Construct Validity of the Anxiety Rating Scale-2 with Individual Sport Athletes.

Russell, William D. ; Cox, Richard H.

The Anxiety Rating Scale-2 (ARS-2) is a shortened version of the Competitive State Anxiety lnventory-2 (CSA 1-2) that has recently displayed greater concurrent validity than the initial Anxiety Rating Scale (Cox & Robb, 1999). It was the purpose of this investigation to further determine the psychometric properties of the ARS-2 by examining the construct validity of the revised short-form inventory with individual sport athletes. Participants in this investigation were 302 college age intramural athletic participants in several individual sport paradigms (table tennis, one-on-one basketball, racquetball and badminton, pickleball, and wrestling). Results from a 2 x 2 (gender x outcome,) MANOVA revealed a significant main effect for outcome but no significant main effect for gender or gender by outcome interaction. Follow-up descriptive discriminant analysis structure coefficients for cognitive anxiety, somatic anxiety and confidence were .12, . 18, and . 90, respectively indicating that confidence was meaningf ul in predicting game outcome. Results demonstrate the construct validity of the ARS-2, in that winning and losing individual sport athletes could be discriminated as a function of scores on the ARS-2. Results are discussed in terms of the importance of short rating scales within precompetitive anxiety measurement and the future examination of the directionality concept.

The development of short-form inventories to assess precompetitive anxiety has shown much empirical progress as various investigations (Krane, 1994; Cox, Russell, & Robb, 1999) have developed short rating scales to be more practical, less intrusive, yet maintain strong psychometric properties in order to assess competitive state anxiety within a multidimensional context (Martens, Vealey, & Burton, 1990). Specifically, the Anxiety Rating Scale was developed as a short rating scale derived, from multiple regression procedures, from the Competitive State Anxiety-2 (CSAI-2; Martens et al., 1990) and was designed within a single statement format to assess precompetitive cognitive state anxiety, somatic state anxiety, and self-confidence. The concurrent validity of the ARS was established with team sport intramural athletes (Cox, Russell, & Robb, 1998; 1999) and with individual sport athletes (Cox, Reed, & Robb, 1997). Results of these investigations have shown the original ARS to be moderately correlated with the subscales of the CSAI-2 (.60 to .70)

Comparative correlations between anxiety subcomponents of the ARS and the Mental Readiness Form-Likert (MRF-L) with the cognitive and somatic state anxiety subscales of the CSAI-2 have consistently favored the ARS (Cox et al., 1997; Cox et al., 1998). Subsequent investigation of ARS revisions which have stemmed from concerns over response confusion in the original ARS have yielded larger concurrent validity coefficients with the CSAI-2 than the original anxiety rating scale.

In creating the three statements associated with cognitive state anxiety, somatic state anxiety, and state self-confidence, the ARS was developed using stepwise multiple regression procedures to identify the "best" three-element model for each subcomponent. These three elements were combined into an overall statement that represented either cognitive anxiety, somatic anxiety, or self-confidence, which were set to a 7-point Likert scale, allowing athletes to rate how they felt immediately prior to competition.

Critics of the ARS have indicated that ARS statements contain more than one psychological construct and that the subscale items may be misconstrued as representing more than one dimension. However, the argument with the ARS is that athletes are able to respond by providing a "global" response to their precompetitive anxiety. The Anxiety Rating Scale -2 (ARS-2) was revised by Cox and Robb (1999) by looking at the nine items for each CSAI-2 subcomponent and selecting three items most logically related to cognitive anxiety, somatic anxiety, or self-confidence. Using this approach the ARS-2, then was based upon CSAI-2 items 13, 16, and 22 to form the aggregate statement for cognitive state anxiety; items 5, 8, and 17 to form the basis of the somatic anxiety statement; and items 6, 12, and 18 to form the self-confidence statement. Recently, Cox and Robb (1999) investigated the concurrent validity of the ARS-2 by comparing it with the original ARS and an initial revision of the ARS with intramural volleyball parti cipants. Results showed that, collapsing across gender, the ARS-2 yielded the highest correlations for cognitive anxiety, somatic anxiety, and self-confidence. They concluded that the ARS-2 exhibited the highest level of concurrent validity.

