Person.
Pergamon
A CONTENT, ANALYTIC
CORRELATIONAL STUDY
OF FOUR
AMBIGUITY
of Psychology,
University
College
AND
FACTOR
TOLERANCE
OF
QUESTIONNAIRES
ADRIAN
Department
D!l/. Vol. 16, No. 3, pp. 403.410. 1994 CopyrIghtf; 1994 ElsevierScienceLtd Printedin Great Britain.All rightsreserved 0191-8869194 $6.00 + 0.00
indwid.
FURNHAM
London,
Gower
Street.
London
WCIE
6BT, England
(Received 7 June 1993) Summary-A review of the scattered and diffuse tolerance of ambiguity (AT) literature from 1946 revealed 4 fairly frequently used self-report measures of this concept (AT). This study set out to compare and contrast these different measures by content, correlational and factor analysis. Nearly 250 people completed all four dimensions. Each scale was then subject to content and factor analyses which agreed with each other, as did correlations with the various factors. First- and higher-order factor analysis yielded a number of interpretable factors. The results are discussed in terms of the multi-dimensional nature of AT beliefs. the psychometric equivalence of these various scales, and the difficulty of measuring the concept in the first place.
INTRODUCTION
The concept of ambiguity tolerance or it’s many synonyms has attracted research from various branches of psychology for over 40 years (Frenkel-Brunswick, 1948). It has been conceived as a personality variable (Budner, 1962) as well as a property of both organizations (Furnham & Gunter, 1993) and cultures (Hofstede, 1980). Tolerance for ambiguity (AT) refers to the way an individual (or group) perceives and processes information about ambiguous situations when they are confronted by an array of unfamiliar, complex or incongruent cues. AT is a variable which is often conceived on an unidimensional scale: the person with low tolerance of ambiguity supposedly experiences stress, reacts prematurely and avoids ambiguous stimuli. At the other extreme of the scale, however, a person with high tolerance for ambiguity perceives ambiguous situations/stimuli as desirable, challenging and interesting and neither denies nor distorts their complexity or incongruity. One of the first references to the AT was published in the American Psychologist in 1948. In a long paper the following year Frenkel-Brunswick (1949) set out her model or theory for AT as an “emotional and perceptual personality variable”. She was clearly influenced by the work of Jaensch (1938) whose early psychophysical and perceptual work was to influence many others (Eysenck, 1954). In her paper she described emotional and perceptual correlates of AT and memory based studies. She concluded: “In the present paper, an attempt was made to discuss denial of emotional ambivalence and intolerance of cognitive ambiguity as but different aspects of what may be a fairly coherent characteristic. An underlying emotional conflict between glorification and hostility in the attitude towards parents, sex and one’s own social identity previously demonstrated in children inclined toward rigid social dichotomizing as revealed by ethnic prejudice is taken as the impetus for experiments in memory, perception, and related topics, devised to test tolerance of ambiguity on an emotionally more neutral ground. There is some indication of a prevalence of premature reduction of ambiguous cognitive patterns to certainty in the prejudiced subjects, as revealed by a clinging to the familiar, or by a superimposition of one or many distorting cliches upon stimuli which are more manageable in a more simple and stereotyped fashion. There is some indication that in the case of distinct intolerance of emotional ambivalence one may as a rule be able to locate at least some aspects of intolerance of cognitive ambiguity although these may often by more apparent on a higher level than that of perception paper.” (P. 140) Frenkel-Brunswick (195 1) set out many behavioural dispositions of AT including resistance to reversal of apparent fluctuating stimuli, the early selection and maintenance of one solution in a 403
404
ADRIANFURNHAM
