performance and persistence of worriers and non-worriers following success and failure feedback

12
Performance and persistence of worriers and non-worriers following success and failure feedback Ted Thompson*, Kate Webber, Iain Montgomery School of Psychology, University of Tasmania, GPO Box 252-30, Hobart, 7001, Tasmania, Australia Received 30 May 2000; received in revised form 16 March 2001 Abstract Worry is a cognitive phenomenon that involves rumination over future events accompanied by feelings of anxiety. This study utilised a 2 (worry status: worrier vs. non-worrier) 2 (performance feedback: suc- cess vs. non-contingent failure) 2 (strategy advice: advice vs. none) between subjects factorial design to investigate the effects of worry on performance and persistence by manipulating performance feedback and strategy advice. n=48 worriers and n=48 non-worriers completed measures of state anxiety, cognitive interference, and perceptions of performance. Following an induction to make participants believed they had failed on a task, worriers showed poor performance, elevated anxiety, and greater cognitive inter- ference than non-worriers. No evidence of nonproductive persistence was evident for worries given strategy advice. The roles of anticipated difficulty levels, item solubility, problem solving confidence, and the per- ceived value of dogged persistence are discussed in assessing situational factors conducive to nonproductive persistence. # 2002 Elsevier Science Ltd. All rights reserved. Keywords: Worry; Cognitive interference; Performance feedback; Nonproductive persistence 1. Introduction Worry is a cognitive phenomenon that involves concern over future events, usually accom- panied by feelings of anxiety. The experience of worry involves intrusive, negative and disruptive thoughts which occur at a high emotional cost (Davey, 1994; Tallis & Eysenck, 1994). According to Borkovec and Inz (1990) it is this experience of intrusive negative cognitive activity that sepa- rates worriers from non-worriers. A frequently reported consequence of worry is impaired performance. For example, worriers have been found to display disrupted processing and slowed response latencies on categorisation tasks (Metzger, Miller, Cohen, Sofka, & Borkovec, 1990). These effects have been attributed to a 0191-8869/01/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved. PII: S0191-8869(01)00076-9 Personality and Individual Differences 33 (2002) 837–848 www.elsevier.com/locate/paid * Corresponding author. Tel.: +61-3-626-2887; fax: +61-3-6226-2883. E-mail address: [email protected] (T. Thompson).

Upload: ted-thompson

Post on 16-Sep-2016

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Performance and persistence of worriers and non-worriers following success and failure feedback

Performance and persistence of worriers and non-worriersfollowing success and failure feedback

Ted Thompson*, Kate Webber, Iain Montgomery

School of Psychology, University of Tasmania, GPO Box 252-30, Hobart, 7001, Tasmania, Australia

Received 30 May 2000; received in revised form 16 March 2001

Abstract

Worry is a cognitive phenomenon that involves rumination over future events accompanied by feelingsof anxiety. This study utilised a 2 (worry status: worrier vs. non-worrier) � 2 (performance feedback: suc-cess vs. non-contingent failure) � 2 (strategy advice: advice vs. none) between subjects factorial design toinvestigate the effects of worry on performance and persistence by manipulating performance feedback andstrategy advice. n=48 worriers and n=48 non-worriers completed measures of state anxiety, cognitiveinterference, and perceptions of performance. Following an induction to make participants believed theyhad failed on a task, worriers showed poor performance, elevated anxiety, and greater cognitive inter-ference than non-worriers. No evidence of nonproductive persistence was evident for worries given strategyadvice. The roles of anticipated difficulty levels, item solubility, problem solving confidence, and the per-ceived value of dogged persistence are discussed in assessing situational factors conducive to nonproductivepersistence. # 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Worry; Cognitive interference; Performance feedback; Nonproductive persistence

1. Introduction

Worry is a cognitive phenomenon that involves concern over future events, usually accom-panied by feelings of anxiety. The experience of worry involves intrusive, negative and disruptivethoughts which occur at a high emotional cost (Davey, 1994; Tallis & Eysenck, 1994). Accordingto Borkovec and Inz (1990) it is this experience of intrusive negative cognitive activity that sepa-rates worriers from non-worriers.A frequently reported consequence of worry is impaired performance. For example, worriers

have been found to display disrupted processing and slowed response latencies on categorisationtasks (Metzger, Miller, Cohen, Sofka, & Borkovec, 1990). These effects have been attributed to a

0191-8869/01/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.

PI I : S0191-8869(01 )00076-9

Personality and Individual Differences 33 (2002) 837–848

www.elsevier.com/locate/paid

* Corresponding author. Tel.: +61-3-626-2887; fax: +61-3-6226-2883.

E-mail address: [email protected] (T. Thompson).

