have been reported in a test manual by Drenth and Kranendonk (1973). ...... and Introverted social shyness came from STUDY 3 in which the first higher-order ...
Adv. Behav. Rex Ther. Vol. 12, pp. 153-282, Printed
in Great Britain.
CORRELATES AND CLINICAL W. A. Arrindell,l* M. J. Pickersgill,
01~2/90 $0.00 + .50 1991 Pergamon Press pk.
OF ASSERTIVENESS IN NORMAL SAMPLES: A MULTIDIMENSIONAL APPROACH R. Sanderman,* W. J. J. M. Hagernan, M. G. T. Kwee,s H. T. Van der Molen,6 and M. M. Lingsma7
1 University of Groningen, Academic Hospital, Department of Clinical Psychology, Oostersingel 59, 9713 EZ Groningen, The Netherlands 2 University of Groningen, Department of Health Sciences, Antonius Deusinglaan 1, 9713 AV Groningen, The Netherlands 3 Vrije Universiteit, Faculty of Medicine, Department of Psychiatry at the Amsterdam Psychiatric Centre, Valeriusplein 9, 1075 BG Amsterdam, The Netherlands 4 University of London, Royal Holloway and Bedford New College, Psychology Department, Egham Hill, Egham, Surrey TW20 OEX, U.K. 5 Joris Psychiatric Centre, Department of Behavioural Psychotherapy, Sint Jorisweg 2, 2612 GA Delft, The Netherlands 6 University of Groningen, Psychologisch Instituut “Heymans”, Department of Personality and Educational Psychology, Grote Kruisstraat 2/l, 9712 TS Groningen, The Netherlands 7 Van Harte & Lingsma Efficiency BV - Communication Courses and Management Development, Gebouw ‘Statenhof, Reaal 5 G, 2353 TK Leiderdorp, The Netherlands Abstract-Recent reviews of the assertiveness literature (e.g., St. Lawrence, 1987) suggest that the majority of the most frequently used measures of self-reported assertion require more empirical attention regarding their psychometric properties before definitive conclusions can be made about their utility for research purposes and for clinical applications. In the present investigation, an attempt was made to expand the construct validity (convergent and divergent) of the Scale for Interpersonal Behavior (SIB), a multidimensional measure of both difficulty and distress in assertiveness. Findings were obtained from eight independent non-patient and clinical samples on a multitude of measures. Considering assertiveness as a subconstruct of the more complex Shyness construct, predictions were formulated as to the kind and degree of associations that ought to emerge in relating the SIB to a large variety of homologous and more or less non-homologous concepts. Among others, these included private and public self-consciousness, social and non-social (e.g., agoraphobic and blood-injury) fears, punitivity. trait shyness and trait social *To whom reprint requests should be addressed. *This monograph was written while the first author was affiliated to the Department of Psychiatry of the Vrije Universiteit, Amsterdam, The Netherlands. 153
anxiety. depression (affect and complaints), anger-hostility. self-esteem, dogmatism aspects, social cognition, aggression, interpersonal values (e.g., leadership, support, conformity) and state and trait anxiety. Statistical analyses involved determining associations from simple correlational and higher-order (factor) analyses and from a multidimensional scaling technique (MINISSA). The results converged in providing clear evidence of the convergent and divergent validity of the SIB measuring constructs. Sex differences in the patterns of assertiveness correlates were either small or negligible, the most outstanding exceptions being correlations involving self-esteem in community volunteers (stronger in females) and such interpersonal values as leadership (significant in males but not in females), support and recognition (both significant in females but not in males) in non-psychiatric social skills trainees. In addition, shyness was confirmed as a higher-order concept, broader than assertiveness, encompassing both its affective and its behavioral components. Different types of shyness relating to the affective, behavioral or cognitive components were identified, of which Neurotic/fearful social shyness and Shyness as an anxiety-behavioral syndrome were most prominent, thus providing further support for the idea that shyness is a fundamental aspect of social/interpersonal behavior, personality organization and structure. The different higher-order types of Shyness were orthogonal to General emotionality/Neuroticism/General psychological distress and also to what may be considered subcomponents of the broader Neuroticism/Anxiety concept (e.g., depressive mood and complaints, phobic and obsessive-compulsive symptoms). The implications which the empirical identification of different types of higher-order Shyness constructs has for assessment, treatment planning and treatment evaluation are explored.