perceptually ambiguous situation, inability to allow for the possibility of good and bad traits in the same person, acceptance of attitude statements representing a rigid, black-white view of life, seeking for certainty, a rigid dichotomizing into fixed categories, premature closure, and remaining closed except to familiar characteristics of stimuli. Indeed this very long list may have attracted researchers to perform studies in the area, as it appears AT is a powerful predictive variable. Most researchers in the AT field based their definition of intolerance of ambiguity upon Frenkel-Brunswick’s (1949, 1951) work, whose own definition of the concept was generated by case study material gleaned from interviews of persons high or low on this variable. She described the characteristics persons at either extreme of the continuum might exhibit. She argued that AT generalizes to the entire emotional and cognitive functioning of the individual, characterizing cognitive style, belief and attitude systems, interpersonal and social functioning and problem solving behaviour. AT also generalizes to the sense modalities, in particular to perception so that the person intolerant of ambiguity in the emotional and cognitive sphere would exhibit similiar characteristics in his perceptual behaviour. She also related AT to other personality variables, predicting a positive relationship with the authoritarian family of personality traits. Since then the topic has attracted considerable research and remains a well-used variable to this day (Anderson & Schwartz, 1992). Over this 40 year period there have been a steady number of published papers using AT as an independent (and very occasionally as a dependent) variable. As is common in the field of personality and individual differences there exist a number of measures all purportedly measuring the same construct. Given that the AT construct has been around for 50 years it is no surprise that a number exist in this area. At about the same time other, unpublished, tests were available (Saunders, 1955) or one’s where little psychometric work was attempted (Eysenck, 1954). The scale developed was a simple 14-item true-false test but appeared to have good concurrent validity. More interestingly perhaps Eysenck reported on a non-verbal pictorial intolerance of ambiguity test which was 8 drawings of a dog turning slowly and by degrees into a cat. The predicted and confirmed hypothesis was that rigid people would continue to cling to the original ‘dog’ concept long after it had turned into a cat. Most of the tests of tolerance of ambiguity are however self-report questionnaires. One of the best known, and well used scales in this area was developed 30 years ago by Budner (1962) who devised a 16-item (half positive, half negative) scale which was described in great detail. His argument, which was derived from a close reading and analysis of the previous literature, suggested that each item had to tap at least one postulated indicator of perceived threat, namely phenomenological submissions or denial, operative submission or denial. Furthermore, items referred to one of either of three types of ambiguous situations: novelty, complexity and insolubility. The scale was validated on 17 different, mainly student, populations and shown to be free of acquiescent and social desirability response tendencies. Although the test correlation was good (0.85 over 2 months) the internal alpha was poor (0.49). Various forms of validity were examined including concurrent and construct. The scale was correlated with rankings of individuals on the basis of short biographies, peer ratings, and measures of conventionality, belief in divine power, attendence at religious services, dogmatism about religious beliefs and attitudes to censorship. The total scale also correlated positively with authoritarianism, idealism of and submission to parents, machiavellianism, career choice in medical students etc. Not all the correlations were significant and most were in the 0.20 to 0.40 range but they were sufficiently consistent to suggest that the measure had content, concurrent and construct validity. Budner (1962) ends on a rather curious note: Given that intolerance
of ambiguity refers to a covert activity--evaluation-and a non-specific goal, and is therefore, an abstraction of many responses to many situations, the varied correlates reported in this study can be viewed not as consequences but as manifestations of intolerance of ambiguity. Being intolerant of ambiguity does not lead an individual to favour such things as censorship; rather favouring censorship (in most situations) is part of being intolerant of ambiguity” (p. 49)
In other words he sees AT as a “non-specific” trait that does not lead to specific behaviours or evaluations that are not manifestations of AT itself. The Budner scale has perhaps attracted most attention and is used most frequently in AT research.