Page 2: Performance and persistence of worriers and non-worriers following success and failure feedback

variety of factors, including a tendency on the part of worriers to overestimate the probability ofnegative events, negative thought processes including counteractive beliefs, appraisals andexpectations (MacLeod, Williams, & Bekerian, 1991), an inability to disengage from intrusivethoughts, elevated evidence requirements, and difficulties with ambiguous stimuli, such as aninability to distinguish soluble from insoluble problems (Tallis & Eysenck, 1994). Poor perfor-mance is also exacerbated by a tendency on the part of worriers to set high standards for self-evaluation and a tendency to overgeneralise a single failure to the whole of the self-concept (Flett& Blankstein, 1994). Such negative cognitions and self-expectations are associated with reducedproblem-solving confidence, delays in decision-making, and poor performance (e.g. Davis &Montgomery, 1997; Dugas, Letarte, Rheaume, & Freeston, 1995).In turn, low problem-solving confidence has been linked with high levels of concern over mis-

takes and self-doubt (Flett & Blankstein, 1994; Frost, Marten, Lahart, & Rosenblate, 1990).Collectively, these attributes contribute to reduced confidence in solving the task at hand, lowdecisional self-esteem, and procrastination (Borkovec, Ray, & Stober, 1998).Worriers appear to have particular difficulties when making decisions about ambiguous stimuli

(Davey, Hampton, Farrell, & Davidson, 1992). Mathews (1990) suggests that this difficulty maybe due to the fact that worriers overestimate potential risk factors in conditions of uncertainty,while Tallis, Eysenck, and Mathews (1991) propose that the difficulties worriers experience in thisregard is related to elevated evidence requirements for making a decision. On the other hand,MacLeod et al. (1991) suggest that worriers are likely to perceive an ambiguous problem asthreatening, causing a rise in anxiety. As anxiety levels rise, attentional resources are allocated tothe threat in order to reduce the attendant anxiety. This results in an increased number of intru-sive worrisome thoughts, directing attention away from the problem at hand, thus delayingdecision making processes, potentially impairing performance (MacLeod et al, 1991).An area that has been seldom researched is the relationship between worry and persistence.

While the beneficial effects of persistence in terms of performance are well documented (e.g.Janoff-Bulman & Brickman, 1982; Shrauger & Sorman, 1997; Tang, Lui, & Vermillion, 1987), forworriers, poor performance may be due to nonproductive persistence, brought about by persist-ing too long on tasks that are insoluble. This assumption is supported by evidence that worriersexperience difficulty detecting whether or not a problem is soluble (e.g. Tallis et al., 1991). Assuch, they are likely to be slower to disengage from insoluble tasks than non-worriers, therebyreducing performance (e.g. Davis & Montgomery, 1997).In addition, the elevated evidence requirements of worriers and their low decisional self-esteem

are likely to lengthen decision-making when working on tasks that involve uncertainty or ambi-guity, delaying disengagement. Finally, the interfering effects of anxiety and cognitive inter-ference, particularly in time-limited or evaluative situations, are likely to diminish worriers’attentional resources. Thus, in a situation in which worriers are given ambiguous informationconcerning task solubility they are likely to engage in nonproductive persistence.

1.1. The present study

On the above bases, the present study explored the effects of worry on performance and non-productive persistence following either success or noncontingent failure feedback. This was doneby manipulating performance feedback and task strategy advice.

838 T. Thompson et al. / Personality and Individual Differences 33 (2002) 837–848

Page 3: Performance and persistence of worriers and non-worriers following success and failure feedback

In order to pursue these investigations, participants were exposed to either success or non-con-tingent failure, after which their performance in solving 20 single-solution anagrams was assessed.We expected that noncontingent failure feedback would initiate intrusive, negative ruminationand increased anxiety for worriers relative to non-worriers. As a consequence, we expected thatworriers would perform poorly relative to non-worriers, indicated by increased mean latenciesand fewer anagrams solved overall. However, no such effect was expected following success. Onthis basis, a significant interaction between performance feedback and worry status (worriers vs.non-worriers) was expected, with these effects mediated by greater predicted cognitive interferenceand anxiety for worriers following exposure to noncontingent failure relative to non-worriers.Following the anagrams, participants were asked to solve 16 unicursal (tracing puzzle) tasks,

eight of which were soluble and eight of which were insoluble. Prior to commencing these tasks,participants were either provided with ambiguous information about task solubility, or noinformation about task solubility. This manipulation was intended to facilitate an assessment ofnonproductive persistence on the part of worriers and non-worriers.Based on the understanding that relative to non-worriers, worriers have less tolerance for

ambiguous information and elevated evidence requirements, and the assumption that exposure tononcontingent failure would initiate intrusive, negative rumination and increased anxiety forworriers relative to non-worriers, we expected worriers would have diminished attentionalresources available to focus on the task in hand and heed strategy advice. On these bases, weexpected that worriers pre-treated to noncontingent failure would show greater nonproductivepersistence than non-worriers, indicated by greater time spent on insoluble items, more attemptsmade at solving insoluble items, and a greater number of nodes traced for insoluble tasks.

