A scale to measure locus of control of behaviour

British Journal of Medical Psychology (1984), 57, 173-180 01984 The British Psychological Society Printed in Great Britain 173 A scale to measure locus of control of behaviour A. R. Craig, J. A. Franklin and Gavin Andrews Many behaviour, psychotherapy and healthy life-style programmes require subjects to take responsibility for the control of the old unwanted behaviours or to be responsible for maintaining new desired behaviours after therapy has ended. A scale to measure the locus of control of behaviour would be valuable if it could predict persons likely to relapse following apparently successful therapy. A 17-item Likert-type scale to measure this construct was developed and shown to have satisfactory internal reliability, to be test-retest reliable in the absence of treatment, to be independent of age, sex and social desirability, and to distinguish clinical disorder from normal non-clinical subjects. Furthermore, change towards internality (a reduced LCB score) during therapy was shown to predict maintenance or, alternatively, change towards externality (an increase or no change in the LCB score) was shown to predict relapse 10 months later in treated stutterers. The scale was shown to be related to the personal control factor of the Rotter I-E scale but to be more powerful a predictor of relapse than this personal control subscale or the full Rotter scale. Individual adherence to therapy is a factor in the success or failure of most treatment plans. We believe it is particularly important in some behaviour therapy programmes and may be critical to the success of healthy life-style programmes which aim to foster weight and diet change, smoking cessation and increased exercise. Some individuals do accept responsibility for their own continued well-being and thus maintain improvement, but others, particularly those who do not accept this responsibility, may be at risk of relapse. There is a need for a measure which would identify such persons so that further steps can be taken to inhibit the relapse before it occurs. This paper describes the development of a scale designed for this purpose. Locus-of-control scales measure the extent to which a person perceives events as being a consequence of his or her own behaviour and therefore potentially under personal control (Lefcourt, 1976). The most widely known scale was developed by Rotter (1966) to measure generalized expectancies for internal vs. external control of reinforcement. In this I-E locus-of-control scale, if the subject attributed the relation to luck or powerful others then the belief was labelled external control. Conversely, if the relation was attributed to personal effort then the belief was labelled internal control. Thus Rotter s scale is potentially capable of predicting relapse. Both Mirels (1970) and Nassi & Abramowitz (1980) showed that Rotter s I-E scale consists of two factors, belief in control over one s own personal behaviour and belief in control over political institutions and world affairs. The personal control factor is based on the responses to 10 forced-choice items and these items seemed relevant to our present purpose. However, the internal reliability of a scale is a function of the number of test items, with at least 20 items being needed to produce an acceptable coefficient alpha (Nunnally, 1967). As the Rotter I-E scale has a coefficient alpha of approximately 0-7 (Rotter, 1966), we estimated that the internal reliability of the 10-item subscale would only be 0.47. On inspection of the 10 personal control items three have reference to perceptions of control over education, a theme unrelated to relapse following therapy. For these reasons we believed the 10 personal control items would not be suitable to measure the locus of control of behaviour. Rotter (1975) and Lefcourt (1976, 1981) emphasized the need for the development of specific locus-of-control scales to measure special aspects of the construct and, indeed, various scales have been developed which measure aspects of the perceived control of

174 A. R Craig, J. A. Franklin and Gavin Andrews personal behaviour. Levenson (1973) developed a Likert-type 24-item multidimensional locus-of-control scale comprising three separate subscales. These were defined as internal, powerful others and chance. Unfortunately, the internal dimension is too short for use in this paper for only four of the eight internal items load substantially on the internal factor. Reid & Ware (1974) developed a 45-item forced-choice I-E scale, which consisted of three main subscales : fatalism, socal-political influence and self-control of impulses. The fatalism subscale concerned general perceptions about luck whilst the self-control subscale concerned perceptions about control of emotions and impulses. Neither seemed appropriate to predict those likely to relapse after therapy. Kirscht (1972) developed a locus-of-control scale measuring perceived control over health. This concept was developed by Wallston