Вы находитесь на странице: 1из 22

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0268-6902.

htm

MAJ 25,2

Time pressure, task complexity, and audit effectiveness


Anthony R. Bowrin
Saginaw Valley State University, Midland, Michigan, USA, and

160
Received 20 April 2009 Revised 26 September 2009 Accepted 2 October 2009

James King II
Formerly of School of Accountancy, Southern Illinois University, Carbondale, Illinois, USA
Abstract
Purpose The purpose of this paper is to examine the relationships among time pressure (TP), task complexity (TC), and audit effectiveness (AE). It is motivated by the conicting results reported in prior TP studies. Design/methodology/approach The research hypotheses are developed using McGraths interactional model of the Yerkes-Dodson Law. Data are collected using a two-treatment eld experiment involving 63 public accountants. Findings The results show a negative, interactional relationship among TP, TC, and AE. Research limitations/implications The rst limitation concerns the non-random procedure used to recruit public accounting rms and auditors. Second, there is the less than perfect operationalization of the TC construct. Practical implications First, the ndings suggest that public accounting rms may need to resist the urge to reduce the time allowed for performing compliance tests, and provide training to improve the detection rate for all type of compliance deviations. Second, the fact that the rate of change in AE, in response to changes in TP, is different for the two audit tasks studied, suggests that it may not be appropriate for audit planners to assume a uniform TP effect across the various tasks involved in an audit. This insight has implication for the trade-offs between the lower direct audit costs associated with tighter time budgets, and possible increases in audit risk associated with lower AE. Originality/value Two unique aspects of this paper are the operationalization of TP as a continuous random variable and the use of z-scores to standardize the AE measure. Keywords Time-based management, Task analysis, Auditing Paper type Research paper

1. Introduction Time budgets are a pervasive and important managerial tool in many contemporary accounting settings because they: . provide a basis for estimating and controlling costs; . promote timely task completion, and performance evaluation; and . provide evidence of compliance with the rst audit eldwork standard, proper audit planning and supervision (Bailey, 1997).
Managerial Auditing Journal Vol. 25 No. 2, 2010 pp. 160-181 q Emerald Group Publishing Limited 0268-6902 DOI 10.1108/02686901011008963

The time budgets have also been associated with auditors job-related stress and behaviors that may reduce audit quality, increase the likelihood of audit failure, and thereby increase the litigation exposure of certied public accounting (CPA) rms. Several researchers have provided evidence that CPAs perceive that time budgets are

becoming unrealistically tight and are negatively affecting audit quality (Azad, 1994; Solomon and Brown, 1992; Pierce and Sweeney, 2004). This situation of increasingly tight budgets, and the time pressure (TP) they generate, has been attributed to increasing price competition among CPA rms since the early 1980s and the changing business focus by CPA rms and their clients towards process re-engineering and right-sizing (DeZoort and Lord, 1997; Sweeney and Pierce, 2004). The topic remains relevant today as rms try to meet the markets audit quality expectations in todays highly regulated and litigious post-Enron business environment. Considerable research has been conducted into the effects of TP on task effectiveness with mixed results. Some research has reported an inverse relationship between TP and task effectiveness (McDaniel, 1990; Kobbeltvedt et al., 2005). Conversely, other research has reported either no relationship (Huchingson, 1973; Topi et al., 2005), or a positive relationship between these variables (Farmer et al., 1984). To complicate matters further, other researchers have proposed and found an inverted U-shaped relationship between TP and task effectiveness (Choo, 1995). These conicting results have been attributed to: . the use of a variety of possibly inconsistent measures to operationalize TP and task effectiveness; . failure to incorporate time limits (budgets) from the full range of conditions found in practice; and . the omission of potential moderator variables from prior studies. This study seeks to clarify these conicting ndings by examining two questions: (1) What is the nature of the relationship between TP and audit effectiveness (AE)? (2) Whether the effect of TP on AE is contingent on the complexity of the task being performed. These questions are examined using a fuller range of TP conditions found in practice than was done in prior studies. A total of 63 auditors from several medium-sized and large public accounting rms participated in an experiment that involved the performance of two independent audit tasks (low- and medium-task complexity (TC)). TP was operationalized between subjects as a continuous random variable and AE was measured as a correct answer to the assigned tasks. The remainder of the paper is organized as follows. In the next section, we develop the theoretical arguments on which the research hypotheses are based. Section 3 describes the experimental method. Section 4 describes the statistical analyses performed and presents the result. The paper concludes with a discussion of the results and their implications for practice and research. 2. Theory and hypotheses 2.1 TP and performance Time (budget) pressure is dened as ones perception regarding her/his ability to perform an assigned task within a specied time limit, given that timely task completion is an important dimension of task performance (DeZoort and Lord, 1997). TP has been shown to inuence task performance both in auditing (Choo, 1995;

Pressure, complexity, and audit 161

MAJ 25,2

162

DeZoort and Lord, 1997; Tan et al., 2002) and psychology studies (Chu and Spires, 2001; Kelly and McGrath, 1985; Walczyk and Grifth-Ross, 2006). To facilitate the study, we adopted the Yerkes and Dodson (1908) Law, which assumes an inverted U-shaped (quadratic) relationship between stress (e.g. TP)[1] and task effectiveness. This decision was based on three factors. First, the Yerkes and Dodson (1908) Law is the most comprehensive of the available TP task effectiveness models. Second, it is consistent with an observed tendency among researchers to assume a quadratic relationship between TP and task effectiveness (Choo, 1995; Kelly and McGrath, 1983). Third, the use of this model is consistent with the conclusion of DeZoort and Lord (1997) that the results of experimental studies in auditing have largely supported the Yerkes-Dodson Law. While the Yerkes-Dodson Law serves as the theoretical base for our discussion, it has a aw. Yerkes-Dodson describes, rather than explains, the relationship between stress (e.g. TP) and task effectiveness (Eysenck, 1973). For this reason McGraths (1976) interactional model, which elucidates the nature of the Yerkes-Dodson relationship, is used to examine the TP task performance relationship and develop the research hypotheses. McGrath suggested that stress may inuence task effectiveness through the interaction of two factors; arousal and goal difculty[2]. Arousal refers to ones state or degree of readiness to act or react (Stratton and Haynes, 1988), while goal difculty refers to the amount of effort required to meet a performance target (Huber, 1985). McGrath (1976) asserted that this interaction derives from two facts. First, both arousal and goal difculty are positively related to stress, and second, arousal is positively related to task effectiveness, while goal difculty is negatively related to task effectiveness[3]. According to McGrath (1976), the combined effect of arousal and goal difculty produces the inverted U-shaped relationship between stress and task performance noted by Yerkes and Dodson (1908). According to McGrath, when stress is low, task effectiveness will be relatively poor because arousal is low (Frankenhaeuser and Gardell, 1976). At moderate levels of stress, both arousal and goal difculty are expected to be at optimal levels, promoting greater on-task attention (Eysenck, 1982), and enhanced task effectiveness (McGrath, 1976). Finally, when stress is high the absolute level of task effectiveness will decline, because high arousal and goal difculty are likely to cause auditors to feel anxious[4] (Bavelas and Lee, 1978; Humphreys and Revelle, 1984). Several researchers have suggested that elevated anxiety may lead to faulty information processing and retard task performance (Ashton, 1990; Chu and Spires, 2001; de Dreu, 2003; Eysenck, 1984; Maule and Hockey, 1993). To date, the empirical evidence regarding the Yerkes-Dodson Law has been equivocal. Several studies have provided evidence supportive of the law (Choo, 1995; Kelly and Margheim, 1990; Kelly and McGrath, 1983; Smith et al., 1997; Sweeney and Pierce, 2004). For instance, Choo (1995) reported an inverted U-shaped relationship between TP and the accuracy of auditors judgments regarding the appropriate sampling strategy for the conrmation of demand deposits. However, other studies have provided evidence that conict with the law (Caballer et al., 2005; Isenberg, 1981; McDaniel, 1990; Kobbeltvedt et al., 2005; Topi et al., 2005). In spite of these conicting ndings, organizational studies researchers generally continue to regard the stress (TP) task effectiveness relationship as being inverted U-shaped in nature. As a result, we present the following hypothesis: HI. In auditing contexts an inverted U-shaped relationship exists between TP and AE.

