Joanna L (1994) - The Effect of Experience

Behavioral Research in AccountingVolume 6, 1994Printed in USA

The Eêct of Experience onConsensus of Going-Concern

JudgmentsJoanna L. Ho

University of CaUfornia, Irvine

ABSTRACTStatement on Auditing Standards No. 59 requires that auditors evaluate tliegoing-concern status of each client. Determining whether a firm will continueas a going concern can be a complex process. Since many data items arepotentially relevant, the task is relatively unstructured. In spite of these cog-nitive demands, previous studies suggest significant agreement among au-ditors' going-concern judgments. Results of this study reveal a lack of con-sensus among both experienced and less experienced auditors who weregiven information for a problem firm. This lack of consensus may explainwhy auditors often disagree on the appropriate audit report for a problemfirm. Also reported are models of the judgment processes of both experi-enced and less experienced auditors. Auditors in both groups placed moreemphasis on the current liquidity and expected profitability of the client thanon other financial indicators. Moreover, experienced auditors generated morepositive going-concern judgments.

Prior to the issuance of Statement on Auditing Standards No. 59(SAS 59), a going-concern evaluation was called for only if an auditordoubted a client's ability to continue in existence. Currently, cin audi-tor must explicitly conclude whether an audit client will continue as agoing concern for up to one year from the date of the audited state-ments [AlCPA, 1988]. If an auditor has substantial doubt about aclient's going-concern status, a modified audit report must be issued.

Auditors have indicated that evaluating a client's going-concern sta-tus can be a difficult audit judgment [Chow et al., 1987]. In question-able going-concern situations, auditors often disagree on the appro-priate audit report [Asare, 1990b; Campisi and Trotman, 1985; Kida,1980; Mutchler, 1986). Kida [1980], for example, found that a firmidentified as having going-concern problems did not necessarily re-ceive a qualiJQed audit opinion. He attributed this finding in part toconflicting economic considerations (e.g., an auditor might want to

I wouid iike to thank Wiiiiam Wright and Elizabeth Davis for their contribution to theproject and to the preliminary drafts of the manuscript, i am also grateful for the help-ful comments of an anonymous reviewer and the editor. The data set used in this studyis available upon request.

The Effect of Experience on Consensus of Going-Concern Judgments 161

retain an audit client but feared potential litigation initiated by theclient's investors). Mutchler [1984] provides related evidence indicat-ing that auditors disagree significantly about their reporting respon-sibility given going-concern uncertainties.

In most audit contexts, the lack of a well-specified criterion eventmakes the evaluation of audit decisions difficult. Thus, audit judg-ment researchers and practitioners often rely on consensus as a sur-rogate measure of audit decision accuracy. Simileurly, regulators andthe courts have often regarded consensus as a surrogate for decisionquality [Libby and Lewis, 1982]. In practice, many public accountingfirms use an internal consulting team to provide advice for difficultaudit decisions. In contrast to Kida's [1980] "conflicting economic con-siderations" assumption, the current study focuses on another hy-pothesis—^that auditors' disagreement on audit report decisions maybe attributed to their lack of consensus on the perceived likelihoodthat a firm will continue as a going concern. If a lack of consensus onauditors' going-concern judgments is found to account for auditors'disagreement on what is the appropriate audit report to issue, thendecision aids (e.g., explicit models and knowledge-based systems) maybe used to overcome this problem.

Much recent debate^ has occurred concerning the role of the audi-tor in considering and reporting on the viability of audit clients [Asare,1990a]. SAS 59 significantly expanded the auditor's reporting obliga-tion in two ways. First, relative to the previous auditing standard (SAS34), the auditor must now go beyond the consideration of asset recov-erability and liability classification and evaluate the client's going-con-cern ability. Second, SAS 59 changed the meaning of, and conditionsfor, and audit report to be modified given a questionable going-con-cern status. Even though prior studies [e.g., Campisi and Trotman,1985; Kida, 1980] observed significant agreement among auditors' go-ing-concern judgments, these findings may not generalize to post-SAS59 judgments as a consequence of differences in auditing standards.

Since the going-concern judgment is relatively unstructured andinformation-intensive, significant experience may be necessary for anappropriate judgment in uncertain client situations. In practice, ex-perienced auditors (managers and partners) are responsible for go-ing-concern judgments. However, after about three to four years ofaudit training and experience, an auditor begins to assume partialresponsibility (as a supervising senior) for going-concern evaluations

'This recent debate includes different issues, such as informational value of the going-concern report [CAR. 1978; Libby. 1979]. the consequences of the going-concern re-port, the ability of an auditor to identify companies experiencing going-concem prob-lems [Altman and McGough. 1974; Kida. 1980]. and the responsibility for reportinggoing-concern uncertainties [AICPA. 1987].