An appropriate next-step research question for examining the ARS-2 is to determine greater evidence of its construct validity. This can be determined by showing that groups of athletes, who theoretically should differ on state anxiety and confidence, do in fact differ on these constructs, as measured by the ARS-2 (Thomas & Nelson, 1996). Recently, the construct validity of the ARS-2 was examined with intramural basketball team participants (Cox, Robb, and Russell, 1999). MANOVA procedures followed up by discriminant analyses indicated that confidence followed by cognitive anxiety were meaningful in predicting game outcome. These results demonstrated construct validity of the ARS-2, in that winning and losing teams could be differentiated as a function of scores on the ARS-2.

Since the ARS-2 was based upon the CSAI-2 and the multidimensional model of competitive anxiety (Martens et al., 1990), support for the ARS-2 could be determined by replicating proposed differences based upon theoretical hypotheses within this model. One line of evidence within this area is the finding that skill is related to anxiety. For example, experienced sport parachutists have demonstrated lower physiological and fear estimates than novice jumpers (Fenz & Epstein, 1967; Fenz & Jones, 1972). This was supported by other researchers for gymnasts (Mahoney & Avener, 1977), racquetball players (Meyers, Cooke, Cullen, & Liles, 1979), and wrestlers (Highlen & Bennett, 1979). It has also been shown that athletes with lower scores in cognitive and somatic anxiety, and higher scores on self-confidence perceived their anxiety as more facilitative toward performance and had higher pre-performance expectations (Wiggins & Brustad, 1996). The rationale for this research question is that more skilled individual athlet es should display lower levels of precompetitive cognitive and somatic anxiety while displaying higher self-confidence compared to less skilled individual athletes. This difference in skill level would then be directly observable through match outcome, with more skilled athletes more likely to be match winners and less-skilled athletes more likely to be match losers.

Individual-athlete sports also provide a good paradigm for determining construct validity, because match outcome is dependent solely upon individuals' performance and because diffusion of responsibility for performance errors is minimized (Scanlan, 1975). After finding evidence for construct validity using intramural basketball team athletes, Cox, Robb, and Russell (1999) recommended that a next step in establishing the construct validity of the ARS2 would be that it be tested using individual sport athletes in various competitive settings. Within this individual sport paradigm, significantly lower cognitive and somatic anxiety scores, and significantly higher self-confidence scores for match winners, compared to match losers, across several individual-athlete sport events would provide additional construct validity for the ARS-2.

It was the purpose of this study to determine the construct validity of the ARS-2 by comparing match outcome with precompetitive responses on the ARS-2. It was hypothesized that for round robin play within intramural competition, the ARS-2 would discriminate between winning and losing athletes in that cognitive anxiety and somatic anxiety scores would be significantly lower for match winners compared to match losers. In addition, it was hypothesized that ARS-2 self-confidence scores would be significantly higher for match winners compared to match losers.

Method

Participants

Participants for this research were 271 male and 31 female (N = 302) intramural athletes (mostly undergraduates) competing in six individual sport competitions (table tennis, one-on-one basketball, racquetball, and badminton, pickleball, and wrestling) at a large American Midwest university. The mean age of the male participants was 20.37 years (SD = 2.68) and the mean age of the female participants was 19.71 years (SD = 1.19). Use of human subjects for the purpose of research was approved by the university and participants' inventories were coded for anonymity and consent provided by inventory completion.

Instrument

For this study, the ARS-2 was used and was based upon previously mentioned procedures. The original ARS has previously demonstrated validity for measuring competitive state anxiety in intramural athletes (Cox, Russell, & Robb, 1999). Recently, the ARS-2 has yielded correlation coefficients of .67, .69, and .75 with the CSAI-2 cognitive anxiety, and self-confidence subscales, respectively (Cox & Robb, 1999). The current ARS-2 is based upon the three aggregate statements set to a 7-point Likert scale ranging from 1 (not at all) to 7 (intensely so) which an enables an athlete to rate how they feel immediately prior to competition. The ARS 2 measures cognitive state anxiety, somatic state anxiety and self-confidence, respectively, using the following statements:

1. I feel concerned about performing poorly, choking under pressure, and that others will be disappointed with my performance.

