Journal of Personality and Social Psychology 1985. Vol. 48. No.

5, 1328-1341

Copyright 1985 by the American Psychological Association, Inc. 0022-3514/8S/$00.75

Rotter's Internal-External Scale: Confirmatory Factor Analysis and Correlation With Social Desirability for Alternative Scale Formats
Neal M. Ashkanasy
University of Queensland, Australia Rotter's (1966) internal-external control of reinforcement scale has been widely used since its introduction. However, problems of dimensionality, social desirability correlation, and scale format have been reported. In this study, 536 psychology undergraduate students completed alternative versions of Rotter's internal-external scale and the Marlowe-Crowne (Crowne & Marlowe, 1960) social desirability scale. Of these subjects, 178 completed the scales in their original formats (forcedchoice and true/false, respectively), and 358 completed a Liken instrument that incorporated both measures. I hypothesized (a) that Watson's (1981) two-factor structure for the forced-choice scale and Collins's (1974) four-factor structure for the Likert version would be confirmed and (b) that both scales would exhibit significant social desirability correlation. A restricted factor analysis of the data confirmed the reported factorial structures, but the variance explained by the factors was small, and there was evidence that the scale may be tapping a unitary construct. Correlation with social desirability was found for the Likert data, but not for the forced-choice results. However, the data indicated that there may be serious problems with interpretation of results that were based on use of the Marlowe-Crowne scale, and, in this respect, the data were inconclusive.

The concept of internal versus external control of reinforcement (also referred to as "locus of control") holds that people, through a lifetime of social learning, acquire a generalized expectation about the source of reinforcement for their actions (Rotter, 1954; Rotter, Chance, & Phares, 1972). Rotter (1966) proposed that this expectation can be represented on an internal-external control continuum. At one end are "internals" who believe that reinforcement derives from within their own personalities; at the other end are "externals" who see reinforcement to be determined by fate, chance, or powerful others. Rotter (1966) argued that the internal-external construct is a stable, unidimensional personality variable, and proposed a 29-item
This study was part of the requirements for a doctorate in the Department of Psychology, University of Queensland, Australia. The author acknowledges and thanks Cindy Gallois, Bob Dick, and Henry Law for assistance in carrying out the study and in preparing this article. Helpful comments by Gordon O'Brien of Flinders University were also received with thanks. Requests for reprints should be sent to Neal M. Ashkanasy, Department of Psychology, University of Queensland, St. Lucia, Queensland 4067, Australia.

forced-choice instrument for its measurement. In subsequent years, the internal-external control construct, and Rotter's scale in particular, has found extensive application in the psychological literature. In 1973 and 1974, for example, Prociuk and Lussier (1975) recorded 277 separate studies, 69% of which entailed use of the Rotter scale; Rotter (1979) noted over 1,000 articles since publication of his original monogram. Joe (1971), Phares (1973), Lefcourt (1972, 1976, 1980), and Prociuk and Lussier (1975) described the application of the construct in clinical, social, educational, and organizational psychology. The general tenor of these findings is that internals are better adjusted than externals, and more able to cope with life. In particular, an external locus of control has been associated with hopelessness and depression (Joe, 1971; Lefcourt, 1976; Miller & Seligman, 1976; Prociuk, Breen, & Lussier, 1976). Despite the large volume of supportive literature, however, a variety of problems have been reported, particularly in relation to the construct validity of Rotter's (1966) scale. These include questions as to the appropriateness of the forced-choice format

1328

ROTTER SCALE ALTERNATIVES

1329

(Collins, 1974; Ray, 1980a), multidimensionality (Abramowitz, 1973; Collins, 1974; Gurin, Gurin, Lao, & Beattie, 1969; Gurin, Gurin, & Morrison, 1978; Lange & Tiggemann, 1981; Lao, 1970; Levenson, 1972; Watson, 1981), correlation with social desirability (Cone, 1971;Kestenbaum, 1976; Kestenbaum & Hammersla, 1976; Stem & Manifold, 1977; Vuchinich & Bass, 1974), culture specificity (Furnham & Henry, 1980; Lao, Chuang, & Yang, 1977; O'Brien & Kabanoff, 1981; Ryckman, Posen, & Kulberg, 1978), and a lack of cross-situational generality (Lefcourt, 1976; Lefcourt, von Baeyer, Ware, & Cox, 1979). Rotter (1975) dismissed these problems as resulting from a failure to understand properly the social learning underpinnings of the construct, and from consequential misapplication of the scale. Rotter (1975) argued that the original scale had been subjected to a rigorous program of validation, and therefore it should be strictly adhered to when addressing internal versus external control of reinforcement as a generalized variable. In particular, subscales and variants should be explicitly acknowledged as applicable only to the context in which they were developed. However, the weight of evidence has continued to accumulate since 1975 and indicates that the problems exist across a broad spectrum of situations and subject populations. An apparent shortcoming of this evidence, however, is that the problems have generally been studied in isolation. In particular, factor analytical studies have been exploratory rather than confirmatory, and are therefore of only limited validity (Armstrong, 1967). The aim of this study was partly to redress this problem in a design incorporating confirmatory factor analysis and correlation with social desirability for alternative scale formats. The issues of scale format, correlation with social desirability, and dimensionality are intimately linked and bear on the ultimate construct validity of the scale. Rotter (1966, 1975) stressed that the forced-choice format was adopted specifically to balance out social desirability. Early versions of the scale, which did in fact involve Likert responses and were intended to be multidimensional, were rejected on the basis of high correlation with the Marlowe-Crowne social desirability scale (Crowne & Marlowe, 1960). Nevertheless,

