You are on page 1of 16

n2e Sport Psychologist, 1990, 4, 285-300

Use of the Test of Attentional and Interpersonal Style (TAIS) in Sport


Robert M. Nideffer Enhanced Performance Systems
Over the past 15 years the Test of Attentional and Interpersonal Style (TAIS) has become increasingly popular in the sport psychology area. More recently, investigatorsfactor-analyzing the six TAIS attentional scales and the information processing scale have raised serious questions about the independence of these measures. Specifically, they have suggested that the six attentional scales measured by the TAIS can be collapsed into two scales, one reflecting scanning (BET, BIT,.INFP) and one reflecting the focusing of attention (NAR, OET, OR). All of the studies reported on can be shown to have methodor o m their analyses. logical flaws and to have drawn inappropriateconclusions f Evidence is provided in the paper demonstratingthe independence of the TAIS scales. Suggestions are made for avoiding the methodological and interpretive problems that have permeated the literature.

In an article that first appeared in 1978, William Morgan reviewed the status of research on personality assessment in sport psychology (Morgan, 1980). At the time of his article, Morgan divided psychological ~ewarchers and practitioners in the area of sport psychology into two camps: The credulous camp whose followers believed that psychological tests could be shown to have a great deal of utility in the prediction of performance, and the skeptical camp whose followers believed, as Brent Rushall so eloquently stated, personality is not a significant predictor of sport performance (Rushall, 1973). Morgan in his paper, and Kane (1980) as well, concluded that research in the area of personality and sport had not provided coherent, unequivocal findings to rely on for predictive purposes. That is, psychological tests did not seem to be differentiating between athletes as a function of skill level within a sport. These authors argued, however, that there was a great deal of descriptive research that differentiated between athletes and nonathletes, and that some of the more recent research using more complex correlational analyses began to differentiate between athletes in different sports (Kroll & Crenshaw, 1970; Schurr, Ashley, & Joy, 1977; Singer, 1969).
Robert M. Nideffer is President of Enhanced Performance Systems, 822 Boylston St., Box 67197, Chestnut H i l l ,MA 02167.

286

Nideffer

It was pointed out, by Morgan and others, that a large number of research issues had clouded literature in the sport personality area. Investigators had failed to control for response bias; they had failed to adequately define both dependent and independent variables. As an example, the term "elite athlete" was often used as a descriptive term for performers who ranged from world champions to age group competitors at state and national levels. Many of the studies tended to be atheoretical, taking a shotgun approach to the assessment of personality. Tests were chosen without apparent logic or reason, simply because they existed. Often these instruments had little theoretical relevance to the population being tested (Cox, 1985). In response to all of these criticisms, it was suggested that investigators find ways to control for response bias (Morgan, 1980), that they attempt to look at both state and trait characteristics andlor situation-person interactions within the same studies, and that inventories be developed that are more "sport specific" in their content (Kroll, 1976; Martens, 1976). Kane suggests the current efforts to develop an interactional model of personality emphasizing the cognitive interpretations of the person in a given situation deserve special attention if only to help clarify the nature of the current objections to the traditional trait model.

Development of the TAIS


The Test of Attentional and Interpersonal Style (TATS) was developed in response to many of the criticisms outlined above (Nideffer, 1976a). The instrument was designed to look at behaviorally relevant attentional and interpersonal characteristics. It was based on a theory that seeks to predict behavior based on the interaction between attentional and interpersonal processes, and physiological arousal. The instrument is multidimensional, looking at cognitive styles (attentional processes) as well as interpersonal characteristics. It is postulated that both attentional and interpersonal characteristics have state and trait components and that these can be examined by manipulating arousal. The TAIS has been shown to have internal consistency and good test-retest reliability (Nideffer, 1976a). Evidence for the instrument's construct validity has been found in many studies (Antonelli, Caldarone, & Gatti, 1982; DePalma & Nideffer, 1977; Feifel, Strack, & Nagy, 1987; Salmela & Ndoye, 1986; Weins, Harper, & Matarazzo, 1980; Whitney & Pratt, 1987). Support for the predictive validity of the TAIS has been found in studies by Zuckerman, Larrance, Hall, DeFrank, and Rosenthal (1979), Kirschenbaum and Bale (1980), Landers and Richards (1981), Reis and Bird (1982), and Nideffer (1987). Table 1 provides a description of the attentional and interpersonal characteristics measured by the TAIS.