et al. (1976) and Wallston et al. (1978) who described a locus-of-control scale to measure perceptions of control over catching disease and remaining healthy. These scales measure a concept different from the present concern. The absence of a suitable instrument to measure locus of personal control over a behavioural problem prompted the development of the Locus of Control of Behaviour (LCB) scale which was designed to measure the extent to which subjects perceive responsibility for their personal problem behaviour. Development of the LCB scale Item generation Twenty items for the LCB were generated to be consistent with the seven non-education items identified in Mirels factor 1. Items were worded in both first and third person. A six-point bipolar Likert-type scale was chosen in order to enhance sensitivity to change. An initial sample of 50 male and 50 female undergraduate university students completed the scale by indicating their degree of agreement with each item. In order to minimize the possible effect of social desirability in responding, half the items were reversed in terms of internality and the scale instructions emphasized that there are no right or wrong answers and that a large number of people agree or disagree with each statement. Item analysis Three items in the data from the 100 students had low correlations (< 0-2) with the total score with that item removed on the data of the 100 subjects above. These three items also showed little response variability as shown by the small standard deviations and were therefore rejected. The remaining 17 items all had item-total correlations between 0.23 and 0.61, and standard deviations greater than one. The 17 items are shown in Table 1. The coefficient alpha for these 17 items was 0.79, which according to Nunnally (1967) is evidence of acceptable internal reliability. Scoring The 17-item test is scored in the same direction as the Rotter I-E scale, that is, high scores indicate externality. Thus, as may be seen from Table 1, the 10 items which relate to externality are tallied from the left-hand column of response boxes and the scores for the seven items relating to internality (items 1, 5, 7, 8, 13 and 16) are transposed so that 5 is scored as 0 (strongly disagree), 4 (generally agree) becomes 1 (generally disagree), etc., in the right-hand column of response boxes. After transposing the seven items the test is scored by summing the scores for all 17 items. Factor analysis Construct validation depends upon all items loading on one common factor (Nunnally, 1967). The 17-item LCB scale was completed by an additional 23 college students and 53

A locus of control of behaviour scale 175 nurses so that the sample of 176 would exceed the 10 subjects per item criterion suggested by Nunnally (1967). The scores were subjected to a principal components factor analysis without rotation. All items loaded greater than 0.3 on the first factor, which accounted for 24 per cent of the variance. On inspection of the factor loadings it was evident that there is a single cluster with no outliers. This then can be considered a general factor representing belief about control over personal behaviour. A scree test for the number of substantive factors (Cattell, 1966) was applied and confirmed the importance of the first factor. The second factor, which accounted for 11 per cent of the variance, has no clear meaning and is probably an artifact of the wording of the items. The item loadings for factors 1 and 2 are shown in Table 1. In order to confirm the factorial structure of the LCB in a clinical population, 1 14 agoraphobic patients completed the 17-item scale. The results were similar to the original factor analysis, with factor 1 accounting for 19 per cent of the variance. The scree test again confirmed that only the first factor was likely to be of interest. Table 1. The locus of control of behaviour scale, showing items and item loadings for the two principal factors (substantial loadings are > 0-30; n = 176) LCB scale Directions: Below are a number of statements about how various topics affect your personal beliefs. There are no right or wrong answers. For every item there are a large number of people who agree and disagree. Could you please put in the appropriate bracket the choice you believe to be true? Answer all the questions. 2. A great deal of what happens to me is probably just a matter of chance....( ) 0.48 0.51 3. Everyone knows that luck or chance determines one's future....( ) 0.37 0.65 4. 1 can control my problem@) only if I have outside support....-( ) 0.47-0.11 5. When I make plans, I am almost certain that I can make them work...( ) 0.49-0.24 6. My problem(s) will dominate me all my life.....( ) 0.50-0.02 7. My mistakes and problems are my responsibility to deal with...( ) 0.30-0.23 8. Becoming a success is a matter of hard work, luck has little or nothing to do 9. 10. 11. 12. 0 1 2 3 4 5 I I I I I I Strongly Generally Somewhat Somewhat Generally Strongly disagree disagree disagree agree agree agree Factor 1 Factor 2 1. I can anticipate difficulties and take action to avoid them... ( ) 0.35-0.29 withit....