subsequent studies have consistently indicated that the Rotter scale is subject to a social acquiescent response set (Cone, 1971; Kestenbaum, 1976; Kestenbaum & Hammersla, 1976; Vuchinich & Bass, 1974). Explanations for this phenomenon include propositions that internality is of itself a positive societal value (Stern & Manifold, 1977) and that locus of control and social acquiescence are interacting variables that define subgroups of "congruent" and "defensive" internals and externals (Evans, 1980; Hochreich, 1974, 1975). The evidence relating to the dimensionality of the Rotter (1966) scale is even more consistent than that for social desirability correlation. Mirels (1970) identified three factors, as follows: (a) a political or powerful-others dimension; (b) a chance, luck, or unpredictable-world dimension; and (c) an internal or personal-control dimension. Subsequently, two- and three-factor solutions have been consistently reported in the literature (e.g., Butler & Burr, 1980; Lange & Tiggemann, 1981; Lao, 1970; Levenson, 1972, 1973, 1974; O'Brien & Kabanoff, 1981). Watson (1981) analyzed two-, four-, and five-factor structures reported in the literature, and concluded that the factor interpretations are stable provided that the number of factors is limited to two (i.e., general and political factors). The dimensions of chance and political control have also been shown to have construct validity in specific situations. For example, the political dimension has been shown to have predictive validity in situations involving social issues (Abramowitz, 1973; Gurin et al., 1969, 1978; Mirels, 1970), whereas the chance dimension has proven useful in studies involving personal orientation and ideology (Gurin et al., 1978; Ray, 1980a). These findings seem overall to provide strong evidence for a multidimensional interpretation of the internal-external control construct that is based on the Rotter scale. On the basis of the forgoing conclusions that the Rotter (1966) scale may be multidimensional and subject to social acquiescence, one could argue that Rotter's (1966, 1975) reasons for abandoning a Likert format may not be good ones. There is also evidence that the Likert format is superior to other formats in other respects. Ray (1980b), for example,

1330

NEAL M. ASHKANASY

found that a Likert instrument was a more valid measure of achievement motivation than a forced-choice equivalent. Collins (1974) noted that item pairs in the Rotter scale are not directly comparable, and he provided evidence that the forced choices are unnecessary. In a comparison of the alternative formats, he concluded that "[the] data clearly demonstrate that the 23-item forced-choice format and the 46-item Likert format measure the same dimension of personality" (p. 383). Collins's analysis of the Likert version yielded the three factors listed in the last paragraph, plus a fourth: belief in a just versus unjust world. He also found that, except for the political responsiveness dimension, item pairs from the forced-choice format load on different factors. This would seem to support Collins's assertion that Rotter's item pairs are not comparable. Subsequent studies have replicated Collins's (1974) dimensions in a variety of social and cultural settings (Barling, 1980; Barling & Bolon, 1980; Duffy, Shiflett, & Downey, 1977; Ryckman et al., 1978; Zuckerman & Gerbasi, 1977a, 1977b; Zuckerman, Gerbasi, & Marion, 1977). In particular, these studies have offered evidence that the dimensions have predictive validity. Scaturo and Smalley (1980) showed that the political dimension correlates with several measures of political attitudes and preferences. Zuckerman and his colleagues (Zuckerman & Gerbasi, 1977a, 1977b; Zuckerman et al., 1977) provided evidence that the just-world dimension may be essentially independent of the other locus of control dimensions. There is also strong evidence that the factorial structure of the Likert format is more stable across situations and cultures. Furnham and Henry (1980) found marked cross-cultural differences in the factorial structure of the forced-choice scale, and O'Brien and Kabanoff(1981) found differences for different social groupings within the same culture. In the case of the Likert format, on the other hand, the evidence from cross-cultural studies is more supportive of the basic structure (e.g., Barring & Bolon, 1980; Ryckrnan et al., 1978), although some changes in factor loadings and the emergence of additional factors has been reported. Duffy et al. (1977) reached a similar

conclusion in respect to differences between civilian and military populations. In summary, it is clear that despite the large volume of data supporting the existence of internal-external control as a personality variable, problems concerning the construct validity of the Rotter (1966) measure still remain. The problems encompass doubts about the appropriateness of the forced-choice format, correlation with social desirability, dimensionality, and sociocultural specificity. The use of a forced-choice format does not seem to have solved the problem of socially desirable responses, perhaps because such responses interact with the internal-external construct. In addition, factor analytical studies have shown that the forced-choice format proposed by Rotter does not tap a unidimensional construct. Investigators of a Likert version of Rotter's questionnaire have elaborated on the dimensionality of the measured construct, and it appears that the Likert format dimensions are more robust across social and cultural situations. It seems therefore that the problems affecting the Rotter scale are intimately linked and are appropriately investigated in a design that incorporates both of the alternative formats. The hypotheses of this study were therefore as follows: 1. Scores on both the Likert and forcedchoice versions of the Rotter internal-external control scale are significantly correlated with Marlowe-Crowne social desirability scores. 2. The four-factor model reported by Collins (1974) for the Likert format is confirmed for the sample in this study. 3. The two-factor model reported by Watson (1981) for the forced-choice format is confirmed for the sample in this study. Method Subjects
Subjects were from a pool of first-year psychology undergraduates, who agreed to participate in exchange for partial credit in an introductory psychology course. A total of 545 subjects participated, of whom 536 successfully completed the questionnaires. Subjects were randomly allocated to two groups: 178 (62 male, 116 female, average age 21.6 years) completed the forcedchoice scale and 358 (129 male, 229 female, average age 22.5 years) completed the Likert version.

ROTTER SCALE ALTERNATIVES

1331

Table 1 Means, Standard Deviations, and Intercorrelations for Alternative Format Locus of Control Scales

No. variable

SD

Forced-choice scale" 1 . External control 2. Social desirability

12.67 16.60 4.09 2.43
.01

Likert format scaleb 1. 2. 3. 4. 5. 6. External control Social desirability Difficult world Unjust world Chancy world Politically unresponsive world
b

191.10 123.18 46.23 43.52 27.28

40.71

24.68 21.28 8.98 7.92 7.62

8.72

-.39** .72** .54** .71**

.58*

-.32** -.13* -.34**

-.16*

.10* .43**
.24*

.34**

.15*

.12*

"n = 158. n = 358. *p<.05. **p<.01.

Materials
Original format scales. The Rotter internal-external control scale and the Marlowe-Crowne social desirability scale were presented respectively in forced-choice and true-false formats as specified by their original authors (Crowne & Marlowe, 1960; Rotter, 1966). (It should be noted that the Rotter instrument includes six filler items, giving a total of 29 forced-choice items.) Liken format instrument. This questionnaire was assembled by means of randomly mixing the 46 decoupled forced-choice items from the Rotter (1966) scale with 33 Marlowe-Crowne items, and 24 items from a self-attributional scale developed by Lefcourt et al. (1979). (The latter scale was not used in the current study; refer to Ashkanasy, 1982). Subjects responded to the resultant 103-item instrument on a 7-point structured Likert-type response scale. Full details of the instrument, including the location of items from the original questionnaire, is given by Ashkanasy (1982).

as quickly as possible, not to linger over individual responses, and not to go back to check on consistency. Most subjects were able to complete the material in 20 min, although the maximum time taken was 55 min.