Development of Sport-Specific Versions of the TAIS


Because of its potential relevance to performance, the TATS has been used a great deal in sport. "The widespread use in sport suggests that many feel the TAIS is valid" (Dewey, Brawley, & Allard, 1989, p. 172). That would certainly seem to be the position of people like Zaichkowsky (1984) and Bemand (1985), who used the TAIS to prepare for the successful capture of the America's Cup Yachting Trophy with Australia 11. In fact, interest in the TAIS has spawned the development of a number of sport-specific versions of the six attentional scales and the information processing scale, including a tennis version (Van Schoyck & Grasha,

Use of the TAIS i n Sport


Table 1 Test of Attentional and lnterpersonal Style Scales
Scale BET Scale description

287

BroadExternalAttention: High scores indicategood environmental awareness and assessment skills ("street sense") Overloadedby Externallnformation: High scores are associated with errors because OET attention is inappropriately focused on irrelevant external stimuli Broad-Internal~iention:~ i g scores h indicate good analytical planning skills BIT OIT Overloadedby lnternallnformation: High scores are associated with errors due to distractions from irrelevant internal sources (e.g., thoughts and feelings) Narrow-FocusedAttention: High scores indicate the ability to remain task oriented, NAR to avoid distractions, and to stay focused on a single job ReducedAttention: High scores are associatedwith errors due to a failure to shift RED attention from an external focus to an internal one, or vice-versa lnformation Processing: High scores are associated with a desire for, and enjoy ment of, a diversity of activity BCON BehaviorControl: High scores are associated with an increased likelihood of "acting out" in impulsive ways andlor a tendency to establish one's own rules rather than strictly adhering to the rules of others CON lnterpersonalControl: High scores are associated with both needingto be in control in interpersonal situations and with actually being in control Self-Esteem: High scores are associated with feelings of self-worth and selfSES confidence Physical Orientation: High scores are associated with having been physically PI0 competitive and with the enjoyment of competitive activities Obsessive: This scale reflects speed of decision making, worry, and anxiety; high OBS scores are associated with increased worry and difficulty making decisions Extroversion: High scores indicate an enjoyment of social involvements and a tendencv to assume leadershio in social situations INT lntroversidn: High scores indicate a need for personal space and privacy IEX IntellectualExpression: High scores indicate a willingness to expressthoughts and ideas in front of others NAE NegativeAffect Expression: High scores indicate a willingness to confront issues, to set limits on others, and to express anger PAE PositiveAffect Expression: High scores indicate a willingness to express support and encouragement to others Depression: A high score is associated with situational (transient) depression DEP

1981), a baseball and softball batting version (Albrecht & Feltz, 1987), and a basketball version (Vallerand, 1983). To validate these sport-specific versions of the TAIS, several authors have subjectedtheir instruments to the rigors of factor analysis in the hopes that factor analysis would confirm the independence of the various attentional scales (Vallerand, 1983; Van Schoyck & Grasha, 1981). Factor analysis of the attentional subscales, however, led to the identification of two general factors, one of which Van Schoyck and Grasha labeled the "scan factor." Subjects scoring high on

this factor also scored high on the scales measuring a broad external (BET) and broad internal (BIT) focus of attention, and high on the information processing subscale (INFP).They scored low on the scale measuring the tendency to make errors of underinclusion, due to a failure to shift the focus of attention (-RED). The second general factor identified consisted of the OET, OIT, and NAR scales. High scorers indicate they have difficulty narrowing attention (-NAR) and are distracted by both external (OET) and internal (OIT) stimuli. Conversely, low scorers indicate they can narrow attention and do not become distracted. Van Schoyck and Grasha labeled this the focusing factor. Dewey et al. (1989) factor-analyzed the seven subscale scores (BET, OET, BIT, OIT, NAR, RED, INFP) using the 40 subjects in their study. They identified two factors (Factor 1 = BET, BIT, and INFP; Factor 2 = OET, OIT, and RED) and found that the NAR scale "had a loading of less than .40 and could not be interpreted as being on either factor" (p. 182). Noting the relatively small difference between their factor structure and that of other researchers, Dewey et al. then compared their factor structure to "Nideffer's original study" and concluded that "A comparison of the scale-to-factor loadings between Nideffer's (1976a) original study and the present study failed to reveal agreement" (p. 182). The article they were referencing was the 1976 article in the Journal of Personality and Social Psychology. In point of fact, that paper did not report on a factor analysis. In each of the studies mentioned, the failure to identify six independent factors reflecting the attentional dimensions of width and direction originally postulated by Nideffer (1976a) resulted in the authors' questioning the validity of the attentional constructs of width and direction of focus. The authors then suggested it might be more valid and useful to begin thinking in terms of the more general concepts of scanning and focusing. The tendency to reduce conceptually independent personality andlor cognitive characteristics into more global measures based on their intercorrelations with each other is not unique to the studies mentioned. Numerous individuals have opted for categorizing any type of attentional distraction as little more than neuroticism as described by Eysenck (1951), as anxiety as described by Spielberger (1972), or as negative affectivity as described by Watson and Clark (1984). Studies like those cited above raise serious questions about the validity and reliability of the attentional dimensions measured by the TAIS. It is my contention that the conclusions drawn by these authors are not justified on the basis of their research. If I am correct, then the suggestion to reduce several highly useful, independent attentional constructs into one or two more global characteristics dramatically reduces the practical utility of the TAIS. Although I would be among the first to recognize that there are distinct limitations to the predictive validity of the TAIS (many of these due to uncontrolled factors such as subject's response sets), one of the last things I would suggest would be that conceptually useful attentional characteristics be subsumed under more general labels whether those be called scanning factors, neuroticism, anxiety, or anything else. Conclusions about the relative independence of TAIS subscales drawn by Vallerand (1983), Van Schoyck and Grasha (1981), and Dewey et al. (1989) are inappropriate for several reasons. First, they fail to recognize that statistical independence is a relative construct. A correlation of .70 between two variables means they are statistically dependent (approximately 50% of the variability between the two scores can be explained). At the same time, a correlation of