162 Ho

(especially since going-concern evaluations are now required for allaudit clients). At this point in their careers, auditors have strong per-sonal incentives to add to their knowledge and to become fully in-formed with regard to the going-concern evaluation process. To ex-plore the effect of experience on going-concern Judgments, auditorsat various levels of experience are evaluated with respect to their go-ing-concern judgments. TTie results of this study reveal considerabledisagreement at all levels of audit experience. Surprisingly, additionalexperience did not improve auditors' going-concern judgment agreement.

The rest of the paper is organized as follows. The following sectionsummarizes conclusions from previous reseeirch on going-concernjudgments. The possible impact of audit experience on going-concernjudgments is considered in the next section. Results are then pre-sented on (1) the extent of judgment consensus and the impact ofdifferentisd experience on the magnitude of going-concern judgmentsand (2) models of the going-concern judgment process, including therelative importance of types of information. A summaiy and discus-sion section concludes the paper.

PREVIOUS RESEARCH ON GOING-CONCERNJUDGMENTS

Kida [1980] obtained going-concern judgments and audit opiniondecisions from 27 audit partners for 40 firms, h£ilf of which had actu-ally failed. To evaluate the ability of the partners to identify problemfirms, Kida converted the 40 going-concern judgments into "problem-firm" or "non-problem-firm" judgments.^ The meein number of correctjudgments was 33.2 (83 percent), with a range of 24 to 37 correct. Amoderate degree of correlational consensus was indicated for the go-ing-concern judgments (a mean inter-subject Spearman correlationof .755, suggesting an r̂ of approximately 57 percent). Concerningthe audit opinions, however, a mean of 25 percent of the firms identi-fied as having problems were assigned unqualified opinions. Addi-tional response data revealed that perceived economic trade-offs suchas losing a client, deterioration of client relations, or investor lawsuit,affected the auditor's willingness to issue a qualified opinion.

Campisi and Trotman [1985] investigated both going-concern judg-ments £ind audit opinion choices. Going-concern judgments for ninefirms were elicited on a ten-point scale. A moderate degree of correla-tioncil consensus (a mean inter-subject correlation of .73, correspond-

[1980] used a stx-polnt scsde to elicit auditors' continuity decisions. The twoendpoints were "I beiieve this firm does/does not have going-concern problems." Rat-ings higher them 2.5 {the midpwint) were converted into "non-probiem-firm" Judgmentsand those iower than 2.5 into "probiem-flrm" judgments.

The Effect of Experience on Consensus of Going-Concem Judgments 163

ing to an r̂ of 53 percent) was exhibited by the 25 auditors. Concern-ing the audit opinions, when the likelihood of continuance was indi-cated to be very high or very low, the auditors reached a high level ofconsensus on the appropriate audit opinion. When a firm's going-con-cem status was clearly uncertain, however, the opinion choice was spreadamong unqualified and qualified, adverse and disclaimed opinions.^

While research on going-concern judgments suggests moderate judg-ment consensus, subsequent work by Kida [1984a] on fgiilure predic-tions'* is Eilso relevant. Kida used 73 audit managers cind partners toprovide probability judgments of firm failure (defined as beinkruptcy,liquidation, or default on bond commitments). A lack of consensus isreported for all base rate conditions [p. 148]. Kida [1984b] also testedthe hypothesis that the way a failure judgment is framed would resultin dissimilar use of confirmatory versus disconfirmatory information.Sixty-six partners £ind managers provided probability judgments offirm failure. A wide range of probability judgments was obtained [seep. 337], with negative information items (items suggesting failure) gen-erally being regarded as more relevant than positive information items(also see Trotman and Sng [1989]).

In summary, prior research suggests that auditors can identify prob-lem firms [Kida, 1980]. Also, Kida [1980, p. 519] concludes that"... asubstantial level of agreement exists between auditors when they makecontinuity decisions from financial statement data." This conclusion,however, was based on a correlation measure of consensus,^ and cor-relational measures address only the relative ordering of judgments

^Mutchler [1984] provides evidence that auditors hold differing beliefs about their re-porting responsibility for going-concem uncertainties. Audit partners provided responsesthat ranged from the belief that violation of the going-concem assumption automati-cally impaired the fair presentation of the financial statements to the belief that a client'sgoing-concern status is not a problem so long as the recoverability of assets is not anissue and the classification of liabilities is correct [also see Kida. 1980]. SAS 59 [AICPA.1988] has expanded the responsibility of the auditor and clarified when a modifiedaudit opinion should be issued.•'A going-concem judgment is different from a judgment of the likelihood of companyfailure. Information that contradicts the going-concem assumption is "...informationthat significantly relates to the entity's inability to continue to meet its obligations asthey become due without substantial disposition of assets outside the ordinary courseof business, restructuring of debt, externally forced revisions of its operations, or simi-lar actions" [AICPA. 1988. p. 341.02] versus information that suggests bankruptcy orliquidation of the company [also see Asare. 1990a. pp. 50-51].Ûsing Kida's dichotomous judgments of whether a firm does/does not have going-concem problems. Ashton [1985. pp. 180-184] reported that the number of times thattwo (of 25) auditors agreed with each other ranged from 16 (40%) to 39 (98%) of the 40case judgments. Average agreement for an auditor with the other 24 auditors rangedfrom 24.96 (62.4%) to 33.75 (84.4%) of the 40 judgments; therefore, based on the mag-nitudes of the dichotomous judgments, a wide range of consensus measurements isindicated. Ashton also documents a positive correlation between consensus and accu-racy for Kida's auditors—also see Keasey and Watson [1989].