Results Personal Data Variables

A check on bivariate correlations between personal data variables and internal-external scores (scored in conventional external direction) indicated a significant (p < .001) correlation between age and externality for both groups (rs = .18 and .24 for the Likert format and original format groups, respectively). The tendency for older people to have a more internal locus of control is well known (Lefcourt, 1976) and does not violate the Procedure social-learning basis of the construct (Rotter Subjects completed the questionnaire in group sessions et al., 1972). Furthermore, the amount of and were told that the study was being undertaken as a preliminary to a larger investigation of leadership and variance accounted for by subject age is small evaluation, with an emphasis on comparing alternative and would not therefore seem to constitute a threat to the validity of the present study. measures of the same construct.
The alternative questionnaire instruments were distributed to subjects in random order, in the ratio of two Likert instruments to one set of original format items. Order effects were controlled for in the Likert questionnaire by means of stapling the pages together in random order. The original format instruments were also distributed in random order within the sets. Subjects were required to complete a section providing personal data before completing the scored instruments. There were asked to complete the items in stapled order

Intercorrelations Between Measured Variables In Table 1, I show bivariate correlations between internal-external control scores and social desirability for both scales, and intercorrelations for Collins's (1974) subscales.

1332

NEAL M, ASHKANASY

Correlation with social desirability scores. As hypothesized, the correlation between Marlowe-Crowne social desirability score and locus of control measures involving a Likert scale format was significant. The correlation between social desirability and the original format scales, however, was not significant. This latter finding may be due to insufficient variance in the true-false format social desirability scores. Closer examination of the raw scores indicated that the overall range of scores was from 11 to 24 (out of a possible 0 to 33); 97% of scores were in the 13-20 interval, and 71% of scores were in the 1519 interval. This lack of variation seems to call into question the utility of the true-false format of the Marlowe-Crowne scale. Correlations between Collins's (1974) subscale scores. In Table 1, I show that the factors were not orthogonal as Collins suggested. Correlations between the predictableversus-chancy-world dimension and (a) the difncult-versus-easy-world dimension and (b) the just-versus-unjust-world dimension were relatively strong. Nevertheless, this does not of itself render implausible the possibility of an underlying factorial structure, because the proportion of shared variance in each case is small. Comparison of Scores Between Alternative Scale Formats Overall scores. In Table 1, overall means and standard deviations for the alternative scale formats are given. One can reduce the mean scores for the Likert format to a figure comparable to the forced-choice format by subtracting 46 and dividing by 12; similarly, one can reduce the variance by dividing by 12. This gives mean scores of 12.67 and 12.09 for the forced-choice and Likert versions, respectively, with corresponding standard deviations of 4.09 and 7.12. The mean scores are similar to those recorded in other recent studies involving Australian undergraduates (O'Brien & Kabanoff, 1981), as is the standard deviation for the forced-choice case. The higher standard deviation obtained for the Likert data results from the structured scale, which tends to spread item response scores. Item scores. In Table 2, item-by-item mean scores for both formats are listed (see the Appendix for item text), along with social desirability correlations for the Likert format

items. The correlation between mean forcedchoice item scores and corresponding mean Likert item pair differences was .90, indicating
Table 2 Item Statistics for Alternative Scales and Likert Scale Item Correlation With Social Desirability Likert scale"
No. variable 2 a b 3 a b 4 a

Forcedchoice11

Item type
E 1 I E I E I E E I E I E 1 I E I E I E I E I E E I E I E I E I E I I E E I E I I E I E
E I

M
4.63 3.78 2.96 4.94 4.00 4.67 3.34 3.61 4.06 3.91 5.33 4.56 4.80 4.85 4.15 3.60 4.43 4.85 3.80 4.63 4.78 3.93 4.57 3.81 3.81 4.51 4.60 4.07 4.17 3.16 4.25 3.93 4.75 3.62 3.28 5.00 4.67 5.58 4.02 3.82 3.57 4.55 4.71 3.97 4.21 4.32

SD
1.76 1.60 1.72 1.94 1.87 1.65 1.76 1.42 1.65 1.79 .47 .66 .43 .65 .97 .81 1.74 1.84 1.84 1.74 1.47 1.76 1.67 1.87 1.72 1.69 1.75 1.69 1.53 1.93 1.85 1.91 .68 .83 .75 .58 .67 .49 1.79 1.86 1.81 1.81 .62 1.83 1.87 1.60

r
-.19** .02 .00 -.04 .11* -.04 .18** -.17** -.16** -.01 -.21** .12* -.04 .20** .12* -.15** .28** .15** .15** -.12* .12* -.19** .25** -.27** -.17** .20** -.11* .13* -.09 .20** -.15** .09 .11* .01 .13* -.07 -.28** .08 -.18** .17** -.02 -.02 .11* -.27** .08 .05

M
0.52 0.76 0.71 0.63 0.41 0.72 0.44 0.31 0.51 0.72 0.37 0.34 0.35 0.66 0.70 0.68 0.76 0.75 0.32 0.57 0.64 0.37 0.46

SD
0.50 0.43 0.45 0.48 0.49 0.45 0.50 0.46 0.50 0.45 0.48 0.47 0.48 0.47 0.46 0.47 0.43 0.43 0.47 0.50 0.48 0.48 0.50

b
5 a

6 a b 7 a b 9 a b 10 a b 11 a b 12 a b 13 a b 15 a b 16 a b 1? a b 18 a b 20 a b 21 a b 22 a b 23 a b 25 a b 26 a b 28 a b 29 a b

Note. Scored in external direction; see Appendix for item text. E = external, I = internal. ' n = 358. b n = 158. * p< .05. ** p < .01.

ROTTER SCALE ALTERNATIVES

1333

that subjects' responses to items were independent of presentation format. This correlation is identical to that reported by Collins (1974). A further observation that accords with Collins's (1974) findings is that several nominally "internal" items scored in an external direction and vice versa. As Table 2 indicates, nine Likert "internal" items exhibited mean scores significantly greater than 4.0 (p < .05), and four "external" items scored significantly in an internal direction. Of 23 forced-choice item pairs, 8 scored significantly in the reverse direction of what would be expected. An extreme example is Item 23b: "There is a direct connection between how hard I study and the grades I get." At face value, affirmation of this statement is clearly an "internal" response. Yet the mean response for Item 23b was 5.58, which is a highly significant external response (p < .001). Clearly, this item must reflect a specific belief by students that study effort is futile. In this instance the response may therefore be sample specific and the item may well remain a valid inclusion in the scale. Finally, with respect to item correlations of Likert responses with social desirability scores, it should be noted that the direction of correlation was an almost perfect indicator of internality versus externality. (The exceptions were Items 2 la and 29a, but correlations in these instances were very small.) This would seem to provide some support for an argument that internality is a positive societal value (Stern & Manifold, 1977). Thus the data seem to support a conclusion that the alternative scale formats measure the same construct. However, score reversals on several items indicate that sample-specific responses are likely contaminants of both scales. The correlation between social desirability and Likert items suggests further that internality may be an intrinsically positive value. Factorial Structure The factorial structure analysis procedure used in this study was based on Joreskog's (1969, 1970) restricted factor analysis method. In this procedure, one uses a maximum likelihood algorithm to determine parameter values for a specified structure that minimizes