Use of the TAIS in Sport

289

.70 demonstrates that the two variables are independent (i-e., 50% of the variability between scores cannot be explained). When your own analyses leave a great deal of the variance unexplained, it is inappropriate to conclude that variables do not measure independent characteristics. While you might conclude that you failed to find evidence within your study for the independence of the variables (e.g., you did not find the orthogonal factors you were looking for), you cannot conclude that it doesn't exist. From a methodological standpoint, there are several reasons to question the data presented by the previously mentioned authors. First, it is unreasonable to assume, as Dewey et al. (1989) have, that a factor analytic solution of 7 of the TAIS scales (BET, OET, BIT, OIT, NAR, RED, INFP) should result in the same factor structure as factor analytical solutions that have looked at all 17 of the TAIS subscales. In addition, Dewey et al. question the stability of previous studies based on their own analysis, yet their analysis was conducted on only 40 subjects. To begin to have confidence in the stability of their own data, these authors should have had a minimum of 10 subjects per variable (70 subjects). In addition, both laboratory and field studies like the ones conducted by Vallerand (1983), Van Schoyck and Grasha (1981), and Dewey et al. (1989) pay little or no attention to the response set issues identified by Morgan (1980) and Nideffer (l981,1987,1989a), which in turn contaminates much of the research data being published. Finally, the conclusions drawn by the previously mentioned authors need to be questioned on the basis that these authors have not understood the theory underlying the development of the TAIS and have not designed studies which would test that theory. An illustration of the failure to understand the theory that underlies the TAIS is reflected in the following quote from the Van Schoyck and Grasha article: "If direction of attention was adequately assessed by the TAIS . . . one would have expected the subscales possessing this dimension to have been negatively correlated. Subscale combinations BET and BIT as well as subscales OET and OIT should have been negatively correlated" @. 156). Few theorists would predict that individuals good at assessing the environment would be bad at analyzing. In fact, good analytical skills require good assessment skills, especially in sport. An athlete cannot adequately analyze a task unless he or she has attended to external task-relevant information (e.g .,position of the defense). To assume that positive attentional abilities should be negatively correlated would be like asserting that individuals who are high on measures of verbal intelligence should score low on measures of quantitative intelligence. Or to use the example that follows, it would be like suggesting that the taller a person is, the less he or she should weigh. In my original article and in all subsequent writing, I have emphasized that the "average" person is capable of developing all four of the attentional styles I identified: broad-external, broad-internal, narrow-external, narrow-internal. I have stated that individuals have attentional preferences and that it is only under conditions of increasing arousal that these preferences begin to affect performance. Before moving into the actual assessment process, however, it's important to point out that everyone's normal attentional style is affected by competitive factors. Anxiety and arousal are natural components of most athletic competitions and have a direct effect on attentional processes. The mere assessment of your attentional style and of the attentional demands on the situation is

not enough to predict how you will perform. If you wish to know when to psych yourself up or psych yourself down, it's critical that you first understand how anxiety affects your attentional style. (Nideffer, 1976b, p. 65)

None of the researchers who have attempted to examine the predictive validity of the TATS have systematically manipulated the level of arousal within tasks. This includes the studies cited in this paper that do provide some support for the TAIS. Unless researchers manipulate arousal, unless they look for withinsubject changes, and unless they control for response set influences, they cannot expect to find the relationships that they have been hypothesizing or to account for much of the variance in those studies designed to examhe the predictive validity and/or reliability of the TAIS. To begin to see the effects that controlling response sets can have on the predictive utility of the TAIS, I would refer you to a paper published in The Sport Psychologist (Nideffer, 1987). This particular study enhanced the predictive validity of the TAIS by manipulating divers' response sets.