2. Method

2.1. Experimental design

Two groups of students: worriers and non-worriers were randomly assigned to one of twoperformance feedback conditions: success versus noncontingent failure, and strategy advice:advice versus none. This rendered the experiment a 2 (worry status: worrier vs. non-worrier) � 2(performance feedback: success vs. non-contingent failure x 2 (strategy advice: advice vs. none)between subjects factorial design.

2.2. Participants

Participants were drawn from a sample of 400 undergraduate students enrolled in a variety ofdegree programs at the University of Tasmania who completed the Penn State Worry Ques-tionnaire (PSWQ: Meyer, Miller, Metzger, & Borkovec, 1990). Worriers were selected from par-ticipants with scores of 62 and above, and non-worriers from those with scores lower than 42(Molina & Borkovec, 1994). These cut-offs identified a potential pool of 53 worriers and 112 non-worriers. Individuals who agreed to participate in the study were randomly selected and assignedto experimental conditions by a person other than the experimenter. As a consequence, theexperimenter was blind to each participant’s group membership.

T. Thompson et al. / Personality and Individual Differences 33 (2002) 837–848 839

Page 4: Performance and persistence of worriers and non-worriers following success and failure feedback

2.3. Instruments

2.3.1. Pre-experimental measures

2.3.1.1. Penn State Worry Questionnaire. The PSWQ was used to allocate participants to experi-mental conditions. This questionnaire comprises 16 items that relate to the intensity, excessive-ness and uncontrollability of worry and its presence over time and across situations (Molina &Borkovec, 1994). Items are formatted on five-point scales with end-point designations Not at alltypical of me (1) and Very typical of me (5). The range of possible scores is from 16 to 80. ThePSWQ has established utility in identifying worried samples and individuals who are found tomeet the diagnostic criteria for Generalised Anxiety Disorder (GAD: Molina & Borkovec, 1994).Brown, Antony, and Barlow (1992) report a Cronbach alpha of 0.96 for 436 anxiety disorderpatients and 0.86 for GAD patients.

2.3.2. Experimental measuresOdd- and even-numbered items from the Spielberger et al. (1983) State-Trait Anxiety Inventory

(STAI) were used to create parallel forms as a means of assessing the impact of performancefeedback on individuals’ anxiety states before and after success and failure pretreatment. Spiel-berger et al. (1983) licence parallel forms based on odd- and even-numbered items. This optionwas followed in order to minimise any tendency on the part of participants to respond to items onthe second administration of the STAI in a manner which was consistent with their responses onthe first occasion. This was a genuine concern as the two administrations were completed within ashort period of time (10 min or so). Thompson and le Fevre (1999) report a KR-20 of 0.89 for thefirst administration of the STAI and 0.88 for the second administration. Possible scores on theSTAI ranged from 10 to 40.

2.3.2.1. The Sarason Cognitive Interference Scale. The Cognitive Interference Scale (Sarason,1978) comprises 11 items which assess the extent to which individuals experience intrusive, dis-tracting thoughts. Items include I thought about how often I got confused, and I thought about howpoorly I was doing. Items are formatted on five-point scales with end-point designations never (1)and very much (5). The range of possible scores on these items is from 11 to 55. A twelfth item,not included in the score total for this scale, provides a global report of cognitive interference byasking respondents to report the degree to which their mind wandered during the task they hadjust completed.

2.3.2.2. Performance perceptions and affective reactions following the simultaneous discriminationtask. Five individual items were specifically developed for this study in order to gather informationconcerning cognitive and affective reactions on the part of worriers and non-worriers followingsuccess and failure. Participants were asked to report how well they performed on the simultaneousdiscrimination task relative to their expectations, how happy they were about their performance,the extent to which they considered their performance to have been a total success versus totalfailure, the extent to which they were in control of their performance, and how satisfied they werewith their performance on the simultaneous discrimination task. These items were formatted onseven-point scales with end-point designations tailored to the requirements of each item.

840 T. Thompson et al. / Personality and Individual Differences 33 (2002) 837–848

Page 5: Performance and persistence of worriers and non-worriers following success and failure feedback

2.3.3. Cognitive tasks2.3.3.1. Simultaneous discrimination task. The success and failure feedback task consisted of amodification of the Levine (1966) simultaneous discrimination task presented on a computer.Four 10-trial, simultaneous discrimination tasks used by Thompson, Davidson, and Barber(1995) were used. Each problem began with the presentation of two figures. These figures con-sisted of two letters of the alphabet, A or T and varied in terms of four features: the letter itself (Aor T), the colour of the letter (black or white), the size of the letter (small or large), and the shapeof the border in which the letters were contained (square or circle). These attributes varied sys-tematically across the 10 displays used in the experiment.Participants were required to identify just one feature which was predetermined by the experi-

menter for each of the four trials (eg. black, square, small, the letter T). In the failure condition,feedback given by the experimenter took no regard of students’ responses. Instead, a fixedsequence of responses was given which varied for each of the four trials. Students in fact received50% reinforcement schedules (Correct or Incorrect) on all four problems. The reinforcementschedules, taken from Thompson et al. (1995), were as follows: (1) C-I-I-C-C-I-I-C-C-I; (2) I-C-I-C-C-I-C-I-C-I; (3) I-C-I-C-I-C-C-I-C-I; (4) C-C-I-C-I-I-C-I-C-I. As a consequence of this feed-back, students failed on all four trials. In the success condition, feedback corresponded withparticipants’ responses.