2.2 TP, TC and performance A major corollary of the Yerkes-Dodson Law states that the relationship between stress (e.g. TP) and task effectiveness is moderated by TC[5]. TC is dened as the manner in which task elements are interrelated and the extent to which task requirements are specied (Simon, 1973). In this study, TC is operationalized by asking auditors to perform two independent tasks that have been shown to exhibit different levels of complexity but are usually performed by auditors with similar levels of experience (Abdolmohammadi, 1999). Two similar tasks were used by Tan and Kao (1999) to operationalize the construct. Both Yerkes and Dodson (1908) and McGrath (1976) asserted that the optimum amount of stress is inversely related to TC. This inverse relationship has been attributed to the fact that superior performance on a complex task requires the exertion of qualitatively greater effort and attention than on a simple task (Campbell, 1988; Kahneman, 1973). Other researchers have argued that tasks which require conguration (complex tasks) tend to require more information processing than tasks, which do not require conguration (simple tasks) (Devine and Kozlowski, 1995; Wood, 1986; Wood et al., 1990). Continuing this logic, one may expect that as tasks become increasingly complex, the resulting qualitative and quantitative demands placed on auditors may begin to exceed their capabilities, creating overload (DeShon et al., 1996; Marsh and Ahn, 2006; Simnett, 1996). In turn, this overload may lead to the sub-optimal use of acquired information and knowledge, and lower task effectiveness (Bonner, 1994; Devine and Kozlowski, 1995). Several studies have provided empirical support for these arguments (Baumeister and Showers, 1986; Bonner, 1994; Chang et al., 1997; Wongpinunwatana, 2003). For instance, Bonner (1994) concluded that, in general, increases in TC are associated with decreases in judgment quality and increases in systematic judgment bias. Similarly, Wongpinunwatana (2003) reported an inverse relationship between TC and the accuracy and certainty of auditors judgments. When the effects of TC are superimposed on the TP-AE relationship, an interactive effect is likely, especially in light of the pre-existing level of arousal and goal difculty (Ashton, 1990; McGrath, 1976). To the extent that TP and TC affect the same mechanism, a more dramatic impairment of task effectiveness may occur when they are jointly imposed (Broadbent, 1971; Forgarty et al., 2000). Since simple tasks tend to make qualitatively smaller demands on the capabilities of decision makers than complex tasks (Wood, 1986), simple tasks are likely to experience smaller performance impairments at all levels of TP[6]. The above arguments regarding the relationship among TP, TC, and AE suggest the following hypothesis: HII. TC moderates the relationship between TP and AE. AE on the more complex (ratio analysis task) will be more adversely affected by increasing TP than that on the relatively simple (compliance testing) task.

Pressure, complexity, and audit 163

3. Experimental design A eld experiment was constructed using a two-treatment, factorial design (Kirk, 1995) to test the research hypotheses. It is depicted in Table I. This design enhances the studys internal validity by facilitating tight control over TP and TC.

MAJ 25,2

164

3.1 Variables 3.1.1 Independent variables. TP was conceptualized as the stringency of the time budget imposed on auditors for task completion. It was operationalized, between subjects, as a continuous random variable with levels selected from the distribution of task completion times developed during a pilot study[7]. An independent time (budget) pressure distribution was established for each task (Smith et al., 1982). The TP distribution for the ratio analysis and compliance-testing tasks were, respectively, mean 21.81 minutes (SD 5.92), and 30.56 minutes (SD 6.27). The distributions were, statistically, signicantly different from each other (t 4.06, p , 0.001, h 2 0.21). The actual time budgets were selected using a stratied, random sampling procedure[8]. The second independent variable, TC, was operationalized at two levels within subjects. Each auditor performed both the ratio analysis and the compliance-testing tasks. Task order was reversed after each administration of the experiment to control for possible order effects. The ratio analysis task required auditors to provide as many plausible explanations for uctuations in the experimental companys three-ratio nancial prole as was possible in the time allowed. Auditors were told that the uctuations were due to a single cause. They were also told that:
[. . .] to be plausible a misstatement or error must be capable of causing an overstatement of net income and either an overstatement of quick assets or an understatement of current liabilities[9].

The specic error seeded in the ratio analysis task was an unrecorded inventory purchase. The compliance-testing task required auditors to nd seeded errors in the experimental companys cash disbursement cycle. The material clearly described the clients control system and indicated that the purpose of the task was to determine whether the established procedures were followed in processing each transaction. Five different deviations were intentionally seeded in the seven voucher packets (voucher and supporting documents) selected for testing. The deviations were: (1) a difference between prices per the purchase order and prices per the invoice; (2) a difference between the remittance address on the invoice and the payee address on the copy of the check; (3) an extension error on the purchase order and invoice; (4) payment of shipping charges on an order with freight on board (FOB) destination terms; and (5) a missing external manager initial on the disbursement verication stamp. No voucher packet had more than one deviation. None of the deviations concerned a direct, material exception, although some did concern small monetary misstatements. Therefore, the materiality of the deviations should not bias the results.
Low TC Table I. High TP Low TP Higher AE Moderate TC Lower AE

3.2 Control variable accountability Accountability, dened as auditors expectation that they will be required to explain and justify their judgments to others (Tetlock, 1985), is a highly salient factor in the audit environment (Glover, 1997; Tan and Kao, 1999). It has been shown to systematically impact auditors decision-making processes and judgments (Peecher and Kleinmuntz, 1991; Messier and Quilliam, 1992) especially when coupled with high knowledge as is the case in this study (Tan et al., 2002). To control for this factor, and to enhance the face validity of the experimental tasks, subjects were instructed that they might be required to explain and justify their responses to the researcher. To increase the salience of this procedure, auditors were required to write their names on the answer sheet for each task and at the completion of the tasks 25 percent of the subjects were interviewed by the researcher. This procedure has been used extensively in the auditing literature with good results (Chang et al., 1997). Subjects were assured that responses to all portions of the task would only be used for the purposes of the experiment and that results would only be presented in aggregate form. All other potential nuisance variables were controlled by randomization. 3.3 Dependent variable AE was operationally dened as the extent to which the auditor achieved the stated objective. 3.3.1 Scoring: ratio analysis task. Auditors were instructed to list plausible explanations for the observed changes in the three-ratio nancial prole of the focal company. Two alternative procedures were used to score the results. Under the rst method, each answer sheet was reviewed to see whether the comments satised the requirements specied for plausibility. Comments that met this requirement (caused an increase in net income and either an increase in quick assets or a decrease in current liabilities) were counted as correct responses and each was given a score of 1. An auditors overall performance was measured as the sum of all correct explanations provided[10]. Overall, 20 unique plausible explanations were provided for uctuations in the three-ratio nancial prole. The second scoring method started with the same procedure as the rst, but adjusted for the number of incorrect explanations provided by each auditor. To facilitate this procedure, each explanation that did not meet the plausibility criteria was counted as incorrect and given a score of 2 1. Under this method, each auditors overall performance was measured as the sum of all correct explanations minus all incorrect explanations[11]. 3.3.2 Scoring: compliance-testing task. For the compliance-testing task, auditors were to have noted any exceptions found while testing the disbursement documents. Performance was measured as the sum of all deviations detected for each auditor[12]. The performance scores for each task were converted to z-scores to facilitate the direct comparison of AE between the two experimental tasks (Moore, 1985). 3.4 Subjects A total of 63 auditors participated in the research. Approximately, half of the subjects were recruited from rms in the Midwest USA (n 33), while the remaining 30 auditors were recruited from the Caribbean[13]. Overall, almost 70 percent of the subjects were from Big-5 rms[14]. Table II provides a demographic prole of the participants. A review of Table II reveals that the US and English-speaking Caribbean