Independence of TAIS Subscales


It is naive to assume that factor analysis of subscale scores on a test should result in the identification of independent factors that reflect the various subscales (e.g., a separate self-esteem factor, separate BET and BIT factors).' To illustrate this fact, I conducted a factor analysis on the height, weight, position, class, and level of sport experience of the 99 players listed in the 1988 San Diego State University media guide. Players were assigned a number based on their position (1 = kickers and punters; 2 = quarterbacks, defensive safety, running backs, wide receivers; 3 = fullbacks, defensive linebackers; 4 = all other offensive and defensive linel l men). Table 2 presents the means, standard deviations, and correlations on a measures. Table 3 presents the results of a principal-components factor analysis using a varimax rotation, and default values (eigenvalues of 1.0) to determine when to stop factoring. As you can see from Table 2, the intercorrelation between the variables of position, height, and weight are very high (.638, .798, and .749), as are the correlations between class and experience (.705). The obvious consequence of these rather strong statistical relationships when subjected to factor analysis can be seen in Table 3. What you have are two factors that account for approximately 83% of the variance. The program stopped factoring as soon as eigenvalues dropped below 1. Because five variables were included in the analysis, there were five eigenvalues available for distribution (one for each variable), each eigenvalue accounting for one fifth of the total variability of scores (20 %). If there had been 20 variables or scales in the analysis, there would have been 20 eigenvalues available, each accounting for 5% of the variance. As you can see, the two factors combined account for over 4 of the eigenvalues in this analysis.
*Froma theoretical standpoint the complete independence of attentional and interpersonal constructs might be highly desirable. Independent scales would make prediction much easier. From a practical standpoint, however, scales that were completely independent would not reflect what we know to be true about human behavior (e.g., that various attentional and interpersonal skills are interrelated).

Use o f the TAIS i n Sport

291

Table 2 Means, Standard Deviations, and Correlations Between Football Players' Height, Weight, Position, Class, and Level of Experience (N = 99)
Correlation Matrix Height Weight Class

Position Position M= 2.98, SD= .92 Height M= 74.06, SD = 2.78 Weight M= 222.27, SD = 36.89 Class M = 2.65, SD = 1.09 Experience M = 2.60, SD- 1.57
1.OOO .638* * .798**

Exper.

1.OOO .749* * -.I75 1.OOO -.123 1.000 ,705"" 1.OW

- .219* - .096

- .092

- .085

Table 3a Principal Components Factor Analysis of 1988 SDSU Football Team (N = 99)
Variable Position Height Weight Class Experience Communality
.808 .767 ,886 .854 356

Factor
1 2

Eigenvalue
2.58 1.59

oh of var.

51.60h 31.8%

Table 3b Rotated Factor Matrix


Experience Weight Position Height Experience Class Factor 1 Factor 2

Based on the responses they made to their own factor analyses, I have to assume that researchers mentioned earlier would conclude from this analysis that position, height, and weight are really all measuring the same thing and we can simplify this by talking about some unitary characteristic. Obviously, telling a tailor or dressmaker that he or she can throw out height or weight since there is so much redundancy in the information would be looked upon by those people in disbelief. Likewise, to tell a coach of a football team that he can ignore the l l three of these variables would be ridiculous. True. inde~endent contribution of a there is statisticaldependence reflected in the analysis, but there is also statisticd independence. Researchers must not forget that a correlation of .71 only accounts for 50% of the variance! Far from being upset by the fact that various attentional andlor interpersonal characteristics are intercorrelated, we should take comfort from the fact that correlations exist, because these increase our predictive abilities (e.g., we are likely to be correct if we assume a tall person is heavier than a short person, just as we are going to be correct if we assume a person who can effectively attend to external information [BET] will be more likely to effectively attend to internal information [BIT]). At the same time, we cannot ignore the tremendous importance the practitioner can and should apply to score of those individuals who deviate from the norm. It is this point that helps demonstrate not only the conceptual independence of attentional and interpersonal characteristics but the statistical independence as well.