the difference between modeled and observed variance-covariance matrices. The maximum likelihood criterion approximates a chi-square statistic and is therefore suitable for significance testing. However, for large sample sizes, the chi-square statistic can discriminate second-order differences that are of no practical value. Joreskog (1970) suggested that as a general rule, a chi-square less than or equal to the number of degrees of freedom represents an overfilled model and lhal more attention should be given to chi-square differences between models. However, from the examples he gave and from further examples discussed by Mulaik (1975), it appears that an appropriate fit is obtained for a chi-square in the range 1 X dfto 2 X df, and a marginal fit is indicated by chi-square in the range 2 X df to 4 X df. Mulaik (1975) stressed the importance of carefully inspecting the matrix of residuals between the observed and calculated variance-covariance matrices. In Table 3, I show chi-square and degrees of freedom for five models fitted to the data obtained from the alternative scale formats. Table 3
Results of Restricted Factor Analysis
Model Likert format* Orthogonal, rotated principal axis Fixed communalilies Maximum likelihood communalities Collins's (1974) factor loadings Collins's (1974) factor structure only Orthogonal factors Oblique structure

df

1081 1035 1035

3943 3538 3850

1005 999

2853 2564

Forced-choice format" Orthogonal, rotated principal axis Fixed communalities Maximum likelihood communalities Watson's (1981) factor loadings Watson's (1981 (factor structure only Orthogonal factors Oblique factors Note. All x2s significant at p < .01. " n - 358; 4-factor solution. "n = 158; 2-factor solution.

276 253 253

503 463 499

246 245

434 416

1334

NEAL M. ASHKANASY

It is clear from these results that all of the models provided a satisfactory fit to the data if the criteria given in the previous paragraph are adopted. Inspection of residual matrices has confirmed this conclusion: 70-80% of residual coefficients were less than .01, only 5-7% were greater than .20, and 1-2% were greater than .30. More detailed results pertaining to each scale format are given in the following paragraphs. Forced-choice format. In Table 4, 1 give details of factor loadings and communalities that I obtained using the various models. The data from the literature were obtained from Watson (1981), who used a sample of Australian undergraduates, and from a two-factor structure defined by loadings (in Watson's study) greater than .40. These factors are similar to those reported elsewhere in the literature (e.g., Mirels, 1970; Lange & Tiggemann, 1981), and they constitute dimensions of personal control and political responsiveness.

The rotated principal axis (RPA) solution yielded factor loadings that were generally close to those given by Watson (1981). However, the factorial model accounted for only 18% of the total score variance, and 14 of the 23 communalities were less than .20. The Joreskog (1969) chi-square statistic for this model was less than 2 X df, indicating an acceptable fit. However, the close fit may simply have reflected the low communality between item scores. This conclusion is reinforced by results of fitting the RPA model with unconstrained communalities. The fit is considerably improved, but only at the expense of still lower communalities: Only 5 of 23 item communalities were greater than .20. Repeating the restricted factor analysis with Watson's (1981) factor loadings resulted in only slight degradation of model fit, indicating an intrinsic stability of factor loadings. However, a clearly better fit was obtained if only the factorial structure obtained by Watson was restricted; this yielded out of 15 corn-

Table 4 Forced-Choice Scale: Factor Loadings and Communalities for Rotated Principal Axis (RPA) and Maximum Likelihood (ML) Solutions
RPA solution Watson's (1981) structure

RPA loads

Communalily
2 .00
.29 .14

ML loads
1 .00 .00 .00 .67 .00 .00 .00 .12 .58 .00 .51 .62 .35 .00
.58 .11

Item no.
2 3 4 5 6 7 9 10 11 12 13 15 16
17

1 .36 .00 .23 .33 .41 .08 .24 .08 .57 .06 .31 .61 .47 -.03 .58 .16 .25 .06 .28 .64 .15 .32 .07

RPA .13 .09 .07 .12 .18 .01 .06 .09 .33 .39 .12 .39 .26 .28 .34 .03 .08 .25 .13
.41

ML .02 .10 .02 .23 .11 .03 .04 .03 .33 .22 .18 .35 .13 .19 .31 .03 .08 .00 .14 .18 .04 .00 .05

2 .00 .30 .00 .00 .00 .00 .00 .00 .00 .62 .00 .00 .00 .54 .00 .00 .00 .03 .00 .00 .00 .00
.07

Communality"
.00 .09 .00 .45 .00 .00 .00 .01 .33 .39 .26 .39 .12 .29 .33 .01 .00 .00 .00 .08 .00 .00

18 20 21 22 23 25

26
28 29

-.07 .08 .03 .04 .30 .04 .62 .16 .11 .18 .53 .08 .09 -.15 .50 .23 .09 -.04 .30 .48

.02 .19 .23

.00 .00 .00 .28 .00 .06 .00

.00

' ML only.

ROTTER SCALE ALTERNATIVES

1335

munalities 7 that were greater than .20. A still further improvement was obtained when the two-factor structure was allowed to be

oblique. However, correlation between the factors in this instance was .70, suggesting a maximum likelihood deduction closer to

Table 5 Likert Format Scale: Factor Loadings and Communalilies for Rotated Principal Axis (RPA) and Maximum Likelihood (ML) Solutions RPA Solution Communality
4 .16 .15 .11 .23 -.18 .43 -.17
.41

Collins's (1974) structure ML solution Factor no.