Pressure, complexity, and audit 165

MAJ 25,2
M Time performed task Ratio analysis Compliance-testing Age Gender Professional certication Firm type Job title Staff auditor Senior auditor Supervisor Manager Total 4.94 7.12 25.03 Male 18 Yes 21 Big 5 22 N 24 7 1 1 33

USA SD 4.22 3.94 3.75 Female 15 No 12 Regional 6 M

English-speaking Caribbean SD 3.80 1.82 7.07 Female 18 No 11 International 9

166

Local 5

Table II. Demographic prole of auditors

7.87 10.27 29.83 Male 12 Yes 19 Big 5 21 N 6 13 5 6 30

samples are comparable on all but one dimension, task experience. US participants had less experience (number of times performed task) with both tasks than their counterparts from the English-speaking Caribbean (t (61) $ 2.88, p # 0.01, h 2 $ 0.12). These differences in task-specic experience did not affect the experimental results. When task-specic experience was introduced into the regression model as a control variable, it was not statistically signicant in either the full or reduced models. Descriptive statistics for AE by region are presented in Table III. A 2 2 ANOVA (geographical region TC) was conducted to examine whether AE on each of the assigned tasks varied by geographic region. The ANOVA results (F (1,121) , 1.0, p . 0.45) indicated that AE did not differ as a function of geographical region. Therefore, the experimental data from both regions were combined to test the research hypotheses. 3.5 Experimental materials Case materials used in this study were adapted from Libby (1985) and Waggoner (1986) and were modied with the assistance of two audit managers from a Big 5
USA Na 32 33 English-speaking Caribbean M SD N Median 4.49 2.83 2.28 0.99 30 30 4.00 3.00

M Ratio analysis task Compliance-testing task 4.37 3.07

SD 2.22 1.26

Median 4.50 3.00

Table III. Descriptive statistics for AE

Notes: aRather than perform the ratio analysis task, one auditor wrote a lengthy explanation of why the uctuations in the three-ratio nancial prole could not be material. Therefore, the sample size for the following analyses of the ratio analysis task is 62. Dependent variable: number of plausible explanations listed for the ratio analysis task; number of seeded errors detected for the compliancetesting task

public accounting rm. All materials went through three pilot tests. The rst and second pilot tests sought to identify and remove typographical and clerical errors, clarify vague statements, help establish preliminary estimates of the time required to review the case material, perform the assigned tasks, and to verify the completeness and validity of the case material. The nal pilot test, which involved 16 staff auditors from a Big-5 rm, was used to establish the TP distribution, assess whether the experimental materials were perceived as effective for the tasks, and determine whether the experience level of auditors was appropriate for the assigned tasks[15]. These subjects indicated that the materials contained no major ambiguities, and clearly communicated both the task context and requirements[16, 17]. 3.6 Manipulation checks 3.6.1 Time pressure. The validity of the TP manipulation was assessed in two ways. First, auditors were asked to rate the amount of time pressure you experienced during the performance of the assigned task. Auditors responded using a six-point, Likert-type scale with responses ranging from 1 extremely low time pressure to 6 extremely high time pressure. For both tasks, responses to this item exhibited a statistically signicant, negative correlation with the imposed time budget (r $ 2 0.48, p # 0.01, n 63). This indicates that, on average, auditors perceived increasing TP stress as the time budget was reduced. As a second measure of the effectiveness of the TP manipulation, auditors were asked whether: . the task required me to work very fast; and . the time allotted was sufcient for proper performance of the task. Auditors responded to each item using a six-point, Likert-type scale with responses ranging form 1 strongly disagree to 6 strongly agree. The responses to these items were used to create a summated variable summated time pressure (r $ 0.73). Next, a Pearson correlation coefcient was computed between the summated variable and the actual time budget allowed. For both tasks there was a statistically signicant negative relationship between the actual time budget and the summated TP variable (ratio analysis: r 2 0.62, p 0.01; compliance testing: r 2 0.42, p 0.01). These ndings suggest that auditors operating under more liberal (demanding) time budgets, reported feeling less (more) rushed during the performance of the assigned tasks. 3.6.2 Task complexity. A dependent-samples t-test was conducted to evaluate auditors perception of the complexity of the thought processes required to complete each task. Auditors mean thought process complexity rating for the compliance-testing task was 2.27 (SD 0.63) on a ve-point scale, while that on the ratio analysis task was 3.21 (SD 0.73), with higher values representing greater complexity. The results indicate that auditors perceived the ratio analysis task as requiring signicantly more complex thought processes than the compliance-testing task (t (61) 27.71, p , 0.001, h 2 0.49). 4. Data analysis and results 4.1 Rationale for using hierarchical regression analysis Hierarchical regression analysis was used to test the research hypotheses (Cohen, 1978). This procedure examines the incremental power of the quadratic

Pressure, complexity, and audit 167

MAJ 25,2

168

terms (TP squared (TP2) and TP2*TC), after the effects of the linear terms have been partialed out (Green et al., 1997). This procedure was used rather than the traditional regression procedure because of the conditional nature of the relationships being examines and the highly correlated nature of several of the independent variables. One of the primary benets of hierarchical regression analysis relative to traditional regression analysis is its ability to more efciently and effectively partial out the unique amount of the total variance in Y (performance) explained by the full set of K independent variables, R2 y:12...k , that is attributable to either one independent variable or one subset of independent variables, when the independent variables being examined are signicantly correlated (Cohen and Cohen, 1983). This benet stems from the fact that in hierarchical multiple regression analysis the independent variables are entered into the model in an order specied in advance by the researcher after considering their causal priority, the risk of spurious relationships, their research relevance and the structural properties of the factors being examined. This pre-specication is the only reliable basis on which the partitioning of variance can occur with correlated variables. Therefore, the technique is well suited to the examination of conditional relationships such as whether TP2, TC2, and TP2*TC2 predict auditors task performance, after controlling for the linear effects of TP and TC which have been shown to be important in prior studies. This procedure has been used in several accounting studies that focused on conditional relationships involving correlated independent variables (Ryan, 2004; Rahim and Minors, 2003; Stanley et al., 2003). For instance, Rahim and Minors (2003) used hierarchical regression analysis to examine the effects of several dimensions of emotional intelligence on managers concern for quality and the problem-solving behaviors of auditors. It was also used by Stanley et al. (2003) to examine issues of justice and commitment in the federal budget decision-making process. 4.2 Diagnostics Examination of the correlation matrix for the variables in the study (Table IV) revealed that the TP and TC variables were highly correlated with higher-order (quadratic) terms of which they are part[18]. While the presence of highly correlated independent variables is usually a cause for concern in regression analysis, it does not present a problem in the current analyses because of the hierarchical procedure used (Cohen and Cohen, 1983). All pertinent assumptions of regression analysis (normality, homoscedasticity, and independence of errors) were assessed for the regression variate after each model was estimated. Examination of the residuals revealed no major violations of these assumptions.
Variables 1 2 3 4 5 6 Mean SD