( My life is controlied by outside actions and events....(... People are victims of circumstance beyond their control.( To continually manage my problems I need professional help...( When I am under stress, the tightness in my muscles is due to things outside mycontrol....( 13. I believe a person can really be the master of his fate...( 14. It is impossible to control my irregular and fast breathing when I am having difficulties.....( 15. I understand why my problem(s) varies so much from one occasion to the next....( 16. I am confident of being able to deal successfully with future problems...( 17. In my case maintaining control over my problem(s) is due mostly to luck ( Score =... +... Sum score =... ) 0.34 ) 059 ) 0.63 ) 0.42 ) 0.60 ) 0.33 ) 0.53 ) 0.43 ) 0-63 ) 0.66 0.42 0.06 0.25 0.04-0.15 0.01-0.30-0.52-0.41 0.37 7 UPS 57

176 A. R Craig, J. A. Franklin and Gavin Andrews Reliabili ty and validity A one-week test-retest reliability score was obtained from the responses of 25 non-clinic adult subjects. The Pearson correlation was 0.90. A six-month test-retest reliability score was obtained from the responses of 25 adult stutterers awaiting treatment. This correlation was 0.73. These results indicate the stability of the personal control construct over time in subjects not receiving treatment. As the LCB is a new test the construct validity has to be inferred from a number of sources. Firstly, as Rotter s I-E scale has been shown to possess construct validity (Rotter, 1966) and the LCB scale is intended to measure the perception of control over personal behaviour, the results should be comparable. Substantial correlations (r = 0-67 for males, r = 0.66 for females) were found to exist between the scores of 123 university students on the LCB and the Rotter I-E scale. Secondly, as the LCB is specifically concerned with locus of personal control, it should discriminate between the Rotter personal and political control items. A high correlation with Mirels factor 1 (perceived control over personal event items) and a lower correlation with Mirels factor 2 (perceived control over political event items) was expected. On the above 123 students the correlations were 0.70 and 0.31 for males (n = 52) and 0.67 and 0.37 for females (n = 71) between the LCB and Mirels factors 1 and 2 respectively. Thirdly, responses to the LCB should not reflect social desirability. The correlation between the LCB and a shortened version of the Marlowe-Crowne Social Desirability Scale (Crowne & Marlowe, 1960; Strahan & Gerbasi, 1972) completed by the above 25 stutterers was low (r = 0.2). Fourthly, LCB scores should be independent of age and sex. For the 123 students, mean age 22 years and 42 per cent male, age and sex were not related to LCB scores (r = -0.25 and 0.04 respectively). For the 25 non-clinic subjects, mean age 31 and 36 per cent male, age and sex were again not related (r = -0.26 and 0.27 respectively). For 70 stutterers, mean age 29 and 81 per cent male, age and sex were also independent of LCB scores (r = -0.12 and 0-09 respectively). For 69 agoraphobics, mean age 38 and 16 per cent male, age and sex were also independent of LCB scores (r = 0.09 and 0.2 respectively). Lastly, a scale which measures locus of control over behaviour should differentiate between persons with and without chronic conditions, for the former group would have learnt through experience that this aspect of their behaviour was not under their personal control. It was predicted that both chronic stutterers and chronic agoraphobics would score higher on the LCB (i.e. be more external) than a normal non-clinic population. Sample 1 consisted of the 123 university students mentioned above, and their mean LCB score was 28.3. Sample 2 consisted of 53 nurses working in a general hospital, mean age 23 years. Their mean LCB score was 27-9. Sample 3 consisted of 70 stutterers mentioned above who had been stuttering for more than 10 years and who were waiting for behavioural treatment for stuttering. The mean LCB score was 31.0. Sample 4 consisted of 69 patients mentioned above waiting for behavioural treatment for agoraphobia, and whose mean duration of illness was eight years. The mean LCB score was 39-4. Table 2 shows the means, standard deviations, skewness of distribution and the 95 per cent confidence intervals for the four samples described above. Differences between the sample means were tested using planned orthogonal contrasts (Keppel, 1973). The clinical samples scored significantly higher than the normal sample scores (t = 6.4, d.f. = 31 1, P < 0.001). The normal samples were not significantly different (t = 0.28, d.f. = 31 I, P = 0-77). The neurotic sample (agoraphobics) scored significantly higher than the non-neurotic sample (stutterers) (t = 5.3, d.f. = 3 11, P -= 0.001). On a separate Student s t test the stutterers were also significantly different from the normal samples (t = -2.2, d.f. = 31 I, P = 0.03).