3 2
4

Item
No 2 3 4 5 6 7 9 10 11 12 13 15 16 17 18 20 21 22 23 25 26 28 29 a b

RPA Loads

Type"
E I I E I E I E E I E I
E 1 I

1 .49 -.12 -.01 -.05 .10 -.02 -.16 .31 .24 .06 .05 -.01 .25 -.38 -.10 .05 -.56 .32 -.07 .11 -.02
.44

3 .07 .06 -.39 .24 -.12 -.04 -.06 -.01 .14 -.10 -.03 .02 -.01 .00 -.06 .11 -.08 .06 -.69 .59 .00 .10 -.04 .07 .15 .00 .45 -.62 .05 -.04 .02 -.02 -.13 -.05 -.42 .52 .00 -.10 .11 .02 -.02 .06 -.05 -.02 .10 -.32

RPA .27 .19 .17 .13 .25 .23 .08 .27 .20 .25 .04 .16 .11 .16 .10 .18
.45 .21

ML .00 .10 .00 .00

.04

Load
-.05 .00 .14 -.11 .15 .19 -.46 .44 -.09 .06 .00 -.18 .01 .00 .03 .16 -.56 .66 -.59 .53 .00 .00 .38 .00 .55 -.12 .59 -.71 .46 -.74 .00 .00 -.65 .21 -.06 .01 .24 .35 .57 -.28 .11 .00 .00 .03 .00 .20

Communality
.00 .00 .02 .01 .02 .06 .21 .20 .01 .00 .00 .01 .00 .00 .00 .03 .31 .43 .34 .28 .00 .00 .14 .00 .30 .01 .35 .51 .21 .54 .00 .00 .43 .04 .00 .00 .06 .12 .32 .08 .01 .00 .00 .00 .00 .04

a b a b
a b a b a b a b a b a b a b

a b
a b a b a b a b a b a b a b a b a b a b a b a b

E I E I E I E I E E I E I E I E I E I I E E ] E 1 I E I E E I

-.02 .39 -.13 .14 .44 -.20 .14 .04 .06 .46 -.02 .40 .19 .10 .27 .04 .28 -.01 .10
.07

-.48
.30 .27

-.21 .10 -.08 .48 -.71 .00 -.08 .16 -.21 -.06 .07 .10 -.21 .13 -.65 -.04 .13 -.03 .13 -.05 -.02

.22 .10 .24 -.02 -.01 .32 .15 .19 .16 .09 .02 .37 .25 .37 .14 .01 -.10 .18 .01 .06 .42 .15 .44 -.02 .11 .17

.35 .16 .19 -.04 .10 -.07 -.11 .41 -.23 .32 -.09 .18 -.15 .34 -.04 .26 .36 -.21 .30 -.08 .36 .05 .35 .09 -.08 .10 -.04 .07 .30 -.02 .54 -.03 .19 -.01 -.20 .38 .20 .10

.50 .40 .07 .33 .29 .17 .23 .19 .32 .43 .38 .51 .12 .15 .11 .19 .20 .28 .11 .09 .33 .43 .21 .04 .24 .16 .06 .15

.06 .11 .22 .00 .05 .00 .00 .00 .02 .03 .08 .19 .24 .34 .29 .06 .25 .00 .00 .22 .06 .25 .33 .29 .33 .13 .00 .00 .04 .00 .00 .05 .00 .24 .11 .05 .00 .00 .00 .05 .00

4 2 1 2 1 1 2 2 1 3 2 1 3 1
4 4

3 1 3
4

4 1 3 2
2

4 4 1 2 1 3 2 2 1
4

E = external, I = internal.

1336

NEAL M. ASHKANASY

Rotter's (1966) original unidimensional concept. l.iken format. In Table 5, I give factor loadings and communalities obtained from the Likert data. Factor loadings obtained from the RPA analysis were generally similar to those obtained by Collins (1974), although, as was the case for the forced-choice data, the factors explained only a small proportion (22%) of total score variance, and 23 out of 46 communalities were less than .20. The general pattern of results obtained by restricted factor analysis for the Likert data followed that obtained with the forced-choice data. Derestricting communalities from the RPA solution resulted in a better fit, but lower item communalities (out of 46 communalities, 11 were greater than .20). Using Collins's (1974) original factor coefficients gave a slightly degraded fit, whereas the maximum likelihood solution based on Collins's factorial structure only gave an improved fit (out of 38 communalities, 14 were greater than .20). The maximum likelihood solution for an oblique structure, however, resulted in relatively strong factor intercorrelations (shown in Table 6). These correlations indicate that, as was the case for the forcedchoice data, the maximum likelihood oblique solution approximates Rotter's (1966) unidimensional ideal. In summary, then, the factorial analysis has indicated that the two-factor structure of the forced-choice format and the four-factor structure of the Likert format scale reported in the literature are consistent with the data from this study. However, the factorial structures accounted in each case for only some 20% of total score variance, and many items in both scales exhibited little communality with score totals. Finally, maximum likelihood estimation of an oblique structure, based on reported factorial structures, indicated that the factors were strongly intercorrelated. Discussion Correlation With Social Desirability The hypotheses with respect to correlation with Marlowe-Crowne social desirability scores were supported only in respect to the Likert version of Rotter's (1966) scale. However, it would seem that the tests on the

Table 6 Maximum Likelihood Estimate of Collins's (1974J Factorial Structure Intercorrelation

Factor Descriptor Predictable vs. chancy world Just vs. unjust world Politically responsive vs. unresponsive world Difficult vs. easy world
I

1
2 3

1.00

1.00
.45

1.00
.48

1.00

.71

.96

.42

1.00

Note. All correlations are significant at p < .01.

alternative scales were not directly comparable because of response differences between the true-false and Likert versions of the MarloweCrowne scale. In particular, variability on the true-false scale was very low, with marked bunching of scores about the middle score. On the evidence from this study that corresponding item pairs on the alternative locus of control scales are highly correlated, and evidence from other work (e.g., Ashkanasy, 1982; Collins, 1974) that shows that the scales measure essentially the same personality construct, it appears that the fault does indeed lie with the true-false format of the MarloweCrowne scale. On the other hand, it is also arguable that the Likert version of the Marlowe-Crowne scale may not give a valid social desirability score. The reasons for this is as follows: A typical item in the Marlowe-Crowne scale is "I never resent being asked to return a favor" (Item 25-T). A "true" or "false" response to this item is an absolute statement because of the inclusion of the word never in the item: At face value, a "true" response to this item must be rooted in social acquiescence. A 1-7 scaled response, however, immediately removes the element of absoluteness and clearly is more akin to a statement of self-esteem. Furthermore, this study has produced results that show that the direction of correlation between Marlowe-Crowne scores and individual item scores in the Likert scale version of Rotter's measure is a predictor of internalily versus externality. Given the