Table IV. Descriptive statistics (mean, standard deviation, and Pearson correlation) for variables in the study

1. Z-score (task performance variable) 2. TP 2 0.42 * 3. TP2 0.37 * 20.98 * 4. TC NA 0.40 * 20.39 * 5. TP TC 2 0.24 * 20.02 0.08 2 0.85 * 6. TP2 TC 0.33 * 20.17 * 0.09 0.68 * 2 0.96 * Notes: *p , 0.05 (two-tailed test); n 125

0.00 1.00 23.12 9.76 629.19 490.15 0.50 0.50 9.51 11.31 217.43 324.14

4.3 Hypothesis testing 4.3.1 H1. The H1 stated that there is a quadratic relationship between TP and AE. In Step 1 of the analysis, the full model comprising TP, TP2, and TC, the interaction between TP and TC (TP*TC), and the interaction between TP2 and TC (TP2*TC) was regressed on the AE variable (AE)[19]. The results of Step 1 are presented in Panel A of Table V. The full model (AE b0 TP TP2 TC TP*TC TP2*TC) was statistically signicant (R 2 0.26, adjusted R 2 0.23, F (5, 119) 8.49, p , 0.00). In Step 2, the quadratic variables (TP2 and TP2*TC) were removed from the regression equation to evaluate whether they accounted for variance in AE over and above that explained by the linear terms. The results of Step 2 are presented in Panel B of Table V and were used to evaluate H1. The restricted model (AE b0 TP TC TP*TC) was signicant (R 2 0.25, adjusted R 2 0.23, F (3, 121) 13.59, p , 0.00). However, the quadratic terms (TP2 and TP2*TC) did not account for a signicant proportion of the variance in AE after controlling for the linear terms (R 2 change 20.01, F (2, 123) 0.87, p 0.42). Based on these results, H1 is not supported. The relationship between TP and AE does not appear to be inverted U-shaped in nature[20]. 4.3.2 H2. In Step 3 of the hierarchical regression analysis, the linear interaction variable (TP*TC) was removed to evaluate whether it accounted for variance in AE
Variables A. Step 1 TP TP2 TC TP * TC TP2 * TC Total R 2 0.263 Total R 2 adjusted 0.232 F (5, 119) 8.46, p , 0.001 B. Step 2 TP TC TP * TC Change in R 2 2 0.011 F (2, 123) 0.87, p 0.421 Total R 2 0.252 Total R 2 (adjusted) 0.233 F (3, 121) 13.59, p , 0.001 C. Step 3 TP TC Change in R 2 2 0.04 F (1, 123) 6.44, p 0.012 Total R 2 0.212 Total R 2 (adjusted) 0.199 F (2, 122) 16.44, p , 0.001 B 2 0.210 2 0.004 0.838 0.078 0.002 SE B 0.173 0.003 1.218 0.106 0.002

Pressure, complexity, and audit 169

b
2 2.055 2 2.015 0.423 1.159 1.438

2 0.013 2 0.611 2 0.045

0.027 0.434 0.018

0.130 2 0.308 2 0.664 *

2 0.051 0.402

0.009 0.174

2 0.503 0.203

Notes: *p , 0.05; dependent variable: standardized performance scores; n 125

Table V. Summary of hierarchical regression analysis

MAJ 25,2

170

over and above that explained by the non-interaction terms. The results of Step 3 are presented in Panel C of Table V and were used to evaluate H2 which stated that AE on the more complex (ratio analysis) task would be more adversely affected by increasing TP than AE on the less complex (compliance-testing) task. The restricted model (AE b0 TP TC) was signicant (R 2 0.21, adjusted 2 R 0.20, F (2, 122) 16.44, p , 0.00). Additionally, the results indicate that the interaction term (TP*TC) accounted for a signicant proportion of the variance in AE after controlling for the non-interaction terms (R 2 change 2 0.04, F (1, 123) 6.44, p 0.01). Based on these results the H2 is supported. The relationship between TP and AE seems to be moderated by TC. Therefore, the Step 2 model seems to provide the best description of the TP, TC, and AE relationship. An examination of the Step 2 regression equation (i.e. AE b0 2 0:0132TP2 0:611TC 2 0:045TP*TC) reveals that only the TP*TC variables were statistically signicant (t (121) # 2 2.54, p , 0.05). As noted above, these results suggest that there is an interactive (linear) relationship between TP and AE. 4.4 Supplementary analyses 4.4.1 Ratio analysis task. Auditors listed 270 plausible explanations for the observed uctuation in the three-ratio nancial prole. These responses were categorized based on Libbys (1985) original list of 20 explanations, and input from three Big 5 audit managers. They were then classied into one of ve major transaction cycles (Arens and Loebbecke, 1994). See Table VI for a summary of the plausible explanations. The seeded error, an unrecorded purchase of inventory, was likely to be detected if any of six of Libbys 20 explanations was pursued by a subject[21]. Overall, 44 of the 62[22] auditors (71 percent) listed at least one of these high-likelihood explanations. As shown in Table VII, the percentage of auditors listing at least one high-likelihood explanation for the seeded error declined as TP increased. About 85, 71, and 64 percent of auditors listed at least one high-likelihood explanation under the low-, moderate- and high-TP conditions, respectively. However, two-way contingency table analysis (x 2 (2, N 62) 1.76, p 0.41), indicates that the proportion of auditors listing at least one high-likelihood explanation was not statistically different across the three TP conditions. 4.4.2 Compliance-testing task. The overall detection rate for the ve seeded deviations was 61 percent. This rate indicates that auditors were not completely accurate, and that non-sampling error was not negligible as usually assumed by practitioners. The detection rates for each of the ve seeded errors are summarized in Table VIII. As shown in Table VIII, the non-detection of deviations varied with the type of deviation tested. Across all TP conditions, authorization and computational errors were much more likely to be detected than errors involving descriptive agreement. 5. Discussion, implications and limitations 5.1 Discussion and implications This study examined two research hypotheses: H1. Whether there is an inverted U-shaped relationship between TP and AE. H2. Whether the relationship between TP and AE is moderated by TC.

Transaction cycle Sales and collection 1. Sales overstated (e.g. ctitious sales; recorded at incorrect (too high) prices; next periods sales recorded in the current period; sales discounts understated) 2. Bad debt expense and allowance for doubtful accounts not recorded or under-recorded 3. Goods returned by customers in the current period not recorded or recorded in the next period 4. Liabilities improperly classied as revenue (e.g. unearned revenue) 5. Receipts on accounts receivable improperly recorded as sales 6. Supplier wrote-off disputed outstanding balance as uncollectible Acquisition and payment 7. Expenses (accrued, general, and miscellaneous) and payables not recorded or understated 8. Marketable securities not written-down to market value 9. Purchase discounts overstated 10. Expense item(s) improperly capitalized Transaction cycle 11. Improper recording of unrealized gains on held-to-maturity securities 12. Unrecorded contingent liabilities 13. Increase in the proportion of other income in current period 14. Taxes and associated liabilities understated (e.g. too low rate; claim inappropriate exemptions/deductions) Inventory and warehousing 15. Inventory improperly relieved (e.g. using FIFO costs) 16. Physical inventory count/valuation overstated (e.g. inventory not adjusted for slowmoving/obsolete items) 17. Sales recorded but inventory (cost of goods sold) not recorded or recorded at incorrect (too low) prices 18. Purchases not recorded or understated (e.g. use of incorrect (too low) prices) Capital acquisition and repayment 19. Interest not accrued on long-term debt Payroll and personnel 20. Payroll taxes understated