2.3.3.2. Anagrams. The criterion task, used to assess performance following failure on theunsolvable problems, consisted of 20 single-solution anagrams each with a median solution timeof 30 s, as determined by Tresselt and Mayzner (1966). All anagrams were disarranged in thesequence 5-3-1-2-4 and were individually printed on index cards 15�10 cm and presented tostudents on a datepad.

2.3.3.3. Unicursal tasks. Sixteen tracing puzzle (unicursal) tasks developed for the present study,based on those used by Feather (1961), were presented on computer and used to assess persis-tence. Participants were required to trace geometric figures in one continuous line by clicking onvertices using the mouse button without re-tracing any given line. The computer supplied feed-back in the form of a blue tick for a correctly traced task. Dependent measures recorded by thecomputer were time (in seconds) taken on each task, number of items solved, number of attemptsat each item, and the number of nodes connected.

2.4. Procedure

On arrival at the laboratory, students were informed that the intention of the experiment was todiscover whether peoples’ ability to discover one type of code was related to their ability to dis-cover another type of code. Students were further advised that they would be given three tasks tocomplete: a simultaneous discrimination task, followed by 20 solvable anagrams, followed in turnby 16 unicursal (tracing puzzle) tasks. Participants were also informed that they would be askedto complete items that assessed perceptions of performance and their reactions to the simulta-neous discrimination task.All participants then completed the first of two parallel forms of the STAI before commencing

the simultaneous discrimination task. The instructions given to students for the simultaneous

T. Thompson et al. / Personality and Individual Differences 33 (2002) 837–848 841

Page 6: Performance and persistence of worriers and non-worriers following success and failure feedback

discrimination task and the anagrams closely resembled those given by Thompson et al. (1995).Participants were informed that the purpose of the experiment was to examine peoples’ ability todiscover a code or underlying principle in order to solve a problem, and to establish whether theability to discover one type of code was related to the ability to discover another type of code.Following the simultaneous discrimination task, students completed the second form of theSTAI, and items assessing perceptions of performance and affective reactions.Participants next completed the 20 single-solution anagrams. They were informed that there

could be a pattern or a principle by which to solve the anagrams and that they were to be alloweda maximum of 100 s to solve each anagram. Upon completion of the anagrams, participantscompleted the Cognitive Interference Scale (Sarason, 1978).Following these measures, participants began the unicursal tasks. Participants were informed

that they could attempt any unicursal task as many times as they liked and move on to any itemas they wished. Participants were told they had approximately 15 min for the entire task, but thatthere were no restrictions on the time they spent on each figure. Participants were then eithergiven strategy advice or not. Participants given strategy advice were informed that the tasks var-ied greatly in difficulty, with some easy to solve, some more difficult, even insoluble. They werefurther informed that as a consequence it may not be wise to persist too long on problems thatthey found very difficult.The computer recorded the number of attempts at soluble and insoluble items, the length of

time spent on each problem, the number of attempts at any given item and the number of correctand incorrect traces overall. Upon completion of the unicursal tasks, participants were debriefed.Those in the failure condition were run through the success condition of the experiment, therebyreinstating any loss of self-esteem.

3. Results

3.1. Analysis strategy

The strategy followed in analysis was to submit each dependent variable to separate 2 (worrystatus: high vs. low) �2 (performance feedback: success vs. failure) ANOVAs. For dependentmeasures associated with the unicursal tasks, 2 (worry status: worrier vs. non-worrier) �2 (per-formance feedback: success vs. failure) �2 (strategy advice: advice vs. none) ANOVAs were used.Post hoc tests (Fisher PLSD) were used to test for significant differences between means where

appropriate. Table 1 shows means and standard deviations for worry status, as well as corre-sponding F-values. In all analyses, the alpha level was set at 0.05. There were no missing valuesfor any dependent measure.