A locus of control of behaviour scale 177 Table 2. Statistics for the four normative samples 95 per cent confidence Skewness of interval Sample Population type Number Mean SD distribution for means 1 University students 123 28.3 8.5 0.01 26.8 to 29.8 2 Nurses 53 27.9 8.1 0.50 25.6 to 30.1 3 Stutterers 70 31.0 9.6-0.29 28.7 to 33.3 4 Agoraphobics 69 39.4 11.2-0.11 36.7 to 42.1 Table 3. Two-by-two contingency tables showing frequency of stutterers who relapsed or maintained 10 months after therapy vs. change in attributions of perceived control during treatment for the LCB, Mirels 10 and the Rotter scale Mirels F1 Rotter I-E LCB Maint. Relapse Maint. Relapse Maint. Relapse Reduced score 13 5 18 7 28 2 No change or increased score 19 8 14 6 4 11 xz = 0.02, d.f. = 1 for Mirels F1. xa = 0.03, d.f. = 1 for Rotter I-E. x2 = 21-7, d.f. = 1, P < 0.01 for LCB. Predictive validity The LCB should be capable of predicting relapse. We now report a comparison of the power of the Rotter I-E scale, the Mirels factor 1 10-item scale and the LCB scale to predict persons who will eventually relapse. Stutterers treated in a behaviour therapy programme need to continue using the skills taught in that programme if they are to remain fluent. It was hypothesized that it is better for the stutterer to regard the improvement during treatment as due to personal effort, therefore enhancing internal expectancies regarding maintaining control over stuttering. A new sample of 36 male and nine female adult stutterers received treatment in a three-week intensive behaviour therapy programme that eliminated stuttering in the short term (Howie et al., 1981). All subjects completed the Rotter I-E and the LCB scales before, at the end, and 10 months after treatment. Frequency of stuttering was measured in percentage syllables stuttered PASS) on the same occasions. By the end of therapy all subjects were basically fluent in that the mean frequency of syllables stuttered had fallen from the pre-treatment mean of 12.3 per cent to a post-treatment group mean of 0-1 per cent syllables stuttered. Relapse was arbitrarily defined as greater than 2 per cent syllables stuttered at the 10-month assessment. Whilst we hypothesized that there would be some relation between pre-treatment locus of control and eventual relapse, some preliminary work (Craig 8z Howie, 1982) had shown that the change in locus of control towards internality was associated with a better long-term outcome. A phi coefficient for dichotomous data was computed between increase in internality during treatment (defined 7-2

178 A. R Craig, J. A. Franklin and Gavin Andrews as at least a 5 per cent reduction in the Rotter, Mirels factor 1 10-item and the LCB scales) and relapse 10 months later (defined as greater than 2 per cent syllables stuttered). The results were as follows: the change in LCB predicted relapse (r = 0.69, 95 per cent confidence interval 0.5 < r < 0.82), whilst neither the change in Mirels factor 1 10-item scale (r = 0.02) nor the change in the Rotter scale (r = 0.03) predicted relapse. Converting these to Fisher z scores the correlations were significantly different (P < 0.01). These differences are displayed in a contingency table (Table 3) which shows the number of stutterers who maintained fluency skills or relapsed in relation to change in attributions of perceived control during treatment. To examine further the extent of change of locus of control of behaviour the group was broken into maintainers (n = 32) and relapsers (n = 13) on the basis of the 2 per cent syllables stuttered criterion at follow-up. The pre- and post-treatment LCB scores for the maintainers were a mean 31 (SD = 8.7) and 23.7 (SD = 9.3). The pre- and post-treatment LCB scores for the relapsers were a mean 26 (SD = 10.6) and 26.7 (SD = 9-6). The mean extent of change for maintainers was 30 per cent in the internal direction (decreasing LCB scores) whilst for the relapsers the mean change was 18 per cent in the external direction. Table 4. Frequency of stuttering and LCB scores for six relapsed patients who received further therapy ~~ After first After remedial Six months relapse treatment later %SS LCB %SS LCB %SS Subject 1 4.3 16 0 14 0 Subject 2 4.7 37 0 34 1.6 Subject 3 8.2 11 0 4 0 Subject 4 6.7 15 0 1 1.3 Subject 5 5.2 21 0 22 4.8 Subject 6 14.2 48 0 40 11.7 Note. Subjects 1 4 reduced their LCB score during treatment and six months later had maintained treatment skills; subject 5 did not reduce her LCB score and had not improved six months later. Because the LCB change during treatment was strongly correlated with subsequent relapse we tested the utility of the scale as an aid to treatment. Preliminary research was conducted on six of the stutterers who had relapsed at the 