Frequency

Pressure, complexity, and audit 171

57 41 15 05 04 03 38 20 14 10 02 02 02 01 20 11 10 09 04 02 Table VI. List of plausible explanations for ratio uctuations by transaction cycle

Time pressure High (6-16 minutes) Moderate (16.01-28 minutes) Low (28.01-35 minutes) Total

Auditor listed high-likelihood explanation Yes No No. % No. % 16 17 11 44 64 71 85 100 09 07 02 18 36 29 15 100

Total 25 24 13 62 Table VII. Two-way contingency table

Only the H2 was supported. The results of the regression analyses indicate that TC moderates the relationship between TP and AE. While the ndings are not totally consistent with the arguments presented in Section 2, they nonetheless enhance our understanding of the relationship between TP and AE. The negative regression coefcient for the TP*TC interaction term (B 2 0.045) is the difference between the change in mean AE for auditors performing the ratio

MAJ 25,2
Error type Authorization 1. Missing external managers authorization initials 2. Difference between price per purchase order and price per invoice paid without required approval Average Computational 3. Freight charges paid on FOB destination shipment 4. Extension error on purchase order Average Descriptive agreement 5. Difference between remittance address on invoice and payee address on check Average

Percentage of auditors detecting 86 68 77 78 51 64.5 24 24

172

Table VIII. Error detection rates for compliance-testing task

analysis task and those performing the compliance-testing task, caused by a one minute change in the time budget. This nding suggests that AE on the ratio analysis task was more adversely affected by increasing TP than AE on the compliance-testing task. This nding is consistent with the assertion of DeZoort and Lord (1997) and Solomon and Brown (1992) that the magnitude of the effect of TP on AE may be contingent on the nature of the task being performed and generalizes their ndings to auditors in the Caribbean. This moderating effect of TC on the TP AE relationship may be due to the greater qualitative complexity of the ratio analysis task, compared to the compliance-testing task (Abdolmohammadi, 1999). This argument is supported by the nding that auditors perceived the thought processes required for the ratio analysis task (mean 3.21, SD 0.73) to be more complex than that required for the compliance-testing task (mean 2.27, SD 0.63). In performing the ratio analysis task, auditors had to rst congure the task, (i.e. recognize relevant cues, develop strategies, and search for alternatives) a step that was not required to the same extent on the compliance-testing task. The ratio analysis task, therefore, involved more complex information processing than the compliance-testing task (Huber, 1985), which by itself, may have acted as a stressor (DeShon et al., 1996). When the stressful effect of this more complex ratio analysis task is combined with the pre-existing stress created by TP, auditors may have experienced greater stress at all levels of TP for the ratio analysis task, than for the compliance-testing task. Since auditors were operating at higher stress at all levels of TP on the ratio analysis task, any increase in stress (due to TP) probably had a larger negative impact on their ability to perform the necessary cognitive activity, relative to that on the less complex compliance-testing task. As a result, AE on the ratio analysis task appears to have been more adversely affected by increasing TP than on the compliance-testing task. The failure to nd the hypothesized inverted U-shaped relationship between TP and AE may be explained in at least two ways. First, the relationship may not have been present in the audit context examined. The moderate level of arousal created by auditors need to justify their decision in their normal high-accountability auditing environment (Chang et al., 1997; Libby and Lipe, 1992) may have prevented auditors

arousal levels from falling to the very low-levels frequently associated with low stress (low TP) conditions. According to this thesis, the overall moderate level of arousal and low level of goal difculty associated with low TP in the auditing setting is likely to be favorable for AE. This argument is consistent with Yates (1990) suggestion that the negative performance consequences of low-stress conditions predicted by the Yerkes-Dodson Law may be mitigated by arousal emanating from other sources (e.g. the task environment). Given a moderate arousal level and a low TP, auditors probably made more/deeper referrals to the case material and thereby obtained a fairly good understanding of the task and its context. It is also likely that under the low-TP condition auditors were able to conduct more iterations of their search/thought processes. The combined effect of these factors would explain the relatively high level of AE achieved under the low-TP conditions. As TP approached a moderate level, the expected negative effect of goal difculty on AE may have begun to dominate the predicted positive effect of increasing arousal (due to TP and accountability) on AE. Auditors understanding of the task may also have been lower under moderate TP than under low TP due to their having less time to assimilate the case materials. Additionally, auditors may not have been able to perform the same number or quality of search/thought process iterations possible under the low-TP conditions (de Dreu, 2003)[23]. This would explain the nding that AE was lower under the moderate TP conditions than under the low-TP conditions. As TP was increased above the moderate level of stringency, the proposed negative effect of goal difculty on AE may have increasingly dominated the expected positive arousal AE relationship. This would explain the nding that AE was lower under the high-TP conditions than under either the low or moderate TP conditions. A second possible explanation for the failure to nd the inverted U-shaped relationship is the high correlations between the linear and quadratic TP variables (r $ 0.98, p 0.05). These high correlations suggest that the amount of unique variation in both quadratic variables was probably very small and of little practical signicance (Cohen and Cohen, 1983). Based on these arguments rms may need to nd ways to minimize the levels of other (non-TP) stressors in the audit environment to improve AE. For instance, they may target the inherent complexity of different audit task by providing decision aids (e.g. expert systems; summaries of previous assignments for similar clients including key issues, ndings and conclusions; lists of cues associates with different patterns of relationships), all of which have the potential to reduce task conguration time. Additionally, rms may track the performance of auditors on various task types and use that information to facilitate task assignment and team composition decisions. The fact that the rate of change in AE, in response to changes in TP, was different for the two audit tasks studied, suggests that it may not be appropriate for audit planners to assume a uniform TP effect across the various tasks involved in an audit. Instead, auditors may need to develop individual estimates of the effects of changing time budgets for the major audit tasks. These estimates may then be used to make more-informed, cost-benet decisions regarding possible trade-offs between the potentially lower direct audit costs associated with tighter time budgets, and possible increases in audit risk and litigation exposure associated with lower AE.