Four
AT questionnaires
405
Rydell and Rosen (1966) and Rydell (1966) reported on the development and validation of another scale. The scale consisted of 16 true-false items which had been constructed on a “u-priori basis” (Rydell & Rosen, 1966, pp. 15CLl51) with relatively limited validation. Test-retest reliabilities over a month with 41 students yielded an r = 0.71 and with 105 students over 2 months r = 0.57 but there was no evidence of the test’s internal reliability. The test was in part validated with the use of semantic differential ratings of contradictory and non-contradictory adjective-noun concept combinations (Rydell, 1966). MacDonald (1970) however, attempted some psychometric evaluation of the Rydell-Rosen scale but added 4 extra items. This larger scale had a test-retest reliability of 0.63 over 6 months and was cross-validated on nearly 800 undergraduates. The test was correlated with Rokeach’s dogmatism scale, the Gough-Sanford Rigidity scale and church attendence but not social desirability. The split-half reliability was also satisfactory at 0.73 MacDonald noted “shows promise of being a useful instrument for the measurement and further investigation of ambiguity tolerance” (p. 797). In the mid 1970s Norton (1975) argued that the extant paper-and-pencil measures of AT were “flawed by low internal reliability and the absence of adequate validity evidence.” (p. 607) This he believed was in part due to ambiguities on the term ambiguous which was used in 8 different ways incompleteness, fragmentation; as a probability; to describe: multiple meanings; vagueness, unstructured; lack of information; uncertainty; inconsistencies, contradiction, contraries and unclearness. He therefore developed a 50-item ‘measure’ (MAT-50) which was tested 7 times to develop high reliability which was r = 0.38 (Kuder-Richardson) and with a test-retest reliability of 0.86 after IO-12 weeks. He also attempted to determine 3 types of validity: content validity (through content analysis and faking studies), criterion-related (through correlations with measures of dogmatism and rigidity) and construct validity (through measures of willingness to volunteer for an ambiguous study; aesthetic judgement; a content analysis of verbal behaviour and behavioural dramatization). As predicted, high AT Ss tended to volunteer more for undefined experiments, to use different aesthetic judgements and be more dramatic in problem-solving groups. Norton (1975) ends his paper suggesting 7 research questions the scale may be used to investigate most of which had been tested before but not when using a self-report AT measure: what is the cognitive process during an ambiguous situation; how can the information theorist account for ambiguity; is it possible to use the measure of AT to help identify therapeutic problems; do groups prefer leaders who are AT; how is trusting behaviour related to AT: to what degree is AT culture bound; what is the function of AT in a conflict resolution situation. More recently Bhusha and Amal (1986) working in India developed a 40-item situational test of AT. It presents people with ambiguous everyday situations and looks at their cognitive affective, avoidance and aggressive responses. Along with content validity the concurrent validity was established by correlations with other measures of AT, rigidity and fascism. A few attempts have been made to assess, compare-and-contrast and analyse the AT measures. Kirton (1981) argued that the literature shows “that the concept has become over extended and its elements remain unsupported by the confused data collected and scales with inadequate psychometric performance have been used” (p. 407). He have over 600 Ss the Budner (1962) and MacDonald (1970) scale along with other measures of dogmatism, inflexibility, conservatism and hostility. Using an item analysis (discarding all items that failed to distribute 8&20%) Kirton reduced the MacDonald scale to 11 items and the Budner scale to 7. All measures were significantly correlated mainly over r = 0.40, but although the correlation between the two AT measures was high (r = 0.56) it was not significantly greater than many of the other correlations in the matrix. Indeed, if treated as 5 individual items of a single scale their internal reliability was 0.86! Kirton believed his study confirmed three primary elements to AT, dogmatism, inflexibility and conservatism. According to Kirton (1981) measures should reinforce and illuminate “parent” theories by demonstrating evidence of containing the elememts defined by the theories and also by demonstrating the expected relationship with correlates adduced from the theory. To a large extent the correlational and experimental findings using the AT concept are dependent on the psychometric robustness and sensitivity of the measures of AT. This study examined the various questionnaire measures frequently used in research. Two sorts of analysis were undertaken. First a content analysis of the items used in the various questionnaires which attempts to ascertain the oblique factors that may be present in their questionnaires. Most, but not all questionnaires,
406
ADRIAN FURNHAM
are undimensional but it seems from empirical and theoretical work that AT is a multi-dimensional concept and is measured as such. Secondly a factor analysis (first and higher-order) of the scales themselves was undertaken to ascertain the orthogonal factors underlying these measures. METHOD
Subjects In all 243 Ss took part in this study of which 168 were female. They were all part of two universities S panels. They ranged in age from 19 to 71 though the mean was 25.74 years (SD 8.90). About half completed secondary schooling and the remainder had some post-work qualification. Three quarters were single and the remainder married, divorced or separated. Politically in all 10% voted Conservative, 22% Social Democrat and 28% Labour which the remainder voted for minor parties or did not vote. Questionnaires The four scales mentioned in the Introduction were used namely Walk’s A Scale (O’Connor, 1952); Budner’s scale (Budner, 1962); Rydell’s scale (Rydell & Rosen, 1966) and Norton’s multi-dimensional scale (Norton, 1975). Four or 5 items were identical in the different scales so replications were omitted. In all Ss completed a lOl-item questionnaire which they completed on a 7-point agree-disagree scale. Procedure Ss completed the questionnaire in their own time. There possible they were debriefed after the study.