10-month point and who also had not reduced their LCB score during therapy. After the 10-month assessment these six were given a one-week treatment programme designed once again to reduce stuttering as well as to enhance internality - meaning a reduction of the LCB score. Treatment consisted of a short version of the three-week intensive treatment followed by a self-control therapy programme which consisted of speech skill monitoring, evaluation of accuracy of monitoring plus anxiety management. Table 4 shows that by the end of therapy all subjects had once again become stutter-free, and while five had reduced their LCB score, one had not. Six months later one of the five who had become more internal was stuttering above the 2 per cent criterion level while the subject whose LCB score did not reduce had relapsed yet again.

A locus of control of behaviour scale 179 Discussion The new 17-item Locus of Control of Behaviour scale was shown to have satisfactory internal reliability, not to be related to sex, age or social desirability and to be stable over time in the absence of treatment. It was also shown to have construct validity, correlating substantially with Rotter s I-E general expectancy scale (r = 0.67). More importantly, the LCB was shown to discriminate between Mirels personal and political subscales of the Rotter I-E scale (r = 0.70 and 0.31 respectively). The normative data presented in this paper illustrate that the scale distinguishes on the LCB construct between persons with and without chronic conditions. Clinical samples scored higher, thus demonstrating greater externality than non-clinic population. On the above assumptions, the clinician would hope to see, as a result of treatment, not only a reduction of the disorder, but also a reduction of the LCB score towards the internal dimension. Such a reduction would imply that patients perception of personal control over their own behaviour had been enhanced. Indeed, lack of change or increase in the LCB scale was shown to be a significantly better predictor of those clients who did relapse 10 months after behaviour therapy than did change in either the Rotter I-E scale or in Mirels factor 1 personal control items. Identifying those likely to relapse is a major problem in treatment. In stuttering, whilst treatment outcome, as might be expected, is strongly related to the type and duration of treatment, within a specific treatment programme it appears to be independent of severity of stuttering, attitudes to speaking, age, sex, social status and even neuroticism (Andrews et al., 1980). Change in the perception of personal control as measured by the present LCB scale is the first strong indicator of those persons who are at risk of relapse. This change is differentially interpreted according to the direction of movement. If a reduction occurs, then expectancies of success are enhanced reducing risk of performance deficit in the long term. If no change or an increase towards the external dimension occurs during treatment then the person has not perceived performance to be a consequence of self-effort and a performance deficit is predicted in the long term. A preliminary study indicates that information on lack of increased internality will enable us to provide extra and specialized treatment to those whose perceptions of control do not internalize. There are several threats to validity. The choice of 5 per cent reduction in the LCB, Rotter and Mirels factor 1 scale as a criterion of change or no change in perceived control should perhaps be discussed. This level was set arbitrarily and it could be argued that a better procedure would be to use the standard error of measurement (Nunnally, 1967). If one does base decisions on this (four-point change), then five subjects no longer scored as becoming more internal. The sense of the data did not alter and the chi-square analysis was still significant Q2 = 10-8, d.f. = 1, P -= 0.01). We feel that the 5 per cent criterion is reasonable, especially as percentile change is a better indicator when scores on the LCB are at the low end of the scale. This work needs replication in other groups of stutterers and generalization to other groups of patients before the utility of the scale will be established. Such work is in progress. In the mean time it appears that changes in perception of personal control over behaviour as measured by the LCB may prove valuable in predicting patients ability to maintain control over problems which they once perceived as chronic and uncontrollable. Being able to identify those likely to relapse means that scarce health service resources can be allocated according to need, with those persons at greatest risk of relapse receiving more and perhaps different treatment.