3.1.1. Anxiety and cognitive interference3.1.1.1. STAI measures. STAI measures were administered prior to experimental instructions(STAI-1) and following the simultaneous discrimination task (STAI-2). As expected, a repeatedmeasures ANOVA revealed a significant interaction between STAI occasion (STAI-1, STAI-2)and performance feedback, with participants reporting greater anxiety at STAI-2 following fail-ure (M=21.94, S.D.=4.96) than they did following success (M=15.83, S.D.=4.67): F (1,

842 T. Thompson et al. / Personality and Individual Differences 33 (2002) 837–848

Page 7: Performance and persistence of worriers and non-worriers following success and failure feedback

92)=45.68, P<0.0001). A trend towards an interaction between STAI occasion, performancefeedback, and worry was also evident: F (1, 92)=3.40, P=0.069.Main effects were also evident for worry status and feedback condition (Ps <0.0001 in each

case), with worriers reporting greater state anxiety than non-worriers, and participants in thefailure condition reporting greater anxiety than participants in the success condition: STAI-1(failure), M=18.56, S.D.=4.46; STAI-2 (failure), M=21.94, S.D.=4.96; STAI-1 (success),M=17.04, S.D.=4.86; STAI-2 (success), M=15.83, S.D.=4.67. Finally, a significant main effectwas evident for STAI occasion (P <0.0019), with STAI-2 scores (M=18.88; S.D.=5.69) beinghigher than STAI-1 scores (M=17.80; S.D.=4.70).

3.1.1.2. Cognitive interference. For the 11-item Cognitive Interference Scale (Sarason, 1978),main effects were evident for worry status and performance feedback, with worriers reportinghigher cognitive interference than non-worriers: (P <0.0001), and participants in the failurecondition greater cognitive interference than those in the success condition (P <0.001; for failure,M=25.15, S.D.=7.48; while for success: M=20.98, S.D.=6.64. Following failure, worriers alsoreported higher levels of cognitive interference on the full scale measure of cognitive interferencethan non-worriers (P=0.0002).Findings for the single-item measure mirrored those for the scale measure, with worriers

reporting higher cognitive interference than non-worriers (P<0.0001), and participants in thefailure condition reporting greater cognitive interference than those in the success condition

Table 1Means and standard deviations (in parentheses) for repeat administrations of the STAI, cognitive interference, ana-gram performance, performance perceptions and affective reactions following the simultaneous discrimination task for

worriers and non-worriers

Worriers (n=48) Non-worriers (n=48) F-values for Worry status

STAI-1 19.77 (5.00) 15.83 (3.45) 20.20 ***STAI-2 21.31 (5.93) 16.46 (4.27) 21.77***

Cognitive interference11-Item scale 26.65 (7.67) 19.48 (4.90) 29.78***Single-item measure 2.44 (1.37) 1.500 (0.58) 19.11***

Anagram performanceMean latency 23.13 (19.29) 14.16 (13.24) 7.06**Failures to solve 2.83 (3.14) 1.35 (1.85) 7.92**

Performance perceptions and affective reactions following the simultaneous discrimination taskPerformance 3.92 (1.89) 3.31 (1.73) 2.67

Happy 4.500 (1.80) 3.54 (1.75) 7.00**Failure/success 4.10 (1.75) 3.56 (1.53) 2.60Control 3.69 (1.87) 3.42 (1.80) 0.52

Satisfaction 4.35 (1.85) 3.69 (1.84) 3.14

**P<0.01.***P<0.001.

T. Thompson et al. / Personality and Individual Differences 33 (2002) 837–848 843

Page 8: Performance and persistence of worriers and non-worriers following success and failure feedback

(P=0.0009; for failure, M=2.31, S.D.=1.26; while for success: M=1.63, S.D.=0.92). There wasalso a trend towards a significant interaction between worry status and performance feedback(P=0.082).

3.1.2. Performance and persistence data3.1.2.1. Anagrams. Performance on the anagrams was assessed in terms of mean latency and thenumber of unsolved anagrams. Significant interactions were evident for mean latency: F (1,92)=7.50, P=0.002; M=36.23, S.D.=16.4 (worriers, failure); M=10.03, S.D.=11.55 (worriers,success); M=18.49, S.D.=15.73 (non-worriers, failure); M=9.83, S.D.=8.46 (non-worriers,success); and failures to solve: F (1, 92)=7.50, P=0.007; M=4.67, S.D.=2.79 (worriers, failure),M=1.00, S.D.=2.30 (worriers, success); M=1.96, S.D.=2.37 (non-worriers, failure); M=0.75,S.D.=0.79 (non-worriers, success). In each case, these interactions arose by virtue of the fact thatworriers had higher mean latencies and greater failures to solve in the failure condition relative tothe success condition (Ps<0.0001 in each case), and higher mean latencies (P=0.0004) and fail-ures to solve (P=0.0007) in the failure condition relative to non-worriers. In addition, worriershad higher mean latencies (P=0.002) and a greater number of failures to solve (P=0.007) thannon-worriers.