Pressure, complexity, and audit 173

MAJ 25,2

174

For the compliance-testing task it was found that all subjected were able to detect at least one of the ve seeded deviations. However, the detection rate for seeded errors was signicantly less that 100 percent and varied with the type of compliance deviation examined and the stringency of the time budget allowed. These ndings are consistent with those reported by Waggoner (1986), and are contrary to auditors usual assumption that non-sampling risk due to non-detection is negligible. The low-overall detection rate for the ve seeded deviations (61 percent) is probably a reection of the stringency of the imposed time constraints. This assertion is supported by the nding (not reported) that the overall detection rate was 90 percent for the authorization deviations compared to only 50 percent for computational and descriptive agreement deviations under the low-TP condition. It suggests that accounting rms need to be aware of the differing time intensities of various simple tasks and adjust time budgets accordingly. The nding that the deviation detection rate varied with the type of deviation examined may be explained by the amount of effort needed for error detection. The detection rate was highest for authorization errors that are generally relatively easy to verify. The auditor need only check for evidence that the transaction represented by the document was approved, usually the initial or signature of the reviewer. Other types of deviations, such as those involving descriptive agreement, which had the lowest detection rate in the current study, generally require several switches among source documents, and may require more time and effort from auditors. These ndings suggest that auditors may need to reassess this assumption with a view to incorporating specic; non-zero estimates of non-sampling risk in their determination of sample sizes for substantive audit procedures. Alternatively, public accounting rms may need to resist the urge to reduce the time allowed for performing compliance tests, and provide training to improve the detection rate for all type of deviations. These strategies are more attractive in the post-Enron business environment where accounting rms are mandated to periodically review and report on the control systems of their clients. 5.2 Limitations of the study This study has several limitations, most of which concern the generalizability of the results. The rst limitation stems from the non-random procedure used to recruit public accounting rms and auditors. It is possible that subjects from other rms and/or ofces would have performed differently. A second limitation relates to the articial nature of the experimental setting. In their everyday work setting auditors are likely to face multiple stressors simultaneously. In this study, attempts were made to control such distracting stimuli and the range of strategies auditors could use to cope with TP stress, such as requesting additional time or resources, or performing the tasks on their own time. While such controls may have enhanced the internal validity of the study, they may have compromised the external validity of the ndings. The potential negative effect of this limitation on the ndings presented above is reduced by the fact that such strategies tend to eventually lose their effectiveness. A third limitation concerns the less than perfect operationalization of the TC construct. By using two different tasks we ran the risk of confounding differences in TC with difference on other dimensions of the tasks such as the knowledge required to

complete the tasks. To minimize the effect of such differences a special effort was made to ensure that auditors participating in the study had the competencies needed to effectively perform both tasks and had done so in practice. 5.3 Future research The ndings of this study suggest at least one avenue for future research. The generalizability of the research could be improved by using subjects from a more diverse group of public accounting rms, and a greater number of ofces from each rm. In addition to improving the generalizability of future research, such a sample will provide a more powerful statistical test of the quadratic and moderating effects examined in this study. It may also be interesting and informative to examine whether other variables that have been suggested as moderators/mediators of the TP AE relationship (such as locus of control), have an effect on the relationship in the auditing context.
Notes 1. Yerkes and Dodson (1908) used the concept of arousal (due to stress) in the presentation of their ideas. Since Bronner (1982, p. 12) described TP as the most frequently cited form of [decision making] stress, it seems reasonable to assume that the arguments presented by Yerkes and Dodson (1908) and in the general stress literature apply to TP. This assumption is supported by the fact that TP has frequently been used to study the stress task effectiveness relationship (Choo, 1995). 2. This interpretation of McGraths (1976) concept of task difculty is consistent with Christensen-Szalanskis (1980) operationalization of the term. 3. While on the surface this assumption seems to conict with the well-documented positive linear relationship between goal difculty and performance, that is not necessarily the case. Several researchers have attributed the positive inuence of goal difculty on task performance to its motivational (arousal) properties (Garland, 1983; Locke et al., 1981; Moussa, 1996). Therefore, if this arousal effect is partialed-out or otherwise controlled, one may observe the negative goal difculty effectiveness relationship suggested by McGrath (1976). 4. The anxiety construct is most often used to describe an unpleasant emotional state or condition which is characterized by subjective feelings of tension, apprehension, and worry; and by the activation of the autonomic nervous system (Spielberger, 1972, p. 482). 5. Yerkes and Dodson (1908) used the concept of task difculty rather than TC in their arguments. However, because the concepts of task difculty and TC have been used interchangeably in the organization studies literature (Chang et al., 1997; Huber, 1985), and the auditing profession has tended to focus on TC rather than task difculty (Abdolmohammadi, 1999; Bonner, 1994), we use the concept of TC throughout this paper. 6. At rst glance this expectation seems at odds with Tan et al. (2002) nding that task performance only declined with increasing TC when there was either high knowledge and low accountability or low knowledge and high accountability. The apparent conict is resolved when one recognizes that it is the presence of varying levels of TP that produces the effect, a factor that was held constant at very low levels by Tan et al. (2002). 7. This is intended to facilitate the objective of considering a wider range of time budgets than would be feasible using the more conventional categorical operationalization. The objective is rooted in the suggestion in prior studies that the failure to detect an inverted U-shaped relationship between TP and AE may be due to the failure to capture the full range of

Pressure, complexity, and audit 175

MAJ 25,2

TP conditions. Given this objective and the constraint of securing an appropriate sample size, the between subjects implementation was necessary. 8. About 20 percent of the time constraints were selected from each of four, one standard deviation segments of the range of values that are within ^ 2 standard deviations of the mean time taken to complete each task during the pilot study. To increase the likelihood of effectively capturing the tight TP condition, the remaining 20 percent of the time constraints was selected from the range of 2 2.5 to 2 2.0 standard deviations below the mean task completion time for each task. Next, each segment of the TP range was divided into ten-second intervals and these points were assigned consecutive numbers to facilitate the randomization process. 9. The plausibility criteria were included to provide cues to assist auditors in the generation of explanations. Their inclusion is justied by the nding of Hirst and Koonce (1996) that auditors routinely receive such cues from clients and colleagues and use them to perform analytical procedures. There is no a priori reason to believe that the plausibility cue would have a negative effect on auditors performance. 10. The primary researcher coded auditors responses to the experimental tasks based on criteria established with the assistance of three audit managers from two Big 5 accounting rms. 11. This scoring procedure was adopted to recognize the potentially negative effect of implausible explanations on AE as they make it less likely that seeded errors will be detected at all or in a timely manner. 12. Only two auditors listed incorrect deviations on the CT task. Therefore, it was not necessary to adjustment for incorrect responses on this task. 13. This mix of subjects was intended to shed light on whether the TP, TC, and AE relationship generalizes across countries. The Caribbean was chosen because of the dearth of research on, and the authors familiarity with, that region. 14. Because of the preponderance of auditors from Big 5-rms in the sample, the analysis was run with rm type (Big 5 versus Other) as a control variable. The results were identical to those reported. 15. Assessment of these factors was important because time budgets were manipulated to induce TP stress. It was thus vital that other sources of stress unrelated to TP (e.g. unfamiliarity with the tasks or faulty guidance) be controlled. 16. Prior to conducting the study in the English-speaking Caribbean, an audit partner from a Big 5 accounting rm operating inTrinidad and Tobago and two audit seniors employed by the Trinidad and Tobago ofce of an international accounting rm reviewed the experimental material and performed both tasks. All three individuals completed the tasks within ^ 0.5 standard deviations of the mean completion times for the US pilot study and indicated that the material was complete and clear. 17. Monetary incentives were not offered to subjects because such incentives were not in keeping with the policies of several participating audit rms. 18. Diagnostic procedures were also conducted to check for violations of the assumptions of regression analysis for the individual variables. None of the relevant assumptions (normality and homoscedasticity) were violated. 19. The analyses reported here are based on the use of the Z-score of the number of plausible explanations provided by auditors on the ratio analysis task, and the Z-score of the number of seeded errors detected on the compliance-testing task, as the dependent variable and the inverse of the actual time allowed as the TP variable. The hypotheses were also tested using

176

auditors perception of TP as the independent variable, and their task performance, as the dependent variable. The results of these analyses were substantially the same as those reported. 20. The small size of the absolute change in explained variance (0.01), also suggests that the change may not be practically signicant. 21. This conclusion is based on the idea that auditors use the perceived explanations of unexpected deviations as hypotheses to determine the accounts to be tested (Hirst and Koonce, 1996). Only the six high-likelihood explanations would have necessarily led auditors to test the accounts that contained the seeded error. 22. One auditor did not attempt the ratio analysis task. 23. This suggestion is supported by two facts. First, after completing their initial reading of the experimental material, many auditors inquired whether they would have access to the material while performing the task. Second, almost all auditors referred back to the experimental material while performing the tasks.