180 A. R Craig, J. A. Franklin and Gavin Andrews References Andrews, G., Guitar, B. & Howie, P. (1980). Meta-analysis of the effects of stuttering treatment. Journal of Speech and Hearing Disorders, 45, 287-307. Cattell, R. B. (1966). The scree test for the number of factors. Multivariate Behavioral Research, 1, 245-276. Craig, A. R. & Howie, P. M. (1982). Locus of control and maintenance of behaviour therapy skills. British Journal of Clinical Psychology, 21,67-68. Crowne. D. P. & Marlow, D. (1960). A new scale of social desirability independent of psychopathology. Journal of Consulting Psychology, 24, 349-354. Howie, P. M., Tanner, S. & Andrews, G. (1981). Short- and long-term outcome in an intensive treatment program for adult stutterers. Journal of Speech and Hearing Disorders, 46, 104-109. Keppel, G. (1973). Design and Analysis: A Researcher s Handbook. New York: Prentice-Hall. Kirscht, J. P. (1972). Perceptions of control and health beliefs. Canadian Journal of Behavioral Science, 4, 225-237. Lefcourt, H. M. (1976). Locus of Control Current Trends: Theory and Research. New York: Wiley. Lefcourt, H. M. (ed.) (1981). In Research with the Locus of Control Construct: vol. 1. Assessment Methods. New York: Academic Press. Levenson, H. (1973). Multidimensional locus of control in psychiatric patients. Journal of Consulting and Clinical Psychology, 41, 397404. Mirels, H. L. (1970). Dimensions of internal versus external control. Journal of Consulting and Clinical Psychology, 34, 226-228. Nassi, A. J. & Abramowitz, S. I. (1980). Discriminant validity of Mirels personal and political factors on Rotter s I-E scale: Does a decade make a difference? Journal of Personality Assessment, 44, 363-367. Nunnally, J. C. (1967). Psychometric Theory. New York: McGraw-Hill. Reid, D. W. & Ware, E. E. (1974). Multidimensionality of internal versus external control: Addition of a third dimension and non-distinction of self versus others. Canadian Journal of Behavioral Science, 6, 131-142. Rotter, J. B. (1966). Generalised expectancies for internal versus external control of reinforcement. Psychological Monographs, 80, 1-28. Rotter, J. B. (1975). Some problems and misconceptions related to the construct of internal vs. external control of reinforcement. Journal of Consulting and Clinical Psychology, 43, 5-7. Strahan, R. & Gerbasi, K. C. (1972). Short, homogeneous versions of the Marlowe-Crowne Social Desirability Scale. Journal of Clinical Psychology, 28, 191-193. Wallston, B. S., Wallston, K. A., Kaplan, G. D. & Maides, S. A. (1976). Development and validation of the Health Locus of Control (HLC) Scale. Journal of Consulting and Clinical Psychology, 44, 58&585. Wallston, K. A., Wallston, B. S. & DeVellis, R. (1978). Development of the Multidimensional Health Locus of Control (MHLC) Scales. Health Education Monographs, 6, 160-1 70. Received 14 March 1983; revised version received 16 June 1983 Requests for reprints should be addresed to Mr Ashley R. Craig, Division of Communication Disorders, Department of Psychiatry, Prince Henry Hospital, University of NSW Teaching Hospital, Little Bay, NSW, Australia 2036.