3.1.2.2. Unicursal tasks. Persistence on the unicursal tasks was assessed in terms of (1) the totalnumber of tasks solved (2) the total time spent on soluble items, (3) the total time spent on inso-luble items, (4) total attempts at soluble items, (5) total attempts at insoluble items (6) the totalnumber of nodes traced for soluble items, (7) the total number of nodes traced for insolubleitems.A 2 (worry status: worriers vs. non-worriers) �2 (performance feedback: failure vs. success) �

(strategy advice: advice vs. none) between-subjects multivariate analysis of variance (MANOVA)was performed on the dependent measures listed above. As there were no significant interactions,separate 2 (worry status: worrier vs. non-worrier) �2 (feedback condition: success vs. non-con-tingent failure) �2 (strategy advice: advice vs. none) ANOVAs were performed for each depen-dent measure.Main effects for strategy advice were evident for number of tasks solved: F (1, 88)=18.82,

P<0.0001 (M=5.45, S.D.=2.24 vs. M=3.62, S.D.=1.82), time spent on soluble items: F (1,88)=6.86, P=0.017 (M=320.68, S.D.=88.37 vs. M=265.43, S.D.=114.52), number of attemptson soluble items: F (1, 88)=14.63, P=0003 (M=13.21, S.D.=6.62 vs. M=8.70, S.D.=4.64),and nodes traced for soluble items: F (1, 88)=22.24, P<0.0001 (M=133.00, S.D.=55.80 vs.M=86.54, S.D.=39.90). In each case, participants spent more time, attempted more items, andconnected more nodes following strategy advice than in the absence of strategy advice.For only one of the dependent measures was a main effect evident for insoluble items. Partici-

pants spent less time solving insoluble tasks when given strategy advice than when no strategyadvice was given: F (1, 88)=4.37, P=0.038 (M=622.70, S.D.=180.84 vs. M=701.47,S.D.=184.67).In addition, participants in the success condition spent less time on insoluble tasks than those in

the failure condition: F (1, 88)=8.05, P=0.006 (M=608.36, S.D.=127.26 vs. M=710.88,S.D.=218. 04). There was also a trend towards a main effect for performance feedback for totalnodes traced: F (1, 88)=3.09, P=0.083), with participants in the success condition tracing a

844 T. Thompson et al. / Personality and Individual Differences 33 (2002) 837–848

Page 9: Performance and persistence of worriers and non-worriers following success and failure feedback

greater number of nodes than participants in the failure condition (M=238.31, S.D.=102.11 vs.M=222.44, S.D.=100.45).

3.1.3. Performance perceptions, affective reactions, and attributions following the simultaneousdiscrimination taskFollowing the simultaneous discrimination task, participants completed items which assessed

performance relative to expectations, their degree of happiness with their performance, percep-tions of performance in terms of success versus failure, perceived control, and satisfaction. Forperceptions of performance relative to expectations, relative to non-worriers, worriers reportedthat they performed poorly relative to expectations (P=0.029). For assessments of how happyparticipats were with their performance, worriers also reported being less happy with their per-formance (P=0.0001), assessed their performance more in terms of failure than success(P=0.009), and reported less satisfaction with their performance (P=0.017) than non-worriers.For each of these dependent measures there were, predictably, main effects for performance

feedback. Relative to participants in the success condition, participants in the failure conditionperceived their performance as worse than their expectations: F (1, 92)=78.97, P<0.0001(M=4.83, S.D.=1.51 vs. M=2.40, S.D.=1.20), reported being less happy with their perfor-mance: F (1, 92)=125.03, P<0.0001 (M=5.40, S.D.=1.09 vs. M=2.65, S.D.=1.31), assessedtheir performance more in terms of failure than success: F (1, 92)=148.82, P<0.0001 (M=5.13,S.D.=1.21 vs. M=2.54, S.D.=0.82), reported less control: F (1, 92)=81.46, P<0.0001(M=4.79, S.D.=1.53 vs. M=2.31, S.D.=1.13), and less satisfaction with their performance: F(1, 92)=78.35, P<0.0001 (M=5.27, S.D.=1.25 vs. M=2.77, S.D.=1.51).

4. Discussion

One of the aims of this study was to test the assumption that worriers would show poorperformance following an experience of failure relative to non-worriers. Significant interactionsbetween worry status and performance feedback were expected for each of the dependentmeasures in question: mean latency and number of unsolved anagrams. These expectationswere based on several assumptions: (1) that relative to non-worriers, worriers would reporthigher levels of state anxiety at pretest and at post-test in the failure condition, and (2) thatworriers would report higher levels of cognitive interference than non-worriers following failurefeedback.