was an 87% response
rate. Where
RESULTS Con tent analysis
Two Ss completed the content analysis. Both were given all 101 items and told to classify them according to any criteria they saw fit. This Q sort methodology led to very similiar conclusions between the two analysers. First it was agreed that 19 were positive statements about ambiguity and 82 negatitle. Of the positive items 10 concerned “neutral objective statements” many taken from about self” mainly taken from the Norton the Budner (1961) scale while 9 were “statements questionnaire. The negative items which reflected a dislike of ambiguity (and constituted nearly 80% of the items) were divided into 3 sections: neutral objective statements (14 items); pathological anxiety statements (18 clients) and items about self (30 items). There seemed to be 5 major factors accounting for nearly all the items. Correlationul
and,factor
analysis
Table 1 shows the 4 scales alpha co-efficient of reliability and their intercorrelations. The Norton and Rosen scale (being the longest) appeared to have the best, quite satisfactory, internal reliability. The two shorter scales had reliabilities just under 0.60 which is regarded by many as the barely Table I. Internal
reliability
No. of items Norton (1975) Walk (O’Connor. 1952) Rydell & Rosen (1966) MacDonald (1970) Budner (1962)
of the four scales Reversed
item
Alpha
69 8
7 3
0.89 0.58
20 I6
5 8
0.78 0.59
Intercorrelations
Norton (N) Walk (W) Rydell & Rosen (R) Budner (8)
N
W
R
0.54 0.82 0.47
0.62 0.44
0.57
Four AT Table 2. Factor
analvsis
questionnaires
407
of the Walk (O’Connor.
1952) scale Factors
Item There is more than one right way to do anything. It is alway better to have a definite course of action than to be vacillating The best leaders give specific enough instructions so that those under them have nothing to worry about. A smart person gets his life into routine so that he is not always being bothered by petty details. Nobody can have feelings of love and hate towards the same person. It is better to keep on with the present method of doing things than to take away that which might lead to chaos. A man can be well informed even if there are many subjects upon which he does not have a definite opimon. It is better to take a chance on being a failure than to let your life get into a rut. Eigenvalue Vai-laIlCe
I
2
3
0.00
0.64
0.04
0.70
0.04
0.77
0.80 -0.06
0.38
-0.08 0.10 2.14 26.8%
-0.12
-0.12
0.16
0.07
0.19
0.02
0.88
0.15
0.65
0.67
0.01
0.67 I.41 17.7%
0.07 I .02 12.9%
acceptable minimum for short scales. However there is some debate as to the necessity for high alphas (Boyle, 1991). The intercorrelations between the four measures were all positive and significant. The highest was 0.82 (which accounted for two-thirds of the variance) while the lowest was 0.44 (which accounted for about a fifth of the variance). This is a fairly typical correlation matrix between questionnaire measures of the same concept (Furnham, 1987, 1990; Furnham & Henderson, 1984). Because the Norton scale had an 8-fold factor structure it was not subject to factor analysis but the other 3 (shorter) measures were treated to a VARIMAX orthogonal rotation. All items were reversed so as given an in tolerance of ambiguity score. Table 2 shows the 3 factors, accounting for nearly 60% variance, that emerged from the analysis of the Walk scale. The first factor is about wise actions, the second about adopting alternative responses and the third about rather conservative points of view. Table 3 shows the 6 factors that emerged from the VARIMAX analysis of the Rosen scale. In all 6 factors emerged also accounting for about 60% of the variance. The first factor had items loading upon it that concerned problem solving and this accounted for nearly a quarter of the variance. The second factor has items loading on it which concerned the anxiety induced by ambiguous stimuli and the third factor the desire to complete or finish a problem. The fourth factor had items loading on it which refered to adventurousness, and the fifth factor uncertainty seeking, while the final factor was concerned