ROTTER SCALE ALTERNATIVES

1337

evidence that internality is positively valued (Stern & Manifold, 1977), one would expect that internal item scores would correlate positively with social desirability (i.e., social values) and vice versa. Against this background, then, it appears that the findings of this study, with respect to social desirability correlation, are equivocal. Clearly, there is a need for more definitive research, such as that undertaken by Stern & Manifold (1977), into the social value aspects of internality and externality. The only clear conclusion from the current study is that the Marlowe-Crowne social desirability scale is of doubtful usefulness in this area of research. Factorial Structure Analysis The specific hypotheses of this investigation were that Watson's (1981) two-factor structure would be confirmed for the forced-choice format of the Rotter scale, and the Collins's (1974) four-factor structure would be confirmed for the Likert scale version of the scale. The empirical results, though not disconfirming the hypothesized factorial structures, were equivocal about the usefulness of the factors. Although the factor structures underlying the respective scales appeared to be stable, they accounted for only about 20% of total score variance in each case. The data also indicated that the factors may not be orthogonal; maximum likelihood fitting of the data to published factorial structures was noticeably improved if the factors were permitted to be strongly oblique. One noticeable characteristic of the factor solutions obtained in this study was that many of the questionnaire items proved to have low levels of communality, particularly for maximum likelihood fitting procedures. Furthermore, many of these items exhibited score reversal (i.e., mean scores opposite to that expected). It seemed therefore to be a worthwhile exercise to attempt to refine the measures by deletion of these items. However, this process resulted in deletion of about 50% of items in each case, and an exploratory factor analysis of the refined scales indicated that the factors still accounted for only about 30% of (refined scale) variance. An additional test of correlation with Ashkanasy's (1982) self-attribution data indicated that the refined

scales produced little improvement over the full-scale results. On the basis of these tests, I concluded that refinement of the scales in the manner suggested is unwarranted. Thus one is left with a relatively stable and well-established underlying structure in the Rotter scale lying half hidden under a burden of irrelevant material. Presumably it is this underlying structure that is tapping the internal-external personality variable. The question that remains is whether this structure defines an essentially unidimensional construct. Several authors (e.g., Zuckerman & Gerbasi, 1977a, 1977b; Zuckerman et al., 1977) have argued strongly that one or another of the (orthogonal) factors should be omitted from the scale. However, the results of the maximum likelihood analysis undertaken in this experiment do not appear to support this conclusion. Instead, the high level of obliqueness between the factors seem to support (a) Rotter's (1966, 1975) stance that the measure is tapping a generalized construct and (b) Collins's (1974) observation that "there is evidence of a common theme of internal-external control of reinforcement running through the 46 alternatives" (p. 387). The reason for the apparent dimensional ambiguity in the Rotter (1966) scale is not clear. Collins (1974) and O'Brien and Kabanoff (1981) suggested that the scale may confuse situation-dependent perceptions with internalized beliefs. This appears to be particularly the case for externality: Several researchers have attempted to differentiate externals into (a) those who express their beliefs as a means of defense against external threat and (b) those who possess true external beliefs (Evans, 1980; Hochreich, 1974, 1975; Rotter, 1975). Evans (1980, 1982), for example, differentiated "defensive" and "congruent" externals on the basis of Marlowe-Crowne social desirability scores. He found that the two groups reacted differently to negative feedback (Evans, 1980), and that they placed different values on success in achievement situations (Evans, 1982). Ashkanasy (1982), however, found that similarly differentiated groups did not differ in self-attribution of causation for academic performance. Clearly, there is considerable scope for research in this respect if the construct validity of Rotter's internalexternal scale is to be further established.

1338

NEAL M. ASHKANASY

From an operational point of view, however, it appears that Rotter's scale, despite an overhead of irrelevant items, measures a useful and essentially unitary personality variable. Comparison Between the Scale Alternatives The results of this study indicate that subject responses to corresponding item pairs are the same. In view of these results and of findings by Ashkanasy (1982) and Collins (1974), it appears that the alternative scale formats are measuring the same personality construct. The data from this study also agree with other published results showing that the underlying factorial structure of each scale, although stable, accounts for only a minority of total score variance. In terms of these findings, therefore, there seems to be little to choose from among the alternatives. The data from this study, however, confirm Collins's (1974) assertion that item pairs in Rotter's forced-choice format are often not strictly comparable. Furthermore, the data support Collins's finding of frequent instances of score reversal. The latter phenomenon in particular is clearly not desirable in a forcedchoice format in which choices are ostensibly between internal and external alternatives. On this basis, and taking into account Ray's (1980b) findings and the greater variance achieved with the Likert format in the current study, it appears that the Likert scale format is the preferable alternative. A further advantage of the Likert scale version of Rotter's scale over the forcedchoice format is the potentially improved insight resulting from the underlying fourfactor structure. Several researchers (e.g.. Scaturo & Smalley, 1980; Zuckerman et al., 1977) have demonstrated that these dimensions provide a potential for additional explanatory power. The results of the current study indicate that these factors may be of only limited usefulness, but, nevertheless, having the additional information contained in the factors on hand could prove advantageous in certain situations. In conclusion, then, this study has failed to produce unequivocal evidence of a relation between internal-external control scores and Marlowe-Crowne social desirability scores.

This failing appears to result from inadequacies in the Marlowe-Crowne scale, and I suggest that future research should be based on more explicit measures of social values, such as in Stern and Manifold's (1977) study. The analysis of factorial structure, on the other hand, has confirmed the fit of published structural models to the data obtained in the instance of this study, and has indicated that the factorial structure, particularly that pertaining to the Likert version, is intrinsically stable. The respective factorial structures accounted in each case for only about 20% of total score variance, suggesting that the Rotter scale, despite the apparent stability of its factorial structure, is not a "clean" measure of the internal-external control of reinforcement construct. Nevertheless, maximum likelihood analysis indicates that factors in the underlying dimensions are strongly oblique, a conclusion that may be interpreted as supportive of Rotter's (1966) original unidimensional internal-external control construct. Finally, the results of this study indicate that although there is little to choose from among the alternative scale formats, the Likert version offers advantages in terms of item clarity and factor interpretability.