INTRODUCTION Like the Assertion Inventory (Gambrill and Richey, 1975), the Scale for Interpersonal Behavior (SIB; Arrindell, De Groot, & Walburg, 1984; Arrindell & Van der Ende, 1985) is unique among self-report inventories in that it measures both discomfort (anxiety) associated with self-assertion in specific situations and the likelihood of engaging in a specific assertive behavior (avoidance or performance). That is, separate scales are used for the measurement of social anxiousness and of the behavior patterns that may express assertive responding. This distinction is one of relevance because there is not a strong relationship between subjective feelings of social distress and overt patterns of avoidant or inhibited behavior, nor any reason to expect one (Leary, 1983a; see also Leary, 1983b, p. 14; compare Marks & Mathews, 1979, p. 265). In fact, Leary (1983a) has noted that all of the self-report measures that are commonly employed in psychological research as measures of dispositional social anxiety share the common problem of confounding one aspect (anxiety) with the other (avoidance/performance). Notwithstanding the significant similarity in scale format between the SIB and the Assertion Inventory, there is at least one major difference between the two. Recent studies have shown that the Assertion Inventory is substantially confounded with social desirability response biases (Kiecolt & McGrath, 1979; Rock, 1981; McNamara & Delamater, 1984). However,
Correlates of Assertiveness
working from a traditional psychometric perspective, to date no evidence of a meaningful overlap of the SIB with social desirability has been found at the correlational level (Arrindell, De Groot, & Walburg, 1984). Beck and Heimberg (1983) have suggested that the majority of the most frequently used measures of self-reported assertion, including the Assertion Inventory, require more empirical attention regarding a number of areas (norms, reliability of measurement, different aspects of validity) before definitive conclusions can be made about the soundness of their psychometric base, and their utility for research purposes and for clinical applications. Swimmer and Ramanaiah (1985) have pointed out that despite the great deal of clinical and research interest in assertion training in recent years, an interest which has resulted in a proliferation of assertiveness inventories, there is a dearth of adequate research examining the construct validity of such measures. St. Lawrence (1987) has noted that this proliferation of instruments appeared and then far outstripped psychometric research when the need for measures exceeded the available assessment instruments: poorly validated instruments appeared to fill the void, established themselves through premature publication, and became perpetuated in the haste to evaluate treatment outcome. To explain this, St. Lawrence (1987) attempts to place behavioral assessment in general in a historical context by referring to Goldfried and Linehan (1977) who observe that: “History has an unfortunate way of repeating itself. Although there admittedly have been numerous conceptual and methodological advances over the past 30 years, behavioral assessment is not immune to many of the pitfalls that have been observed in the past. Once any measure appears in the literature, it becomes capable of developing its own momentum. If the procedure is clearly specified and easily administered, researchers and clinicians are likely to use it. At that point it becomes a “frequently used” assessment procedure, thereby justifying its utility by assessors in the future. It then only requires a factor analysis-and perhaps a short form-to provide it with a completely independent life of its own” (p. 16).