with problem fragmentation. Table 4 shows the VARIMAX factor analytic results of the Budner scale. Four factors emerged which accounted for about half of the common variance. The first, bi-polar factor, which accounted for a fifth of the variance concerned issues about predictability. The second factor had 5 items loading on it all concerning variety and originality, the third clarity and the fourth regularity. Following the individual first level factor analysis, a higher order factor analysis was performed on the 21 factor scores derived from the 4 measures (see Table 5). The first factor, which accounted for nearly a third of the variance had 6 of the 8 Norton and 2 Rydell subscales loading on it. The second factor had 1 subscale from each of the 4 scales while the third had 3 subscales from the Rosen measure. The fourth and fifth factors, both of which accounted for only about 5% of the variance, each had 1 subscale factor from the Walk and 1 from the Budner measure. DISCUSSION The proliferation of measures purporting to measure the same psychological construct poses various problems for researchers. The first is which measure to choose and on what basis to make the choice e.g. psychometric, historical. The second is the overlap or intercorrelation between the
408
ADRIAN FURNHAM Table 3. Factor analvsis of the Rvdell and Rosen (1966) scale Factors
Items A problem has little attraction for me if I don’t think it has a solution. I amjust a little uncomfortable with people unless I feel that I can understand their behaviour. There’s a right way and a wrong way to do almost everything. I would rather bet I to 6 on a long shot than 3 to 1 on a probable winner. The way to understand complex problems is to be concerned with their larger aspects instead of breaking them into smaller pieces. I get pretty anxious when I’m in a social situation over which I have no control. Practically every problem has a solution. It bothers me when J am unable to follow another person’s train of thought. J have always felt that there is a clear differences between right and wrong. It bothers me when I don’t know how other people react to me. Nothing gets accomplished in this world unless you stick to some basic rules. Vague and impressionistic pictures really have httle appeal for me. Before an examination. I feel much less anxious if I know how many questions there wll be. Sometimes I rather enjoy going against the rules and doing things I’m not supposed to do. I like to fool around with new ideas, even If they turn out later to be a total waste of time. Perfect balance is the essence of all good compostion. If I were a doctor. I would prefer the uncertainties of a psychiatrist lo the clear and definite work of someone like a surgeon or X-ray specialist. If I were a scientist, I might become frustrated because my work would never be completed (science wll always make new discoveries). I don’t like to work on a problem unless there IS a possibility of coming out with a clear-cut and unambiguous answer. The best part of working a jigsaw puzzle LSputting in that last piece. Eigenvalue Variance
I
2
3
4
5
6
0.39
0.21
0.54
0.09
0.21
0.09
0.24 0.67
0.73 0.14
0.27 0.08
-0.07 0.00
0.07 0.23
0.05 0.14
-0.15
0.60
0.19
0.2 I
-0.04
0.06
0.05
0.79
0.06 -0.24
0.04 -0.03
0.04 0.00
-0.14 -0.12
0.07 -0.13
0.26 0.69
-0.65 0.06
0.09
-0.86
0.29 0.00 -0.16
0.04
0.00 -0.05 0.08
-0.17 0.02 0.08
0.05 0.39
0.04 0.46
0.69 0.06
0.07 0.77 -
0.20 0.17
0.19 0.10
0.66 0.38
0.28 0.08
-0.04 0.09
0.20 -0.03
0.06
0.13
0.12
0.20
0.21
0.1 I
0.01
0.76
0.05
0.00 0.43
- 0.05 0.12
0.17 0.14
0.75 0.09
0.18 0.15
-0.10 -0.40
0.00
-0.03
- 0.06
0.13
0.78
-0.20
0.33
-0.06
-0.09 0.42 0.1 I 4.78 23.9%
-0.44
0.23
0.67 ~~
-0.03
0.25
0.36
9.2%
7.7%
~0.00 0.04 1.33 6.7%
0.20 0.02
0.43 -0.20 I.15 5.8%
0.00 0.00 I.10 5.5%