2. I feel jittery, my boy feels tense, and my heart is racing.

3. I feel comfortable, secure, and confident about performing well.

Unlike the CSAI-2, which takes athletes 3-5 minutes to complete, the rating scale format of the ARS-2 enables particpants to complete it within seconds.

Procedures

Participants of the selected individual sport competitions were approached approximately 15 mm prior to the start of their round robin competition and asked if they would volunteer for the study. Participants who agreed were asked to respond to ARS-2 items relative to how they felt at that moment in time. Since precompetitive ARS-2 measures were compared with that individual's match outcome (win, loss) their performance was tracked and their match result recorded on their coded inventory. It has been argued that repeated assessment of precompetitive psychological constructs can be influenced by previous match outcome (Sonstroem, Harlow, & Salisbury, 1993) and performance expectations (Wiggins & Brustad, 1996). Repeated assessment of precompetitive self-confidence might be expected to increase prior to a second match as a result of successful match outcome for their first match. Thus, participants were only assessed once during their round robin competition.

Analysis

Data collected during this investigation were analyzed using a gender x outcome (2 x 2) multivariate analysis of variance (MANOVA) with outcome (win, loss) and gender (male, female) serving as categorical independent variables and the three ARS-2 subscales (cognitive state anxiety, somatic state anxiety, and self-confidence) serving as continuous dependent variables. An a priori decision was made not to include sport as an independent variable because all of the events were individual sport paradigms. Because the dependent variables of interest within this study (cognitive state anxiety, somatic state anxiety, and self-confidence) have shown to be correlated from the literature (Martens et al., 1990), it is considered inappropriate to examine them univariately (Pedhazur, 1997). Significant multivariate F-ratios were followed up by descriptive discriminant analysis procedures using a combination of ARS-2 subscale scores to discriminate among levels of outcome and gender. Total structure coefficients were anal yzed from a descriptive discriminant analysis to determine which component of the ARS-2 was the most important in discriminating between match outcome (winning or losing). In addition, intercorrelations among the three dependent variables (ARS-2 cognitive anxiety, somatic anxiety, and self-confidence) were calculated to verify these subscales were moderately significantly (p [less than] .05) correlated, as measured by the ARS-2.

Results

Pearson product moment correlations indicated that there were moderate and significant correlations between cognitive and somatic anxiety (r = .59, p [less than] .001), cognitive anxiety and self-confidence (r = .24, p [less than] .001) and somatic anxiety and self-confidence (r = -.23, p [less than] .001), respectively. Table 1 contains means and standard deviations for gender and match outcome for cognitive anxiety, somatic anxiety, and self-confidence as measured by the ARS-2.

Multivariate Analysis of Variance

A gender by outcome (2 x 2) MANOVA was conducted on the three ARS-2 subscales. Results of the MANOVA yielded a nonsignificant Wilks's Lambda statistic for the interaction, F (3,296) = 1.18, p = .32 and for the main effect of gender, F (3,296) = .99, p = .40. However, MANOVA results did yield a significant main effect for outcome and the three ARS-2 subscales, F (3,296) = 2.78, p = .04. The resultant Wilks's Lambda for this main effect was .973.

Discriminant Analysis

In order to determine the relative contribution of the dependent variables in the significant main effect for outcome, the MANOVA was followed by a descriptive discriminant analysis to determine the relative importance of the dependent variable in discriminating between winners and losers. The discriminant function resulting from this analysis is used in calculating a discriminant score for each participant. This discriminant score is correlated with each participant's original dependent variable and the resulting correlations are called structure coefficients of [greater than] .30 are considered meaningful (Pedhazur, 1997). In this study, the structure coefficients were based upon total sums of scores and cross products matrix and are referred to as total structure coefficients. The total structure coefficients for cognitive anxiety, somatic anxiety, and self-confidence were -.12, .18, and .90, respectively. These coefficients suggest that only self-confidence was meaningful in discriminating between winnin g and losing atheltes.