The SIB has good reliability and adequate construct validity (Arrindell, De Groot, & Walburg, 1984; Arrindell & Van der Ende, 1985; Arrindell, Sanderman, Van der Molen, Van der Ende, & Mersch, 1988), and has been proven to be a sensitive measure of change (e.g., Emmelkamp, Van der Hout, & De Vries, 1983; Van der Molen, 1985). However, there is a need to extend certain aspects of validity of the SIB further (see below). This will be the main aim of the present article, which is preceded by a specification of the different aspects that are tapped by the Scale. Measuring Assertion,
Aspects of the SIB
as the term is usually used in the research literature,
multidimensional construct which encompasses a relatively wide range of response classes (cf., e.g., Furnham & Henderson, 1983). Definitions often specify not only which types of responses are to be included under the definition of the term but also which are to be excluded. Two important aspects of the definition of assertion that were considered to be important in constructing the forerunner (i.e., the experimental version) of the SIB (De Groot & Walburg, 1977) and that are also represented in the present SIB are Negative and Positive Assertion. The definition of assertion presently used is a broad one in the sense that it refers to all those responses in interpersonal relations by means of which individuals assert themselves in a normal manner. By “normal” is meant that such individuals will respond without anxiety and without expressing aggression or anger with undue force; also they will be self-confident, will have the courage of their convictions, will defend their rights and interests, and will be able to express positive feelings, such as affiliative needs and affection for other persons (cf., De Moor & Orlemans, 1971). According to Linehan and Egan (1979), the two most frequently included response classes are self-expressiveness and standing up for one’s rights (see also Furnham & Henderson, 1983; Henderson & Furnham, 1983), both of which are captured by the SIB. The two response classes most typically excluded are the emotional responses of anxiety or general emotionality and hostility, two aspects that are not tapped by the SIB and which have consequences for the SIB’s theoretical relations with external measures. Specifically, the SIB measures the following factorially-derived constructs of assertion: (1) Display of negative feelings (negative assertion): requesting change in another person’s irritating behavior; standing up for one’s rights in a public situation (defense of rights and interests); behavior that calls for exercise of initiative to resolve problems and to satisfy needs; ability to refuse requests. (2) Expression of and dealing with personal limitations: admitting ignorance about a topic; recognition of one’s failure or shortcomings; ability to deal with criticism and pressure; requesting help and attention. (3) Initiating assertiveness: social assertiveness in the sense of Lorr and More (1980); expressing one’s own opinion. (4) Praising others and the ability to deal with complimentslpraise of others (expression of positive feelings or positive assertion-in the sense of
Henderson & Furnham, 1983): giving and receiving praise or compliments; display of feelings. In addition to these subscales, a fifth scale termed General Assertiveness can be employed as an indication of a person’s level of assertiveness across
various situations and various types of assertive behavior. The General Scale is scored for both distress and performance (see Appendix 1 for a listing of the SIB items). CONSTRUCT
OF THE SIB
To date, attempts at establishing the construct validity of the SIB have mainly dealt with the investigation of (1) the convergent validity of SIB scales as compared with those of Social Fears, Social Inadequacy and Shyness; (2) the differentiation (that is, divergent or discriminant validity) of SIB scales from aggression (and other aspects of Extrapunitivity), social desirability and (certain components of) anxiety or general emotionality; (3) the factorial validity of the SIB (that is, the generalizability of its measuring constructs across distinct samples of subjects). While the bulk of the available findings could give the impression that an extensive and therefore sufficient body of evidence in support of the construct validity of the SIB has been obtained, such is not actually the case and the affective, behavioral and cognitive correlates of the SIB Anxiety and Performance constructs need to be determined. Thus, correlations of the SIB with such theoretically-relevant measures as selfattention (which includes public and private self-consciousness), negative self-evaluation (self-concept) and self-reported behavioral indicators of social and interpersonal skill have yet to be examined and such correlations could further expand its construct validity. Establishment of construct validity is essential if an instrument is to be interpreted as a measure of the construct which it purports to measure (e.g., Cronbach & Meehl, 1955). In fact, Messick (1975) has argued that construct validation should be viewed as the evidential basis for inferring a measure’s meaning. However, construct validation is the most difficult and demanding (that is, complex) form of validity to achieve. To declare that such an achievement has been accomplished, a judgment is required which is based on many kinds of information: procedures followed in developing the test, results of experiments testing specific implications of the construct, and
patterns of correlations with other measures (e.g., Guion, 1974, p. 289). Derogatis and Cleary (1977) note that the programmatic requirements for construct validation create the impression of something of a psychometric decathlon. They claim: “Investigations and exercises that demonstrate various aspects of domain sampling, convergent and discriminant correlations. multi-method trait assessments. and a spectrum of predictive, criterion-oriented evaluations must all be accomplished to demonstrate that, in fact. measurement operations are indicative of the underlying construct purported to exist. In addition. empirical findings derived from such procedures must coincide with the nomological
W.A. network construct”
of relationships deducible from whatever (Derogatis & Cleary. 1977. p. 981).