Table 4. Factor analysis of the Budner (1962) scale Factors Jtems An expert who doesn’t come up with a definite answer probably doesn’t know too much. There is really no such thing as a problem that can’t be solved. A good job is one where what is to be done and how it is to be done are always clear. In the long run it is possible to get more done by tackling small. simple problems rather than larger and complicated ones. What we are used to is always preferable to what is unfamiliar. A person who leads an even, regular life in which few surprises or unexpected happenmgs arise, really has a lot to be grateful for. J like parties where I know most of the people more than ones where all or most of the people are complete strangers. The sooner we all acquire similar values and Ideas the better. I would like to live in a foreign country for a while. People who fit their lives to schedules probably miss most of the joy of living. It is more fun to tackle a complicated problem than to solve a simple one Often the most interesting and stimulating people are those who don’t mind being different and original. People who insist upon a yes or no answer just don’t know how complicated things really are. Many of OUTmost important dewions are based upon insufficient informatmn. Teachers or supervisors who hand out vague assignments give a chance for one to show initiative and originality. A good teacher is one who makes you wonder about your way of looking at things. Eigenvalue Variance
I
2
3
4
0.73 0.09
0.20 -0.01
0.08 0.44
0.27 0.48
0.46
0.00
0.57
0.00
0.05 m
0.0 I 0.01
0.75 0.44
-0.07 -0.14
0.25
-0.14
m
-0.11
0.20 -0.64 -0.57
-0.06 -0.17 0.05
0.52 -0.30 0.15
0.00 ~ 0.08 0.45
-0.04 -0.03
-0.03 0.32
-0.06 0.20
0.72 _0.48
-0.27
-0.19
0.20
0.48
-0.02
0.67
0.00
-0.04
0.63
0.09
-0.12
-0.02
0.56
0.05
-0.23
0.61 2.15 13.5%
-0.21 I .33 8.3%
0.27 3.24 20.3%
0.1 I
0.03 I.14 7.2%
Four
AT questionnaires
Table 5. Hieber order factor analvsis of the 21 factors resultiw
409 from analvsis of the four scales Factors
I
NI Philosophy N2 Interpersonal communication N3 Public image N4 Job-related N5 Problem solving N6 Social N7 Habit N8 Art-form WI Wise actions W2 Alternative responses W3 Conservatism Bl Predictability B2 Variety and originality B3 Clarity B4 Regularity RI Problem solvability R2 Ambiguity anxiety R3 Completer-finisher R4 Adventurousness RS Uncertainty seeking R6 Problem-fragmentation Eigenvalue Variance
0.09 -0.43 -0.75 0.66 2 0.66 0.68 0.14 0.24 0.13 0.09 0.34 -0.22 0.31 -0.09 0.19 0.83 -0.59 0.05 0.21 0.01 6.61 31.5%
2 0.80 0.36 0.05 0.10 0.47 0.42 0.19 0.21 0.58 0.11 0.14 0.45 -0.15 0.62 0.16 E 0.25 0.13 0.19 0.00 0.24 2.33 11.1%
3 0.10 -0.30 -0.05 0.16 0.09 - 0.08 0.19 0.00 0.16 0.03 0.15 -0.01 0.37 0.10 -0.64 0.02 0.04 0.08 -0.77 0.55 0.72 I.55 7.4%
4 0.1 I -0.05 0.07 0.32 0.36 0.05 0.25 0.66 0.14 0.15 0.64 0.52 0.27 0.20 0.33 0.11 - 0.02 0.48 0.06 0.31 0.13 I .26 6.0%
5 0.17 0.07 0.20 -0.33 -0.04 - 0.05 -0.13 0.10 -0.27 0.79 0.15 0.20 (@ -0.06 -0.15 0.05 0.14 -0.15 0.15 -0.17 -0.01 I .09 5.2%
various measures because of the correlations not being high (e.g. 0.80 plus) one has to account for at least half of the variance by some other factor. Thirdly when doing a review, or even more importantly a meta-analysis, it is particularly difficult to compare and contrast the results of studies if they have used different scales. The pattern of intercorrelations in this research (r = 0.44 to 0.82) are fairly similar to other “housekeeping” studies that examined as this study did, through content, correlational and factor the relationship between various questionnaires all supposedly measuring the same thing (Furnham, 1987,199O; Furnham & Henderson, 1984). The four questionnaires examined in this study differed on a number of criteria. First two appeared to have acceptable internal reliability while two did not. Whilst there has been some debate over the necessity of high internal reliability (Boyle, 1991) these results do help explain the correlational findings. The highest intercorrelation is between the most internally reliable scale, the lowest between the most unreliable scales. Indeed, the size of even the biggest correlation suggests at least 30% of the variance is not accounted for. Perhaps the factor analytic results were more interesting and important. They showed firstly that each of the three supposedly unidimensional scales in fact have a clear factor structure