4.1. Performance, anxiety and cognitive interference

Both the expected differences in performances and the bases of these expectations were con-firmed. On the one hand, significant interactions between worry status and performance feedbackwere evident for each of the dependent measures in question. Following failure, worriers solvedfewer anagrams and took longer to solve the anagrams than non-worriers, while in the successcondition, the two groups were essentially undifferentiated.Correspondingly, predictions in relation to state anxiety and cognitive interference were sub-

stantiated. Relative to non-worriers, worriers reported higher anxiety at pretest, and higher

T. Thompson et al. / Personality and Individual Differences 33 (2002) 837–848 845

Page 10: Performance and persistence of worriers and non-worriers following success and failure feedback

anxiety following failure feedback, supporting previous research demonstrating links betweenworry and anxiety (e.g. Breakwell, Fife-Shaw, & Devereux, 1988). This pattern was repeated forcognitive interference, with worriers reporting higher cognitive interference relative to non-wor-riers following failure. This was true for both the full scale and single-item measures of cognitiveinterference, supporting findings from Borkovec and Inz (1990) that the experience of negative,intrusive thoughts separates worriers from non-worriers.

4.2. Nonproductive persistence

While participants in the success condition spent less time on insoluble tasks than those in thefailure condition, the prediction that worriers would show greater nonproductive persistence thannon-worriers was not substantiated. This prediction was based on reports by Tallis and Eysenck(1994) of elevated evidence requirements and difficulties with ambiguous stimuli, as well asreduced problem-solving confidence for worriers and non-worriers, and a diminished ability todiscriminate soluble from insoluble items (e.g. Davis & Montgomery, 1997; Dugas et al., 1995),which seemed conducive to nonproductive persistence. It was also based on assumed differencesbetween worriers and non-worriers in terms of cognitive interference and anxiety, substantiatedin this study.Worriers did not differ from non-worriers in terms of attempts at insoluble items, time spent on

insoluble items or total number of nodes traced for insoluble items. Nor were there performancedifferences between worriers and non-worriers on the unicursal tasks. While in the present studyworriers were apparently beset with high levels of anxiety and cognitive interference, theymimicked the behaviour of non-worriers across the two strategy advice conditions. Nevertheless,significant main effects were evident for strategy advice. Worriers and non-worriers spent moretime on insoluble items, made more attempts at these items, and traced more nodes when theywere given strategy advice than when no advice was given. Seemingly, they did so in order toascertain solubility. This activity had productive consequences, evident in the fact that partici-pants solved more unicursal tasks when they were given strategy advice than when they were not.Presumably, the indication that some of the items may be insoluble motivated greater productiveeffort on soluble items.

4.3. Implications and conclusions

Several possibilities need to be considered in the light of these results. On the one hand, it maybe that the findings from the present study can be depended upon in that non-worriers do notengage in nonproductive persistence. Against this, the assumed preconditions for nonproductivepersistence on the part of worriers were evident in the form of greater anxiety, greater cognitiveinterference, diminished perceptions of control, and perceptual biases, together with tangibleevidence of the consequences of these effects in terms of poorer performance on the part of wor-riers relative to non-worriers following failure. An alternative explanation is that the interpolatedexperiences of (a) working on the anagrams, and (b) completing items assessing perceptions ofperformance and affective reactions following the simultaneous discrimination task effectivelydiluted the negative effects of failure so that failure pretreatment did not impact on persistence inthe predicted ways.

846 T. Thompson et al. / Personality and Individual Differences 33 (2002) 837–848

Page 11: Performance and persistence of worriers and non-worriers following success and failure feedback

An additional possibility is that detection of nonproductive persistence among worriers requiresan alternative, more subtle advice manipulation which leaves the potential solubility of itemsopen to discovery in the process of problem solving. The assumption in the literature (e.g. Tallis& Eysenck, 1994) is that nonproductive persistence on the part of worriers is mediated by aninability to disengage from tasks, and an inability to distinguish between solvable and unsolvableproblems. As a consequence, they do not disengage from unsolvable problems as quickly as non-worriers, perhaps because they are less sensitive to contextual cues indicating solubility. Forworriers, it appears that nonproductive persistence requires a belief that items are potentiallysoluble, but that difficulty levels may vary. Where these conditions obtain, it is likely that lowproblem solving confidence and anxiety drive dogged persistence in disregard of contextual cuessuggesting item insolubility. These ideas invite investigation in further studies.

References

Bokovec, T. D., & Inz, J. (1990). The nature of worry in generalised anxiety disorder: a predominance of thoughtactivity. Behaviour, Research and Therapy, 31, 321–324.

Borkovec, T. D., Ray, W. J., & Stober, J. (1998). Worry: a cognitive phenomenon intimately linked to affective

physiological and interpersonal behavioural processes. Cognitive Therapy and Research, 22, 561–576.Breakwell, G. M., Fife-Schaw, C., & Devereux, J. B. (1988). The relationship of self-esteem and attributional style toyoung people’s worries. The Journal of Psychology, 122, 207–215.