Pressure, complexity, and audit 177

References Abdolmohammadi, M. (1999), A comprehensive taxonomy of audit task structure, professional rank, and decision aids for behavioral research, Behavioral Research in Accounting, Vol. 11, pp. 51-82. Arens, A.A. and Loebbecke, J.K. (1994), Auditing: An Integrated Approach, 6th ed., Prentice-Hall, Englewood Cliffs, NJ. Ashton, R. (1990), Pressure and performance in accounting decision settings: paradoxical effects of incentives, feedback, and justication, Journal of Accounting Research, pp. 148-80 (supplement). Azad, A.N. (1994), Time budget pressure and ltering of time practices in internal auditing: a survey, Managerial Auditing Journal, Vol. 9, pp. 17-25. Bailey, J.M. (1997), Miller: GAAS Guide, Harcourt Brace Professional Publishing, San Diego, CA. Baumeister, R. and Showers, C.J. (1986), A review of paradoxical performance effects: choking under pressure in sports and mental tests, European Journal of Social Psychology, Vol. 16, pp. 361-83. Bavelas, J. and Lee, E.S. (1978), Effects of goal level on performance: a trade-off of quantity and quality, Canadian Journal of Psychology, Vol. 32, pp. 219-40. Bonner, S.E. (1994), A model of the effect of audit task complexity, Accounting, Organizations & Society, Vol. 19 No. 3, pp. 213-34. Broadbent, D.E. (1971), Decision and Stress, Academic Press, London. Bronner, R. (1982), Decision Making under Time Pressure, Lexington Books, Lexington, MA. Caballer, A., Garcia, F. and Peiro, J.-M. (2005), Affective responses to work process and outcomes in virtual teams: effects of communication media and time pressure, Journal of Managerial Psychology, Vol. 20 Nos 3/4, pp. 245-60. Campbell, D.J. (1988), Task complexity: a review and analysis, Academy of Management Review, Vol. 13 No. 1, pp. 40-52. Chang, C.J., Ho, J.L. and Liao, W.M. (1997), The effect of justication, task complexity and experience/training on problem-solving performance, Behavioral Research in Accounting, Vol. 9, pp. 98-122 (supplement).

MAJ 25,2

178

Choo, F. (1995), Auditors judgment performance under stress: a test of the relationship predicted by three theoretical models, Journal of Accounting, Auditing & Finance, Vol. 10 No. 3, pp. 611-41. Christensen-Szalanski, J. (1980), A further examination of the selection of problem-solving strategies: the effects of deadlines and analytical aptitudes, Organizational Behavior and Human Performance, Vol. 25, pp. 107-22. Chu, P.C. and Spires, E.E. (2001), Does time constraint on users negate the efcacy of decision support systems?, Organizational Behavior and Human Decision Processes, Vol. 85, pp. 226-49. Cohen, J. (1978), Partialed products are interactions; partialed powers are curved components, Psychological Bulletin, Vol. 85 No. 4, pp. 858-66. Cohen, J. and Cohen, P. (1983), Applied Regression/Correlation Analysis for the Behavioral Sciences, 3rd ed., Lawrence Erlbaum Associates, Hillsdale, NJ. de Dreu, C.K.W. (2003), Time pressure and closing of the mind in negotiations, Organizational Behavior and Human Decision Processes, Vol. 91, pp. 280-95. DeShon, R.P., Brown, K.G. and Greenis, J.L. (1996), Does self-regulation require cognitive resources? Evaluation of resource allocation models of goal setting, Journal of Applied Psychology, Vol. 81 No. 5, pp. 595-608. Devine, D.J. and Kozlowski, S.W.J. (1995), Domain-specic knowledge and task characteristics in decision making, Organizational Behavior and Human Decision Processes, Vol. 64, pp. 294-306. DeZoort, F.T. and Lord, A.T. (1997), A review and synthesis of pressure effects research in accounting, Journal of Accounting Literature, Vol. 16, pp. 28-85. Eysenck, H.J. (1973), Personality, learning, and anxiety, in Eysenck, H.J. (Ed.), Handbook of Abnormal Psychology, 2nd ed., Robert R. Knapp, San Diego, CA, pp. 390-419. Eysenck, M.W. (1982), Attention and Arousal: Cognition and Performance, Springer, Berlin. Eysenck, M.W. (1984), A Handbook of Cognitive Psychology, Lawrence Erlbaum Associates, London. Farmer, E.W., Hunter, J. and Belyavin, A.J. (1984), Performance under time constraint: the role of personality, Perceptual and Motor Skills, Vol. 59, pp. 875-84. Fogarty, T.J., Singh, J., Rhoads, G. and Moore, R. (2000), Antecedents and consequences of burnout in accounting: beyond the role stress model, Behavioral Research in Accounting, Vol. 12, pp. 31-67. Frankenhaeuser, M. and Gardell, B. (1976), Underload and overload in working life: outline of a multidisciplinary approach, Journal of Human Stress, Vol. 2 No. 3, pp. 35-46. Garland, H. (1983), Inuence of ability, assigned goals, and normative information on personal goals and performance: a challenge to the goal attainability assumption, Journal of Applied Psychology, Vol. 68 No. 1, pp. 20-30. Glover, S.M. (1997), The inuence of time pressure and accountability on auditors processing of non-diagnostic information, Journal of Accounting Research, Vol. 35 No. 2, pp. 213-26. Green, S.B., Salkind, N.J. and Akey, T.M. (1997), Using SPSS for Windows, Prentice-Hall, Englewood Cliffs, NJ. Hirst, D.E. and Koonce, L. (1996), Audit analytical procedures: a eld investigation, Contemporary Accounting Research, Vol. 13 No. 2, pp. 457-86. Huber, V. (1985), Effects of task difculty, goal setting and strategy on performance of a heuristic task, Journal of Applied Psychology, Vol. 70 No. 3, pp. 492-504.