which is more than simply the distinction between tolerance and intolerance of ambiguity. In fact the 20-item Rydell and Rosen (1966) scale had 6 distinguishable factors. Further a higher order factor analysis of the subscales for all four questionnaires (21 in all) suggested a clear multidimensional structure. Four of the 6 factors were concerned with intolerance of ambiguity factors related to preferences, others to anxieties and yet others to philosophic, epistemological preferences. The fact that different studies have used different measures makes a review of the tolerance of ambiguity literature complicated because if there is evidence of non- or poor relication of experimental findings using different and clearly non-equivalent measures of the AT, one cannot be sure whether the findings are not robust, whether the scales are indeed not measuring the same things, or whether the measures are marked by construct irrelevancies and psychometric deficiencies. Convergent, discriminant and nomological validity are related issues in construct validation, but they are nevertheless unique. By contrast, the replication of a finding using different AT measures is a testament to their robustness. Psychologists’ major contribution to the AT literature has been to provide psychometricallysound measures of the AT to be used in empirical research. Although these measures are themselves open to fairly serious psychometric criticism, particularly with respect to the inadequate, indeed, poor validity data, that have been used successfully in a number of research programs.
410
ADRIAN FURNHAM REFERENCES
Anderson, S. & Schwartz, A. (1992). Intolerance of ambiguity and depression. Socicrl Cognirion, 10. 271 298. Bhusha, L. & Amal, S. (1986). A situational test of intolerance ambiguity. fs~chologia, 179, 254 261. Boyle, G. (1991). Does item homogeneity indicate internal consistency or item redundancy in psychometric Personality
and Individual
Differences,
Budner, S. (1962). Intolerance of ambiguity as a personality variable. Journal of Personalirq: 30, 29 59. Eysenck, H. (1954). The psychology of politics. London: Routledge. Frenkel-Brunswick, E. (1948). Intolerance of ambiguity as an emotional and perceptual personality variable. Personality,
scales?
12, 29 l-294.
Journtrl of
18, 108-123.
E. (1949). Tolerance toward ambiguity as a personality variable. Anirrican Ps~chologisr, 3. 268. E. (1951). Personality theory and perception. In Blake. R. & Ramsey. E. (Eds), Perception: An upproach to personality. New York: Ronald. Furnham, A. (1987). A content and correlations analysis of seven locus of control scales. Currenl P.sycho/ogicu/ Resecrrch Frenkel-Brunswick, Frenkel-Brunswick,
and Revie++ls, 6, 244-255.
Furnham, A. (1990). A content, correlational and factor analytic study of seven questionnaire measures of the Protestant Work Ethic. Human Relations, 43, 383-399. Furnham, A. & Gunter, B. (1993). Corporate assessment. London: Routledge. Furnham, A. & Henderson, M. (1984). Assessing assertiveness: A content and correlational analysis of live assertiveness inventors. Behavioural Assessment. 6, 244-255. Hofstede, G. (1980). Cultures’ consequences. California: Sage. Jaensch, E. (1938). Der gegenryppus. Leipzig: Barth. Kirton, M. (1981). A reanalysis of two scales of tolerance of ambiguity. Journal of Personcrlity Asxx~ment. 4S, 407 415. MacDonald, A. (1970). Revised scale for ambiguity tolerance: Reliability and validity. P.sj~chokogical Reports. 26, 791-798. Norton. R. (1975). Measurement of ambiguity tolerance. Journal of‘ Personuli~y Asses~menr. 39, 607 619. O’Connor. P. (1952). Ethnocentrism, “intolerance of ambiguity” and abstract reasoning ability. Journal of Ahnormal cmd Social
Psychology,
47, 526-530.
Rydell, S. (1966). Tolerance of ambiguity and semantic differential ratings. Psvchologicul Report.\. IY, 1303 13 12. Rydell, S. & Rosen, E. (1966). Measurement and some correlates of need cognition. P.sycho/ogical Reports, 19, 139- 165. Saunders, D. (1955). Some preliminary interpretive material for the PRI Research Memorandum SS. Educational Testing Service. October.