Brown, T. A., Antony, M. M., & Barlow (1992). Psychometric properties of the Penn State Worry Questionnaire in a

clinical anxiety disorders sample. Behaviour, Research and Therapy, 30, 33–37.Davey, G. C. L. (1994). Pathological worrying as exacerbated problem-solving. In G. C. L. Davey, & F. Tallis (Eds.),Worrying: perspectives on theory, assessment and treatment (pp. 35–59). West Sussex: John Wiley .

Davey, G. C. L., Hampton, J., Farrell, J., & Davidson, S. (1992). Some characteristics of worrying: evidence for wor-rying and anxiety as separate constructs. Personality and Individual Differences, 13, 133–147.

Davis, M., & Montgomery, I. (1997). Ruminations on worry: issues related to the study of an elusive construct.

Behaviour Change, 14, 193–199.Dugas, M. J., Letarte, H., Rheaume, J., & Freeston, M. H. (1995). Worry and problem solving: evidence of a specificrelationship. Cognitive Therapy and Research, 19, 109–120.

Feather, N. (1961). After affects of periodic noise. In D. C. Glass, & J. E. Singer (Eds.), Urban stress (pp. 48–49). NewYork: Academic Press.

Flett, G. L., & Blankstein, K. R. (1994). Worry as a component of test anxiety: a multidimensional analysis. InG. C. L. Davey, & F. Tallis (Eds.), Worrying: perspectives on theory, assessment and treatment (pp. 35–59). West

Sussex: John Wiley.Frost, R. O., Marten, P., Lahart, C., & Rosenblate, R. (1990). The dimensions of perfectionism. Cognitive Therapy andResearch, 14, 449–468.

Janoff-Bulman, R., & Brickman, P. (1982). Expectations and what people learn from failure. In N. T. Feather (Ed.),Expectations and actions: expectancy value models in psychology. Hillsdale, NJ: Lawrence Erlbaum.

Levine, M. (1966). Hypothesis behavior by humans during discrimination learning. Journal of Experimental Psychol-

ogy, 71, 331–338.MacLeod, A. K., Williams, M. G., & Bekerian, D. A. (1991). Worry is reasonable: the role of expectations in pessi-mism about future personal events. Journal of Abnormal Psychology, 100, 478–486.

Mathews, A. (1990). Why worry? The cognitive function of anxiety. Behaviour, Research and Therapy, 28, 455–468.

Metzger, R., Miller, M. L., Cohen, M., Sofka, M., & Borkovec, T. D. (1990). Worry changes decision-making. Theeffect of negative thoughts on cognitive processing. Journal of Clinical Psychology, 46, 78–88.

Meyer, T., Miller, M., Metzger, R., & Borkovec, T. D. (1990). Development and validation of the Penn State Worry

Questionnaire. Behaviour Research and Therapy, 28, 487–495.

T. Thompson et al. / Personality and Individual Differences 33 (2002) 837–848 847

Page 12: Performance and persistence of worriers and non-worriers following success and failure feedback

Molina, S., & Borkovec, T. D. (1994). The Penn State Worry Questionnaire: psychometric properties and associated

characteristics. In G. C. L. Davey, & F. Tallis (Eds.), Worrying: perspectives on theory, assessment and treatment(pp. 265–283). West Sussex: John Wiley.

Sarason, I. G. (1978). The Test Anxiety Scale: concept and research. Stress and Anxiety, 5, 193–216.

Shrauger, J. S., & Sorman, P. B. (1977). Self-evaluation, initial success and failure, and improvement as determinantsof persistence. Journal of Clinical Psychology, 45, 784–795.

Spielberger, C. D., Gorsuch, R. L., Lushene, R., Vagg, P. R., & Jacobs, G. A. (1983). State-trait anxiety inventory for

adults. California: Consulting Psychologists Press.Tallis, F., & Eysenck, M. (1994). Worry: mechanisms and modulating influences. Behavioural and Cognitive Psy-chotherapy, 22, 37–56.

Tallis, F., Eysenck, & Matthews, A. (1991). Elevated evidence requirements and worry. Personality and Individual

Differences, 13, 161–168.Tang, T. L., Lui, H., & Vermillion, J. R. (1987). Effects of self-esteem and task labels (difficult vs. easy) on intrinsicmotivation, goal setting and task performance. Journal of General Psychology, 114, 249–262.

Thompson, T., Davidson, J. A., & Barber, J. G. (1995). Self-worth protection in achievement motivation: performanceeffects and attributional behaviour. Journal of Educational Psychology, 87, 598–610.

Thompson, T., & le Fevre, C. (1999). Implications of manipulating anticipatory attributions on the strategy use of

defensive pessimists and strategic optimists. Personality and Individual Differences, 26, 887–904.Tresselt, M., & Mayzner, M. (1966). Normative solution times for a sample of 134 solution words and 378 associatedanagrams. Psychonomic Monograph Supplements, 1, 293–298.

848 T. Thompson et al. / Personality and Individual Differences 33 (2002) 837–848