Factor Analysis of the 17 TAIS Subscales


Although no factor analysis data were reported in the 1976a article on the TAIS, several studies have factor-analyzed subscale scores. These have been summarized in the book, Predicting Human Behavior (Nideffer, 1989b). Table 4 provides a summary of these principal-components factor analyses, using varirnax rotations, and helps illustrate the effects that response sets can have on the factor structure of psychological inventories. As you can see from Table 4, Factor 1 resembles the scanning factor described earlier. The fact that additional scales load on this factor (e.g., SES l l of the TAIS and CON) is a function of the fact that the analysis included a subscales, not just the attentional scales. It is important to point out that the amount of variance accounted for by this factor differs dramatically across the five studies, from 15.9 to 45.7%. As was pointed out earlier, the higher the intercorrelationsbetween scales loading on a factor, the higher the eigenvalue for that factor, and the more variance accounted for. Because 17 scales were analyzed in each of the studies presented in Table 4, an eigenvalue of 1.0 accounts for approximately 5.9% of the variance. This means that for the police applicant population this particular factor had an eigenvalue of 45.715.9 or 7.74. It also means that we are less likely to find as many independent factors in the police analysis as we are in the other analyses because so much of the variance has been accounted for by the attentionally effective, or scanning, factor. You should notice that although Table 4 lists six factors, only four of the factors for police applicants accounted for enough of the variance (5.9%) to be classified as independent factors on the basis of traditional criteria.

Use of the TAIS in Sport


Table 4 Principal Components Factor Analysis of TAIS Subscales
U.S. college German college Music students Police applicants

293

Scales

Psychiatric patients

Factor f = Confident/attentionaIly effective BET .79 .83 .78 INFP .88 .86 .87 CON .41 .56 .47 .34 .51 .33 SES BET -81 .82 .85 IEX EXT SlGSQ 16.1% 20.8% 15.8%~ Factor 2 = Performance anxiety/speed of decision meking .87 .88 .84 RED .81 OBS -91 .89 .43 .34 .67 OIT OET .58 NAR .34 SlGSQ 12.7% 14?6O/b 13.9% Factor 3 = Extroverted EXT .77 .63 .79 INT .& .89 .86 .89 PAE .88 .89 .46 .31 SES SIGSQ 14.8% 13.6% 14.4% Factor 4 = Overloaded-impulsive/distractibie OIT .73 .73 .47 .81 .79 .91 NAR .79 .52 OET .81 .38 .81 BCON 8.9% 12.3% SlGSQ 17.2% Factor 5 = Physical orientation/leadership CON .60 -61 .50 .38 .50 SES .42 .95 PI0 .94 .94 8.4% 10.3% SIGSQ 9.6% Factor 6 = Angry-impulsive/intellectoally critical .89 NAE .73 .93 .81 .77 BCON .83 IEX SlGSQ 10.4% 10.6% 11.O%

.90 .84 .52 .54 .61 .54 .51 45.7% .88 .81 .70 .71

"

-81 .82 .83 .69 .86 .74 .75 31.0% .85 .89 .70

24.5%

18.2%

- .47 .81 8.7%

.66

.41

- .71
.50

.43

- .85

4.0%

8.7%

.n
.47 .69 12.4% .77 .58

.83 7.3% .68 .65


d

4.7%

15.0%

Look at Factor 2 in Table 4. This factor seems to be similar to the "focused" factor described by Van Schoyck and Grasha (1981). Notice that for the student groups, the amount of variance accounted for by Factors 1 and 2 ranges from about 29 to 35 % . In contrast, the amount of variance accounted for by police applicants and psychiatric patients for these two factors is 70 and 49%, respectively. So what does this mean? What it means is that it was much more common for these particular scale patterns to rise and fall together for the police applicants and psychiatric patients than it was for students. So how do we interpret that? Basically there are two possible interpretations. Before presenting those, however, I must point out that the average police applicant scores high on Factor 1 and low on Factor 2. The average psychiatric patient scores in the opposite direction. Thus, police applicants tell us they are attentionally effective, have a high level of self-esteem, and do not get overloaded and/or make errors of underinclusion (RED). Psychiatric patients indicate they are not attentionally effective, have low levels of self-esteem, are highly distractible and make errors of underinclusion. One interpretation for these findings is that police applicants are more effective in general than college students, and that is what the pattern of responses shows. This same argument would state that by definition psychiatric patients are less effective and it should show in their test scores. A second interpretation, however, is that police applicants and psychiatric patients were motivated to distort their responses. The former wanted a job and so they tried to look good. The latter wanted sympathy, so they tried to look bad. It should be obvious that both interpretations are viable. That means it is extremely important for the researcher and/or practitioner to diicriminate between these two possibilities. One does that by interviewing individuals, by consensually validating responses, by looking for extreme profiles, by attempting to ensure cooperation and honesty, and by providing subjects with an appropriate response set (Nideffer, 1981, 1987, 1989a). Controlling response sets, and attempting to determine when response distortions are present, is something that none of the investigators have provided any evidence of doing! This makes it very difficult to interpret their data and almost impossible to compare it to data collected by other investigators. Setting aside the issues of the inappropriate interpretation of factor analysis and poorly controlled studies, where does all of this leave us with respect to the validity of the attentional scales on the TAIS? Does the fact that most factor analyses of the 17 TAIS scales result in five factors, or the fact that the analysis of the six attentional scales results in two factors, mean that we should limit our interpretations to those factors? Does it mean that there are only two attentional scales or attentional characteristics measured by the TAIS? Absolutely not! We do, however, need to demonstrate the predictive and/or discriminant validity of the various scales. For example we need to identify situations in which the scanning factor is broken down (e.g., where subjects score high on BET but not BIT, or where BIT is related to some criterion variable and BET is not). Several of the studies mentioned at the beginning of this article have done just that. I would like to discuss a recent study conducted at the Australian Institute for Sport (Nideffer & Bond, 1989).