Discussion

The three subscales of the ARS-2 were examined for utility in explaining the relationship with match outcome (winning or losing) for individual sport events to provide evidence for construct validity of the ARS-2. Since the main effect for outcome was significant and there were only two levels of outcome, total structure coefficients provided evidence for the construct validity of the ARS-2 in that one can discriminate between winning and losing based upon ARS-2 scores. Results of the MANOVA indicated that outcome was related to ARS-2 scores and the descriptive discriminant analysis showed that precompetitive self-confidence was meaningful in this relationship. Results from this study varied from previous findings (Cox, Robb, & Russell, 1999) that found self-confidence as well as cognitive anxiety were meaningful in discriminating between winning and losing athletes. Yet, it is possible that the cognitive and somatic components as assessed by the ARS-2 were interpreted as facilitative by some athletes and deb ilitative by others, thus did not discriminate match outcome with these participants. This differential interpretation of anxiety was not considered in this study but has become important in determining relative contribution to explaining performance variance (Swain & Jones, 1996).

In this data set, the evidence suggests that one's precompetitive self-confidence was much more relevant to match outcome, with higher levels of precompetitive self-confidence associated with successful match outcome more than lower levels of precompetitive self-confidence. This supports the notion that athletes who indicate higher confidence perceive anxiety as more facilitative (Wiggins & Brustad, 1996). The mean ARS self-confidence score for winning outcome was 5.32, compared to a mean ARS self-confidence score of 4.99 for losing outcome. A previous criticism of the ARS-2 (Cox & Robb, 1999) was that some of the psychological constructs included in each statement seemed to cause confusion in respondents relative to whether the item was a cognitive item, a somatic item, or a self-confidence item. However, results from this investigation indicated that although the subscales showed low to moderate intercorrelations, athletes were able to distinguish among these subscales that was reflected in the main effect for match outcome. Based upon these findings, it appears that while the ARS-2 was able to discriminate between winning and losing athletes, the weak strength of cognitive and somatic anxiety subscales with match outcome may have been apparent due to intensity and direction issues (Cox, 1998).

It has been noted that some athletes view intensity responses to be much more facilitative than other athletes (Jones, Hanton, & Swain, 1994; Jones & Swain, 1995). Since the ARS-2 was derived from the CSAI-2, it is based upon the notion of measuring intensity of one's precompetitive state anxiety however, the current scale does not measure an athlete's interpretation of anxiety direction (debilitative or facilitative) (Edwards & Hardy, 1996). Edwards and Hardy (1996) used the CSAI-2 with netballers and found that self-confidence intensity was related to the direction of cognitive and somatic anxiety, and that as self-confidence increased, athletes perceived their anxiety to be more facilitative to their performance. It may have been that with cognitive and somatic anxiety in the current investigation, individual differences in intensity may have been dictated by whether athletes viewed their cognitive anxiety and somatic anxiety as debilitating or facilitating, thereby affecting the relationship between thes e subscales and subsequent outcome. In fact, it may be necessary to assess direction and intensity of anxiety regardless of the brevity of anxiety scale, to provide more accurate predictions of performance variance (Jones, 1995)

Recently, Wiggins and Brustad (1996) showed that athletes who had high self-confidence and low cognitive and somatic anxiety scores perceived anxiety as more facilitative and had higher pre-competitive expectations of performance results. Thus, the stronger relationship between ARS-self-confidence and outcome may have been due to the fact that athletes high in self-confidence perceived their cognitive and somatic anxiety as more facilitative, thus accounting for the nonsignificant relationship between cognitive anxiety, somatic anxiety, and match outcome. This would also support the view (Hardy, 1990) that high self-confidence may protect against the potentially debilitative effects of anxiety. By integrating a direction scale into the ARS-2, the construct validity of the scale could be supported if, for example, cognitive anxiety was rated as significantly more facilitative to performance by the "better" performers as measured by successful (win) match outcome (Jones, Swain, & Hardy, 1993). Based upon these observations, it is recommended that future construct validity studies examining the ARS-2 attempt to use a direction scale to examine whether both intensity and direction of cognitive anxiety, somatic anxiety, and self-confidence are more effective in explaining performance differences.

References

Cox, R.H. (1998). Sport psychology: Concepts and applications. (4th ed.) Dubuque: William C. Brown-McGraw-Hill.

Cox, R.H., Reed, C., & Robb, M. (1997). Comparative concurrent validity of the MRF-L and ARS competitive state anxiety ratings scales for individual sport athletes. Journal of Applied Sport Psychology 8, Supplement, S23.