Huchingson, R.D. (1973), Effects of time constraints on closed-loop tracking performance and change in heart rate, Perceptual and Motor Skills, Vol. 36, pp. 1195-8. Humphreys, M.S. and Revelle, W. (1984), Personality, motivation and performance: a theory of the relationship between individual differences and information processing, Psychological Review, Vol. 91, pp. 153-84. Isenberg, D. (1981), Some effects of time pressure on vertical structure and decision making accuracy in small groups, Organizational Behavior and Human Performance, Vol. 27, pp. 119-34. Kahneman, D. (1973), Attention and Effort, Prentice-Hall, Englewood Cliffs, NJ. Kelly, J. and McGrath, J. (1983), Task load, time limit and production rates: does work expand to ll the time available?, Report No. SPOT 83-1, The Social Psychology of Time Research Program Psychology Department, University of Illinois, Urbana, IL. Kelly, J. and McGrath, J. (1985), Effects of time limits and task type on task performance and interaction of four-person groups, Journal of Personality and Social Psychology, Vol. 49, pp. 395-407. Kelly, T. and Margheim, L. (1990), The impact of time budget pressure, personality and leadership variables on dysfunctional behavior, Auditing: A Journal of Practice & Theory, Vol. 9 No. 2, pp. 21-41. Kirk, R.E. (1995), Experimental Design: Procedures for the Behavioral Sciences, Brooks/Cole, Pacic Grove, CA. Kobbeltvedt, T., Brun, W. and Laberg, J.C. (2005), Cognitive processes in planning and judgments under sleep deprivation and time pressure, Organizational Behaviour and Human Decision Processes, Vol. 98, pp. 1-14. Libby, R. (1985), Availability and the generation of hypotheses in analytical review, Journal of Accounting Research, Vol. 23 No. 2, pp. 648-67. Libby, R. and Lipe, M. (1992), Incentives, effort, and the cognitive processes involved in accounting-related judgments, Journal of Accounting Research, Vol. 30, pp. 249-73. Locke, E.A., Shaw, K.N., Saari, L.M. and Latham, G.P. (1981), Goal setting and task performance: 1969-1980, Psychological Bulletin, Vol. 90, pp. 125-52. McDaniel, L.S. (1990), The effects of time pressure and audit program structure on audit performance, Journal of Accounting Research, Vol. 28 No. 2, pp. 267-85. McGrath, J.E. (1976), Stress and behavior in organizations, in Dunnette, M.D. (Ed.), Handbook of Industrial and Organizational Psychology, Rand McNally College Publishing, Chicago, IL, pp. 1351-95. Marsh, J.K. and Ahn, W. (2006), Order effects in contingency learning: the role of task complexity, Memory & Cognition, Vol. 34 No. 3, pp. 568-76. Maule, A.J. and Hockey, G.R. (1993), State, stress, and time pressure, in Svenson, O. and Maule, A.J. (Eds), Time Pressure and Stress in Human Judgment and Decision Making, Plenum Press, New York, NY, pp. 83-101. Messier, W.F. and Quilliam, W.C. (1992), The effect of accountability on judgment: development of hypotheses for auditing, Auditing: A Journal of Practice & Theory, Vol. 11, pp. 123-38 (supplement). Moore, D.S. (1985), Statistics: Concepts and Controversies, 2nd ed., W.H. Freeman and Company, New York, NY. Moussa, F.M. (1996), Determinants and process of the choice of goal difculty, Group & Organization Management, Vol. 21 No. 4, pp. 414-38.

Pressure, complexity, and audit 179

MAJ 25,2

180

Peecher, M.E. and Kleinmuntz, D.N. (1991), Discussion of experimental evidence on the effects of accountability on auditor judgments, Auditing: A Journal of Practice & Theory, Vol. 10, pp. 108-13 (supplement). Pierce, B. and Sweeney, B. (2004), Cost-quality conicts in audit rms: an empirical investigation, European Accounting Review, Vol. 13 No. 3, pp. 415-41. Rahim, M.A. and Minors, P. (2003), Effects of emotional intelligence on concern for quality and problem solving, Managerial Auditing Journal, Vol. 18 Nos 1/2, pp. 150-5. Ryan, J.J. (2004), Moral reasoning as a determinant of organizational citizenship behaviors: a study in the public accounting profession, Journal of Business Ethics, Vol. 33 No. 3, pp. 233-44. Simnett, R. (1996), The effect of information selection, information processing and task complexity on predictive accuracy of auditors, Accounting, Organizations and Society, Vol. 21 Nos 7/8, pp. 699-719. Simon, H. (1973), The structure of ill-structured problems, Articial Intelligence, Vol. 4, pp. 181-201. Smith, C.A.P., Arnold, V. and Sutton, S.G. (1997), The effect of time pressure on decision-making for choice and judgment tasks, Accounting & Business Review, Vol. 4 No. 2, pp. 365-83. Smith, J.F., Mitchell, T.R. and Beach, L.R. (1982), A cost-benet mechanism for selecting problem-solving strategies: some extensions and empirical tests, Organizational Behavior and Human Performance, Vol. 29, pp. 370-96. Solomon, I. and Brown, C. (1992), Auditors judgments and decisions under time pressure: an illustration and agenda for research, in Srivastava, R.P. (Ed.), Proceedings of the 1992 Deloitte & Touche/University of Kansas Symposium on Auditing Problems, University of Kansas, Lawrence, KS, pp. 73-91. Spielberger, C.D. (1972), Conceptual and methodological issues in anxiety research, in Spielberger, C.D. (Ed.), Anxiety: Current Trends in Theory and Research, Vol. 2, Academic Press, London, pp. 481-93. Stanley, A.B., Dastoor, B., Magner, N. and Stolp, C. (2003), The contribution of organizational justice in budget decision-making to federal managers organizational commitment, Journal of Public Budgeting, Accounting & Financial Management, Vol. 15 No. 4, pp. 505-14. Stratton, P. and Hayes, N. (1988), A Students Dictionary of Psychology, Edward Arnold, London. Sweeney, B. and Pierce, B. (2004), Management control in audit rms: a qualitative examination, Accounting, Auditing & Accountability Journal, Vol. 17 No. 5, pp. 779-812. Tan, H.-T. and Kao, A. (1999), Accountability effects on auditors performance: the inuence of knowledge, problem-solving ability and task complexity, Journal of Accounting Research, Vol. 37 No. 1, pp. 209-23. Tan, H.-T., Ng, T.B. and Mak, B.W. (2002), The effects of task complexity on auditors performance: the impact of accountability and knowledge, Auditing: A Journal of Practice & Theory, Vol. 21 No. 2, pp. 81-95. Tetlock, P. (1985), Accountability: the neglected social context of judgment and choice, in Staw, B. (Ed.), Research in Organizational Behavior, Vol. 1, JAI Press, Greenwich, CT, pp. 297-332. Topi, H., Valacich, J.S. and Hoffer, J.A. (2005), The effects of task complexity and time availability limitations on human performance in database query tasks, International Journal of Human-Computer Studies, Vol. 62 No. 3, pp. 349-79. Waggoner, J.B. (1986), Nonsampling risk: auditor detection of compliance deviations, unpublished doctoral dissertation, University of Cincinnati, Cincinnati, OH.

Walczyk, J.J. and Grifth-Ross, D.A. (2006), Time restrictions and the linkage between subcomponent efciency and algebraic inequality success, Journal of Educational Psychology, Vol. 98 No. 3, pp. 617-27. Wongpinunwatana, N. (2003), Using computer technologies to disseminate business knowledge, Managerial Auditing Journal, Vol. 18 No. 9, pp. 761-71. Wood, R.E. (1986), Task complexity: denition of the construct, Organizational Behavior and Human Decision Processes, Vol. 37, pp. 60-82. Wood, R.E., Bandura, A. and Bailey, T. (1990), Mechanisms governing organizational performance in complex decision-making environments, Organizational Behavior and Human Decision Processes, Vol. 46 No. 2, pp. 181-201. Yates, J.F. (1990), Judgment and Decision Making, Prentice-Hall, Englewood Cliffs, NJ. Yerkes, R.M. and Dodson, J. (1908), The relation of strength of stimulus to rapidity of habit formation, Journal of Comparative Neurology of Psychology, Vol. 18, pp. 459-82. Further reading Waggoner, J.B. and Cashell, J. (1991), The impact of time pressure on auditors performance, The Ohio CPA Journal, January/April, pp. 27-32. About the authors Anthony R. Bowrin is a Lecturer in Accounting at the University of the West Indies, St Augustine Campus. He earned his PhD in Business Administration with a specialization in Accounting from Southern Illinois University Carbondale. His current research interests include corporate governance and international nancial reporting. Anthony R. Bowrin is the corresponding author and can be contacted at: abowrin@svsu.edu; anthony.bowrin@sta. uwi.edu James King II (deceased) was the KPMG Distinguished Teaching Professor at Southern Illinois University Carbondale until his recent passing following a brief illness. He published extensively in rst-rate academic journals and was the recipient of several teaching awards.

Pressure, complexity, and audit 181

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com Or visit our web site for further details: www.emeraldinsight.com/reprints

Вам также может понравиться