The present study attempts to examine the convergent and discriminant validity of the SIB measuring constructs by studying validational data obtained from eight independent non-patient and clinical samples. The relevance of such an endeavor lies in the observation of, among others, Swimmer and Ramanaiah (1985) and St. Lawrence (1987) that, at present, there is a great need for the assessment of convergent and divergent validity of several measures of assertiveness within the seff-report mode that are widely used in both research and clinical work. Such information is essential for integrating the results of various studies based on different assertiveness inventories (cf., Swimmer & Ramanaiah, 1985). The procedure for evaluating convergent and discriminant validity that was followed here corresponds with the traditional one in which correlations between the scale in question and at least two other scales that differ in their conceptual similarity to the first are examined (e.g., Campbell & Fiske, 1959). A larger correlation with a scale assessing the same or a conceptually similar construct rather than with a scale assessing a conceptually unrelated or less closely related construct provides evidence of convergent and discriminant validity (cf., Smith, Pope, Rhodewalt, & Poulton, 1989, for a practical application). Unique to the methodology employed in the present study is that multidimensional scaling (MDS) techniques are used to systematize data by representing the similarities between the variables/constructs spatially, as in a map. In addition, MDS procedures are utilized in addition to exploratory factor analyses (provided that sample size is sufficiently large) and, where possible, the results obtained with both methods are compared descriptively. METHOD Samples
Eight studies were conducted independently of each other for the purpose of validating the SIB and other instruments not further specified here. These investigations included five samples of university students who volunteered to participate independent of any course requirements, one sample of community subjects (Ss) who responded to notices put in newspapers, one sample predominantly of junior and senior executives who participated in optional courses for increasing specific areas of social skills (i.e., non-psychiatric social skills trainees), and an unselected sample of psychiatric inpatients whose diagnoses were based on consensus between a clinical psychologist and a psychiatrist who followed DSM-III guidelines.
The Ns for each sample are shown in Table 1, as are demographic data describing each in terms of sex and age. The trainees completed their questionnaires prior to the initiation of the training, while the patients completed theirs prior to hospitalization. Instruments1
The following measures were used in one or more of the studies: (1) The Scale for Interpersonal Behavior: Different measuring aspects of the Scale were outlined above. It suffices to mention that the measure contains 50 items (Appendix l), 46 of which are classified (in a non-overlapping fashion) into the four categories of assertive responding referred to earlier. The respondent evaluates each item on two separate 5-point (l-5) Likert-type scales, one for discomfort (to the left of the item), and the other for the probability of engaging in a specific assertive behavior (to the right of the item). Reliability and validity data pertaining to this measure have been summarized in Arrindell and Van der Ende (1985). (2) The Fear Questionnaire (FQ; Marks & Mathews, 1979) yields scores on three categories of clinical phobias: Agoraphobic Fears, Social Fears and Blood-Injury Fears. In addition, the questionnaire also contains an Anxiety-Depression dimension. Data supporting the reliability and validity of the Dutch version of this measure are given in Arrindell, Emmelkamp, and Van der Ende (1984). (3) The Self-C onsciousness Scale (SCS; Fenigstein, Scheier, & Buss, 1975) measures two different aspects of self-consciousness, which is defined as the degree to which an individual tends to select himself/herself as an object of his/her own attention. Private Self-Consciousness is the process of attending to one’s inner thoughts and feelings. Public Self-Consciousness refers to a general awareness of oneself as a social object that may have an effect on others. A third subscale, Social Anxiety, is defined by ‘Due to the low frequency of endorsement and to the poor variance of specific scale items, not all subscales of particular multiscale measures could be included in the analyses presented below since they revealed too low internal consistency reliabilities (