Use o f the TAIS i n Sport

295

Attentional and Interpersonal Style as a Function of Type of Sport


The Australian Institute for Sport (AIS), Australia's training center for Olympic and elite level athletes, has been using the TAIS as a counseling tool with individuals training at the institute since late 1983. In December 1987 they began analyzing the data that had been collected on 1,799 athletes. Earlier research had suggested that personality characteristics such as extroversion and introversion and the need for control should discriminate between sport groups (Kroll & Crenshaw, 1970; Schurr et al., 1977; Singer, 1969). In addition, our concepts of the attentional demands of the different sports would lead us to hypothesize that there would be differences in attentional abilities as a function of type of sport. We believed, as Morgan (1974) suggested, that individuals either gravitate toward certain sports based on various personality characteristics or develop these characteristics as a function of their involvement in those sports. Table 5 presents the results of an SPSSX discriminant function analysis that was conducted on 814 athletes (Nideffer & Bond, 1989). These athletes were subdivided into three sport categories: (a) closed-skill sports that included diving, gymnastics, shooting, golf, archery, and skating; (b) open-skilledtindividual sports that involve one-on-one competition such as judo, wrestling, fencing, squash,and tennis, and sports that require the individual to react to the environment such as cycling, canoeing, kayaking, and skiing; and (c) team sports such as baseball, basketball, soccer, hockey, netball, lacrosse, and volleyball. Figures 1 and 2 have been
Table 5a Discriminant Analysis of Australian Athletes Involved In Closed Skill, Open Skill, and Team Sports (summary table)
Wlks' lambda

TAlS scale PI0 UCT CON RED SES NAR OBS BET OIT BCON IEX BIT

Sig.

Label Competitiveness Extroversion Leadership-control Underinclusive Self-esteemlconfidence Focusedldisciplined Worrylanxiety "Street sense" aware Internal distractions lmpulsivelown rules Intellectual expression Analytical

Table 5b Classification Results


Actual group Closed skill Open skill Team sport Predicted group
N

Closed

Open

Team

225 325 264

129 57.3% 89 27.4% 60 22.7%

45 20% 148 45.5% 70 26.5%

51 22.7% 88 27.1 % 134 50.8%

Percent of grouped cases correctly classified: 50.5%.

w
Z +-

{ so
w

Closed Skill Open Skill HTearn Sport

40 Aware Analytical Focused External Distractability

Figure 1

- Attentional style as a function of type of sport.

// /
;/ 1
II

w I2
Z

60

I/IIli'
I
I

{ so
w

40

Closed Skill m o p e n Skill H T e s r n Sport

Iil I
30 Control
Competitive

/11I I
(I I

Extroverted

Introverted

Figure 2

- Interpersonal style as a function of type of sport.