Cox, R.H., & Robb, M. & Russell, W.D. (2000). Concurrent validity of the revised Anxiety Rating Scale. Journal of Sport Behavior, 23,4, 327-3 34.

Cox, R.H., Robb, M., & Russell, W.D. (1999). Construct validity of the revised anxiety rating scale (ARS-2). Manuscript submitted for publication. Journal of Sport Behavior

Cox, R.H., Russell, W.D., & Robb, M. (1998). Development of a CSAI-2 short form for assessing competitive state anxiety during and immediately prior to competition. Journal of Sport Behavior, 21, 30-40.

Cox, R.H., Russell, W.D., & Robb, M. (1999). Comparative concurrent validity of the MRF-L and ARS competitive state anxiety rating scales for volleyball and basketball. Journal of Sport Behavior, 22, 1-11.

Edwards, T., & Hardy, L. (1996). The interactive effects of intensity and direction of cognitive and somatic anxiety and self-confidence upon somatic performance. Journal of Sport and Exercise Psychology 18, 29 6-312.

Fenz, W.D., & Epstein, S. (1967). Gradients of physiological arousal in parachutists as a function of an approaching jump. Psychosomatic Medicine, 29, 33-51.

Fenz, W.D., & Jones, G.B. (1972). Individual differences in physiologic arousal and performance in sport parachutists. Psychosomatic Medicine, 34, 1-8.

Hardy, L. (1990). A catastrophe model of performance in sport. In G. Jones, & L. Hardy (Eds.), Stress and performance in sport (pp. 81-106). Chichester: Wiley.

Highlen, P.S., & Bennett, B.B. (1979). Psychological characteristics of successful and unsuccessful elite wrestlers: An exploratory study. Journal of Sport Psychology 1, 123-127.

Jones, G. (1995). More than just a game: Research developments and issues in competitive anxiety in sport. British Journal of Psychology 86,449-478.

Jones, J.G., Hanton, S., & Swain, A.B.J. (1994). Intensity and interpretation of anxiety symptoms in elite and non-elite performers. Personality and Individual Differences, 17,657-663.

Jones, G., & Swain, A. (1995). Predispositions to experience debilitative and facilitative anxiety in elite and nonelite performers. The Sport Psychologist, 9, 20l-211.

Jones, G., Swain, A., & Hardy, L. (1993). Intensity and direction dimensions of competitive state anxiety and relationships with performance. Journal of Sport Sciences, 11, 525-532.

Krane, V. (1994). The MRF as a measure of competitive state anxiety. The Sport Psychologist, 8, 189-202.

Mahoney, M.J., & Avener, M. (1977). Psychology of the elite athlete: An exploratory study. Cognitive Therapy and Research, 1, 135-141.

Martens, R., Vealey, R.S., & Burton, D. (1990). Competitive anxiety in sport. Champaign, IL: Human Kinetics.

Meyers, A. W., Cooke, C.J., Cullen, J., & Liles, L. (1979). Psychological aspects of athletic competitors: A replication across sports. Cognitive Therapy and Research, 3, 361-366.

Pedhazur, E.J. (1997). Multiple regression in behavioral research: Explanation and prediction (3rd edition). New York, NY: Harcourt Brace.

Scanlan, T.K. (1975). The effect of competition trait anxiety and success-failure on the perception of threat in a competitive situation. Unpublished doctoral dissertation, University of Illinois.

Sonstroem, R.J., Harlow, L.L., & Salisbury, K.S. (1993). Path analysis of a self-esteem model across a competitive swim season. Research Quarterly for Exercise and Sport, 64, 335-342.

Swain, A., & Jones, G. (1996). Explaining performance variance: The relative contribution of intensity and direction dimensions of competitive state anxiety. Anxiety, Stress, and Coping, 9, 1-18.

Thomas, J.R., & Nelson, J.K. (1996). Research methods in physical activity. (3rd edition). Champaign, IL: Human Kinetics.

Wiggins, M.S., & Brustad, R.J. (1996). Perception of anxiety and expectations of performance. Perceptual and Motor Skills, 83, 1071-1074.