References
Abramowitz, S. L. (1973). Internal-external control and political activism: A test of the dimensionality of Rotter's internal-external scale. Journal oj Consulting and Clinical Psychology, 40, 196-201. Armstrong, J. S. (1967). Derivation of theory by means of factor analysis, or Tom Swift and his electric factor analysis machine. American Statistician, 21, \ 7-21. Ashkanasy, N. M. (1982). Internal and external validity of alternative versions of the Rotter locus of control scale. Unpublished manuscript (available from the Department of Psychology, University of Queensland, St. Lucia, Queensland 4067, Australia.). Barling, J. (1980). Multidimensional locus of control beliefs among English-speaking South African mothers. The Journal of Social Psychology, 111, 139-140. Barling, J., & Bolon, J. (1980). Multidimensional locus of control: The case of white South African students. The Journal of Social Psychology, 111, 295-296. Butler, M. C., & Burr, R. G. (1980). Utility of a multidimensional locus of control scale in predicting health and job-related outcomes in military environments. Psychological Reports, 47, 719-728. Collins, B. E. (1974). Four components of the Rotter internal-external scale: Belief in a difficult world, a just world, a predictable world, and a politically re-

ROTTER SCALE ALTERNATIVES sponsive world. Journal of Personality and Social Psychology, 29, 381-391. Cone, J. D. (1971). Locus of control and social desirability. Journal of Consulting and Clinical Psychology, 36, 449. Crowne, D. P., & Marlowe, D. (1960). A scale of social desirability independent of psychopathology. Journal of Counseling Psychology, 24, 349-354. Dufiy, P. J., Shiflett, S., & Downey, R. G. (1977). Locus of control: Dimensionality and predictability using Likert scales. Journal of Applied Psychology, 62, 214219. Evans, R. G. (1980). Reactions to threat by defensive and congruent internals and externals: A self-esteem analysis. Journal of Research in Personality. 14, 76^90. Evans, R. G. (1982). Skill versus chance tasks: Comparison of locus of control, defensive externality, and persistence. Personality and Social Psychology Bulletin, 8, 129133. Furnham, A., & Henry, J. (1980). Cross-cultural locus of control studies: Experiment and critique. Psychological Reports, 47, 23-29. Gurin, P., Gurin, G., Lao, R. C., & Beattie, M. (1969). Internal-external control in the motivational dynamics of Negro youth. Journal of Social Issues, 25(3), 29-53. Gurin, P., Gurin, G., & Morrison, B. M. (1978). Personal and ideological aspects of internal and external control. Social Psychology, 41, 275-296. Hochreich, D. J. (1974). Defensive externality and the attribution of responsibility. Journal of Personality, 42, 543-557. Hochreich, D. J. (1975). Defensive externality and blame following failure. Journal of Personality and Social Psychology, 32, 540-546. Joe, V. C. (1971). Review of the internal-external control construct as a personality variable. Psychological Reports. 28. 619-640. Joreskog, K. G. (1969). A general approach to confirmatory maximum likelihood factor analysis. Psychometrika, 34, 183-202. Joreskog, K. G. (1970). A general method for analysis of covariance structures. Biometrika, 57, 239-251. Kestenbaum, J. M. (1976). Social desirability and scale values of locus of control scale items. Journal of Personality Assessment, 40, 306-309. Kestenbaum, J. M., & Hammersla, J. (1976). Filler items and social desirability in Rotter's locus of control scale. Journal of Personality Assessment, 40, 162-168. Lange, R. V., & Tiggemann, M. (1981). Dimensionality and reliability of the Rotter I-E locus of control scale. Journal of Personality Assessment, 45, 398-406. Lao, R. C. (1970). Internal-external control and competent and innovative behavior among Negro college students. Journal of Personality and Social Psychology, 14, 263-270. Lao, R. C., Chuang, C. J., & Yang, K. A. (1977). Locus of control and Chinese college students. Journal of Cross-Cultural Psychology, 8, 299-313. Lefcourt, H. M. (1972). Recent developments in the study of locus of control. In B. A. Maher (Ed.),

1339

Progress in experimental personality research (pp. I 39). New York: Academic Press. Lefcourt, H. M. (1976). IJKUS of control: Current trends in theory and research. New York: Wiley. Lefcourt, H. M. (1980). Personality and locus of control. In J. Garber & M. E. P. Seligman (Eds.), Human helplessness: Theory and applications (pp. 245-259). New York: Academic Press. Lefcourt, H. M., von Baeyer, C. L., Ware, E. E., & Cox, D. J. (1979). The multidimensional-multiattributional causality scale: The development of a goal-specific locus of control scale. Canadian Journal of Behavioural Science, 11, 286-304. Levenson, H. (1972). Distinctions within the concept of internal-external control: Development of a new scale. Proceedings of the 80th Annual Convention of the American Psychological Association (pp. 261-262). Levenson, H. (1973). Multidimensional locus of control in psychiatric patients. Journal of Consulting and Clinical Psychology. 41, 397-404. Levenson, H. (1974). Activism and powerful others: Distinctions within the concept of internal-external control. Journal of Personality Assessment, 38, 377383. Miller, W. R., & Seligman. M. E. P. (1976). Learned helplessness, depression and the perception of reinforcement. Behaviour Research and Therapy, 14, 7-17. Mirels, H. L. (1970). Dimensions of internal versus external control. Journal of Consulting and Clinical Psychology, 34, 226-228. Mulaik, S. A. (1975). Confirmatory factor analysis. In D. J. Amick & H. J. Walberg (Eds.), Introductory multivariate analysis (pp. 170-207). Berkeley, CA: McCutchan Publishing. O'Brien, G. E., & Kabanoff, B. (1981). Australian norms and factor analysis of Rotter's internal-external control scale. Australian Psychologist, 16, 184-202. Phares, E. J. (1973). Locus of control: A personality determinant of behavior. Morristown, NJ: General Learning Press. Prociuk, T. J., Breen, L. J., & Lussier, R. J. (1976). Hopelessness, internal-external locus of control, and depression. Journal of Clinical Psychology, 32, 299300. Prociuk, T. J., & Lussier, R. J. (1975). Internal-external locus of control: An analysis and bibliography of two years of research (1973-1974). Psychological Reports, 37, 1323-1337. Ray, J. J. (1980a). Belief in luck and locus of control. The Journal of Social Psychology, 111, 299-300. Ray, J. J. (1980b). The comparative validity of Likert, projective, and forced-choice indices of achievement motivation. The Journal of Social Psychology, 111, 63-72. Rotter, J. B. (1954). Social learning and clinical psychology. Englewood Cliffs, NJ: Prentice-Hall. Rotter, J. B. (1966). Internal-external control of reinforcement. Psychological Monographs, 80(1, Whole No. 609). Rotter, J. B. (1975). Some problems and misconceptions related to the construct of internal versus external