Use of the TAIS i n Sport

297

included to show the differences in both attentional and interpersonal characteristics as a function of sport. The percentiles presented in the figures were obtained by computing a Z-score and then converting the Z-score to a percentile. Z-scores were calculated by taking the mean for each group, on a particular subscale, and then subtracting it from the mean score for the entire Australian sample (1,799 athletes). The remainder was then divided by the standard deviation for the combined male and female sample. As you can see, 12 of the 17 TAIS subscales contributed significantly to the equation separating athletes on the basis of their type of sport. The interpersonal differences found between groups were similar to those described by Schurr et al. (1977), thus providing some consensual validation for their results. The attentional differences are consistent with the theory proposed by Nideffer (1976a, 1981,1989a) and provide support for the construct validity of the TAIS. As you can see, individuals involved in closed-skill sports tend to be more introverted and less extroverted, to have a narrower focus of attention, and to be less externally distractible. Individuals involved in open-skill sports are more competitive (PfO) and score higher on the control scale on the TAIS (CON). Individuals involved in team sports have the highest scores on extroversion and, along with those involved in open-skill sports, score higher on the attentional scale measuring a broad-external focus of attention. It is important to note that this group scores lowest on the broad-internal scale (evidence contradictoryto the notion of a simple scanning factor). Those involved in team sports also score lower in terms of need for control. What would be lost by collapsing scales? Based on Table 5, BET and BIT made independent contributions to the prediction equation, as did NAR, OIT, and RED. To have collapsed those scales into two global measures would have resulted in the loss of extremely useful data. This point is also true of RED, OBS, and OIT, since these three scales are intercorrelated, and since all three correlate with anxiety, some would suggest that they should be collapsed into one variable. It is quite likely that anxious persons will have difficulty making decisions (OBS), will be distracted by internal stimuli (OIT), and will make errors of underinclusion because their attention narrows too much (RED). When this pattern occurs frequently enough across individuals, a factor is created that increases our ability to make predictions about behavior. The greater the amount of variance accounted for, the more likely we can predict an individual's score on OIT based on hisfher score on RED. This does not preclude the fact that a great amount of the variance remains unexplained. It is the unexplained variance that we seek to explain by breaking the factor scores down into their individual components. Just as it is possible for a person to score high or low on all three of the scales mentioned (OBS, RED, OIT), it is also possible, and indeed it happens, that athletes will score high on one or two of these scales and low on others. For example, athletes who make mistakes because they overanalyze (a condition coaches refer to as paralysis by analysis) typically are not anxious and do not have difficulty making decisions. They score high on OIT but not on the other two. In fact, these individuals have a positive correlation between OIT and BIT. Likewise, some athletes are extremely dedicated and make errors of underinclusion (RED) because they let other things slide such as personal relationships or academic work. For some this is a conscious choice. They score high on RED but low on the other two.

Conclusions
Many other examples could be given, especially when breaking subjects into groups on the basis of age and sex, that demonstrate the conceptual, statistical, and practical independence of the attentional scales measured by the TAIS (Nideffer & Bond, 1989). The most important point to be made here, however, is the fact that the information lost by collapsing scales tears the heart out of any training program a practitioner would create based on test results. People in sport are well aware of the need to become more specific in their training programs, not less specific. Collapsing scales may make sense for a researcher who needs to reduce the number of variables helshe is trying to investigate, it does not make sense for the practitioner who is trying to provide feedback to an individual. Nor does it make sense in those research situations where a sufficient number of subjects can be tested, where response sets can be controlled, and where within-subject changes on attentional scores can be examined as a function of the level of stress and arousal in performance settings. A major criticism in the field in recent years has concerned the blanket administration of general procedures such as progressive relaxation, autogenic training, positive thinking, mental rehearsal, and so forth. Attempts have been made to just@ training on the basis of a global need to relax in sport. The absurdity of such assumptions has been demonstrated time and time again. Let's not regress, let's progress! We can begin to do that by designing better studies, by being sensitive to response bias and underlying theory, and by becoming more responsible with respect to the conclusions drawn about the stability andtor independence of variables on the basis of factor analysis.

References
Albrecht, R.R., & Feltz, D.L. (1987). Generality and specificity of attention related to competitive anxiety and sport performance. Journal of Sport Psychology, 9,231-248. Antonelli, F . , Caldarone, L., & Gatti, M. (1982). Prof1e psychologique du gyrnnaste national Italien. In Mental trainingfor coaches and athletes @p. 100-101). Ottawa: Coaching Association of Canada. Bertrand, J. (1985). Born to win. New York: Bantam Books. Cox, R.H. (1985). Sportpsychology: Concepts and applications. Dubuque, LA: Brown. DePalma, D.M., & Nideffer, R.M. (1977). Relationshipsbetween the Test of Attentional and Interpersonal Style and psychiatric subclassification. Journal of Personality Assessment, 41, 622-63 1. Dewey, D., Brawley, L., & Allard, F. (1989). Do the TAIS attentional-style scales predict how information is processed? Journal ofqort & Exercise Psychology, 11,171-186. Eysenck, H.J. (1951). The organization or-ty, Journal sfPemomlity, 20, 101-117. Feifel, H., Strack, S., & Nagy, V.T. (1987). Degree of life-threat and differential use of coping modes. Journal of Psychosomatic Research, 31(1), 91-99. Kane, J.E. (1980). Personality research: The current controversy and implications for sports studies. In W.F. Straub (Ed.), Sportpsychology: An analysis of athlete behavior (2nd ed.). Ithaca, NY: Mouvement.

