block (1957, for example, reported median (uncorrected) correlations of 95 and 79 between ipsative and normative scores under two conditions Likewise, Kogan and Fordyce (1962) obtained correlations between the two types of measures that ranged from 63 to 70 and concluded that the measures had "practical equivalence " For the idiothetic model, this equivalence is often mandated by the pragmatics of the measurement situation Specifically, because the maximum and minimum responses that a subject could hypothetieally make are typically operational lzed as the endpoints ofthe actual rating scale, ldiothetic transformations are necessanly linear ones (Paunonen & Jackson, 1986) The result of this equivalence is that the different measurement models cannot be fruitfully compared using the traditional psychometric standards of validity coefficients and predictive utility Moreover, such standards are, by their nature, nomothetic, and as such are com of httle value in the ldiographic realm (Lamiell, 1981, Runyan, 1983, Rychlak, 1981 Instead, a more acceptable cntenon is the extent to which a measurement model captures the meaning intended by the rater We will compare the nonnative, ipsative, and idiothetic models on this altemative cntenon…”

Personality and social psychology: towards a synthesis

“…These criticisms have concentrated primarily on the empirical procedures and findings in support of the idiothetic model (for example, Conger, 1983;Paunonen and Jackson, 1986a, b;Woody, 1983). Among other points, Paunonen and Jackson (1986a) object that idiothetically defined behavioural tendencies fail to take into account differences in the base rates of behavioural criteria. To illustrate their point, they choose the example of a person scoring 0.50 on an idiothetic measure of aggressive behaviour which suggests that the person has performed about half of the aggressive acts covered by the measure.…”