1340

NEAL M. ASHKANASY Vuchinich. R. E., & Bass, B. A. (1974). Social desirability in Rotter's locus of control scale. Psychological Reports, 32, 1124-1126. Watson, J. F. (1981). A note on the dimensionality of the Rotter locus of control scale. Australian Journal of Psychology, 33. 319-330. Zuckerman, M., & Gerbasi, K. C. (1977a). Dimensions of the I-E scale and their relationship to other personality measures. Educational and Psychological Measurement. 37, 159-175. Zuckerman, M., & Gerbasi, K. C. (1977b). Belief in internal control or belief in a just world: The use and misuse of the I-E scale in prediction of attitudes. Journal of Personality, 45, 356-378. Zuckerman, M., Gerbasi, K. C., & Marion, S. P. (1977). Correlates of the just-world factor of Rotter's I-E scale. Educational and Psychological Measurement, 37, 375-

control of reinforcement. Journal of Consulting and Clinical Psychology, 43. 56-67. Rotter, J. B. (1979). Comments on Section IV: Individual differences and perceived control. In L. C. Perlmuter & R. A. Monty (Eds.). Choice and perceived control (pp. 263-269). Hillsdale, NJ: Erlbaum. Rotter, J. B., Chance J., & Phares, E. J. (Eds.) (1972). Applications of a nodal learning theory of personality. New York: Holl, Rinehart & Winston. Ryckman, R. M., Posen, C. E, & Kulberg, G. E. (1978). Locus of control among American and Rhodesian students. The Journal of Social Psvchologv, 104. 165173. Scaturo, D. J., & Smalley, N. S. (1980). IXKUS of control as a multidimensional correlate of political involvement. The Journal of Psychology, 105. 83-92. Stern, G. S., & Manifold, B. (1977). Internal locus of control as a value. Journal oj Research in Personality, 11. 237-242.

381.

Appendix Text of Scoring Locus of Control Items 2. (a) Many of the unhappy things in people's lives are partly due to had luck, (b) People's misfortunes result from the mistakes they make. 3. (a) One of the major reasons we have wars is because people don't take enough interest in politics, (b) There will always be wars, no matter how hard people try to prevent them. 4. (a) In the long run people get the respect they deserve in this world, (b) Unfortunately, an individual's worth often passes unrecognized no matter how hard he [or she] tries. 5. (a) The idea that teachers are unfair to students is nonsense, (b) Most students don't realize the extent to which their grades are influenced by accidental happenings. 6. (a) Without the right breaks one cannot be an effective leader, (b) Capable people who fail to become leaders have not taken advantage of their opportunities. 7. (a) No matter how hard you try some people just don't like you. (b) People who can't get others to like them don't understand how to get along with others. 9. (a) I have often found that what is going to happen will happen, (b) Trusting to fate has never turned out as well for me as making a decision to take a definite course of action. 10. (a) In the case of the well-prepared student, there is rarely if ever such a thing as an unfair test, (b) Many times exam questions tend to be so unrelated to course work that studying is really useless. 11. (a) Becoming a success is a matter of hard work: luck has little or nothing to do with it. (b) Getting a good job depends mainly on being in the right place at the right time. 12. (a) The average citizen can have an influence in government decisions, (b) This world is run by the few people in power, and there is not much the "little guy" can do about it. 13. (a) When 1 make plans, 1 am almost certain that I can make them work, (b) It is not always wise to plan too far ahead because many things turn out to be a matter of good or bad fortune anyhow. 15. (a) In my case, getting what I want has little or nothing to do with luck, (b) Many times we might just as well decide what to do by flipping a coin. 16. (a) Who gets to be boss often depends on who was lucky enough to be in the right place first, (b) Getting people to do the right thing depends upon ability; luck has little or nothing to do with it. 17. (a) As far as world affairs are concerned, most of us are the victims of forces we can neither understand nor control, (b) By taking an active part in political and social affairs, the people can control world events. 18. (a) Most people don't realize the extent to which their lives are controlled by accidental happenings, (b) There is really no such thing as "luck." 20. (a) It is hard to know whether or not a

ROTTER SCALE ALTERNATIVES person really likes you. (b) How many friends you have depends upon how nice a person you are. 21. (a) In the long run the bad things that happen to us are balanced by the good ones, (b) Most misfortunes are the result of lack of ability, ignorance, laziness, or all three. 22. (a) With enough effort we can wipe out political corruption, (b) It is difficult for people to have much control over the things politicians do in office. 23. (a) Sometimes I can't understand how teachers arrive at the grades they give, (b) There is a direct connection between how hard I study and the grades I get. 25. (a) Many times I feel that I have little influence over the things that happen to me. (b) It

1341

is impossible for me to believe that chance or luck plays an important role in my life. 26. (a) People are lonely because they don't try to be friendly, (b) There's not much use in trying to please people; if they like you, they like you. 28. (a) What happens to me is my own doing. (b) Sometimes I feel that I don't have enough control over the direction my life is taking. 29. (a) Most of the time I can't understand why politicians behave the way they do. (b) In the long run the people are responsible for bad government on a national as well as local level.

Received July 5, 1983 Revision received January 10, 1984

Instructions to Authors Authors should prepare manuscripts according to the Publication Manual of the American Psychological Association (3rd ed.). Articles not prepared according to the guidelines of the Manual will not be reviewed. All manuscripts must include an abstract of 100-150 words typed on a separate sheet of paper. Typing instructions (all copy must be double-spaced) and instructions on preparing tables, figures, references, metrics, and abstracts appear in the Manual. Also, all manuscripts are subject to editing for sexist language. APA policy prohibits an author from submitting the same manuscript for concurrent consideration by two or more journals. Also, authors of manuscripts submitted to APA journals are expected to have available their raw data throughout the editorial review process and for at least 5 years after the date of publication. For further information on content, authors should refer to the editorial in the March 1979 issue of this journal (Vol. 37, No. 3, pp. 468-469). The reference citation for any article in any JPSP section follows APA's standard reference style for journal articles; that is, authors, year of publication, article title, journal title, volume number, and page numbers. The citation does not include the section title. Anonymous reviews are optional, and authors who wish anonymous reviews must specifically request them when submitting their manuscripts. Each copy of a manuscript to be anonymously reviewed should include a separate title page with authors' names and affiliations, and these should not appear anywhere else on the manuscript. Footnotes that identify the authors should be typed on a separate page. Authors should make every effort to see that the manuscript itself contains no clues to their identities. Manuscripts should be submitted in quadruplicate (the original and three photocopies), and all copies should be clear, readable, and on paper of good quality. Authors should keep a copy of the manuscript to guard against loss. Mail manuscripts to the appropriate section editor. Editors' addresses appear on the inside front cover of the journal. Section editors reserve the right to redirect papers among themselves as appropriate unless an author specifically requests otherwise. Rejection by one section editor is considered rejection by all, therefore a manuscript rejected by one section editor should not be submitted to another.