Use of the TAIS in Sport

299

Kirschenbaum, D.S., & Bale, R.M. (1980). Cognitive-behavioral skills in golf: Brain power golf. In R.M. Suinn (Ed.), Psychology in sports: Methods andapplicatiom. Minneapolis: Burgess. Kroll, W. (1976). Reactions to Morgan's paper: Psychological consequences of vigorous physical activity in sport. In M.G. Scott (Ed.), The Academy Papers. Iowa City: American Academy of Physical Education. Kroll, W., & Crenshaw, W. (1970). Multivariatepersonality profile analysis of four athlete groups. In G.S. Kenyon (Ed.), Contemporarypsychologyof sport: Second Znternational Congress of Sport Psychology. Chicago: Athletic Institute. Landers, D.M., & Richards, D.E. (1981). Test of Attentional and Interpersonal Style scores of shooters. In G.C. Roberts & D.M. Landers (Eds.), Psychology of motor bekmior and sport. Champaign, IL: Human Kinetics. Martens, R. (1976). Sport compmmtion anxiety test. Champaign, IL: Human Kinetics. Morgan, W.P. (1974). Selected psychological considerations in sport. Research Quarterly, 45, 324-339. Morgan, W.P. (1980). Sport personology: The credulous-skeptical argument in perspective. In W.F. Straub (Ed.), Sportpsychology: An analysis ofathlete behavior (2nd ed.). Ithaca, NY: Mouvement. Nideffer, R.M. (1976a). T e s t of Attentional and Interpersonal Style, J m m l of Personality and Social Psychology, 34, 394-404. Nideffer, R.M. (1976b). The inner athlete. New York: Thomas Crowell. Nideffer, R.M. (1981). The ethics andpractice of applied sportpsychology. Ithaca, NY: Mouvement. Nideffer, R.M. (1987). Issues in the use of psychological tests in applied settings. The Sport Psychologist, 1, 18-28. Nideffer, R.M. (1989a). Anxiety, attention and performance in sports: Theoretical and practical considerations. In D. Hackfort & C.D. Spielberger (Eds.), Anxiety in sports: An international perspective. New York: Hemisphere. Nideffer, R.M. (1989b). Predicting h u m behavior (2nd 4.). Oakland, CA: Enhanced Performance Services. Nideffer, R.M., & Bond, J. (1989). Test of attentional and interpersonal style-Cultural and seam1 dzyerences. (Manuscript to appear in Proceedings of the X)(I Banff International Conference on Psychology, Sport and Health Promotion, Banff, Canada.) Reis, J., & Bird, A. (1982). Cue processing as a function of breadth of attention. J o u m l of Sport Psychology, 4, 64-72. Rushall, B.S. (1973). The status of personality research and application in sports and physical education. J a m 1 of Sports Medicine and Physical Fitness, 13,281-290. Salrnela, J.H., & Ndoye, O.D. (1986). Cognitive distortions during progressive exercise. Percephlal and Motor Skills, 63, 1067-1072. Schurr, K.T.; Ashley, M.A., &Joy, K.L. (1977). A multivariate analysis of male athlete characteristics: Sport type and success. Multivariate Erperiimental ClinicalResearch, 3, 53-68. Singer, R.N. (1969). Personality differences between and within baseball and tennis players. Research Quarterly. 40, 582-587. Spielberger, C.D. (1972). Anxiety as an emotional state. In C.D. Spielberger (Ed.), Anxiety: Current trends in theory and research (Vol. 1). New York: Academic Press.

Vallerand, R.J. (1983). Attention and decision making: A test of the predictive validity of the Test of Attentional and Interpersonal Style (TAIS) in a sport setting. Journal of Sport Psychology, 5 , 449-459. Van Schoyck, R.S., & Grasha, A.F. (1981). Attentional style variations and athletic ability: The advantages of a sport-specific test. Journal of Sport Psychology, 3, 149-165. Whitney, D.L., & Pratt, R.W. (1987). Paying attention: A study of attentional and interpersonal styles of hotel and restaurant administration students. Hospitality and Education Research Journal, 11(3), 43-58. Wiens, A.N., Harper, R., & Matarazzo, J.D. (1980). Personality correlates of nonverbal behavior. Journal of Clinical Psychology, 36(1), 205-215. Zaichkowsky, L.D. (1984). Attentional style. In W.F. Straub & J.M. Williams (Eds.), Cognitive sport psychology. Lansing, NY: Sport Science Associates. Zuckerman, M., Larrance, D., Hall, J., DeFrank, R., & Rosenthal, R. (1979). Posed and spontaneous communication of emotion via facial and vocal cues. Journal of Personality, 47, 712-733.

Manuscript submitted: July 7, 1989 Revision received: November 28, 1989

You might also like