164 Ho

Without considering whether the judgment distributions are actuallysimilar (i.e., a high correlation may exist even when two judgmentdistributions are very different in terms of their central tendency, dis-persion, and other distribution parameters).^ One could mistakenlyconclude that considerable consensus exists when the judgments are,in fact, very different [of. Pincus, 1990, pp. 3-4]. Kida [1984a, 1984b]reports a lack of consensus when somewhat similar company fsiilurejudgments were made. Also, most research (except Asare, 1990b) wasconducted prior to implementation of the significant changes prescribedby SAS 59. In contrast to SAS 34's emphasis on asset recovery andliability classification, SAS 59 imposes a going-concern evaluationwhich requires broad information assimilation. Therefore, the degreeof consensus of current going-concern judgments remains to be as-certained. The current study uses scale-dependent measures of con-sensus.

With respect to the audit report, since auditors can identify prob-lem firms and there is correlational evidence of moderate consensuson going-concern judgments, why, then, do auditors disagree on theaudit report to be issued to problem firms [Campisi and Trotman,1985]? This lack of agreement may arise from the conflicting economictradeoffs that auditors face when issuing modified audit reports, oralternatively, because auditors do not agree on the likelihood that anidentified problem firm will continue as a going concern and thesedifferences affect their choice of audit reports.

THE GOING-CONCERN JUDGMENT AND AUDITEXPERIENCE

The going-concern judgment requires scrupulous mental integra-tion of numerous facts. It is a difficult and error-prone process (Ashton,1982; Einhorn, 1972; Libby and Libby, 1989]. Auditors must choose,weigh and combine predictive facts into the going-concern judgment,a mental process not prescribed in audit manuals or indicated bydecision aids. Adding to the difficulty, available data often provideambiguous and/or conflicting implications concerning firm continu-ity. Knowledge of how to mentally Integrate such information in a validand reliable way miay come only from experience. Given the complex-ity of the going-concern judgment, however, even auditors with con-siderable experience may fail to reach consensus.

To improve the quality of their work, auditors have personal incen-tives to learn how to appropriately integrate predictive facts into go-ing-concern judgments. While learning from past decisions can be

^For example, the product-moment correlation is computed as the covarlance (adjustingfor two possibly very different meEin Judgments) divided by the product of the standarddeviations (adjusting for two possibly very different degrees of Judgment dispersion).


difficult [Einhorn, 1980], experienced professionals shovild be betterinformed about which facts Eire predictive, and they may produce morevalid Judgments [Blattberg and Hoch, 1990]. Auditors also increasetheir knowledge as they participate in common, CPA firm-sponsoredtreiining courses. Further, experienced auditors who have served cli-ents having financial difficulties or who have been made aware of suchclient situations may recognize more cleeirly just how severe problemsmust be before a firm can no longer be considered to be a going con-cern [however see Ashton, 1991].'̂ Therefore, significant experienceas well as training may permit experienced auditors to have reason-ably well-developed schemata for going-concern situations while lessexperienced auditors may have less well-developed schemata [Chooand Trotman, 1991, p. 469]. This discussion leads to the followinghypotheses:H1: Consensus will be higher among experienced auditors than among

less experienced auditors.H2: Experienced auditors may weigh and combine going-concern in-

formation differently than less experienced auditors, hence thetwo groups will produce different going-concern judgments.

EXPERIMENTSubjects

One hundred and fifty-six auditors from four multinational account-ing firms completed the experimental materials. The mean (median)audit experience was 54 (32) months and the range was 12 to 300months. One hundred and fourteen auditors from one firm partici-pated during two staff training sessions,^ whereas 42 managers andpartners, from four multinational firms, were contacted directly bythe researcher. In practice, at approximately 42 months of experi-ence, a supervising senior becomes at least pcirtially responsible forgoing-concern considerations (especially, given the new requirementof SAS 59 that the going-concern status of̂ each audit client must beevaluated). To address the experience effect on going-concern judg-ment agreement and to understand the judgments of auditors whohave reached the threshold of training, experience, and responsibilitywhere going-concern assessment is an explicit part of their profes-

"Client circumstances that raise substantial doubt about continuance as a going con-cern are relatively infrequent; therefore, even managers and partners may have limitedactual experience making (difficult) going-concern judgments.^Statistics show no significant differences between these two groups for the three rat-ings of interest. The mean going-concern judgments for auditors in the first and thesecond training sessions are 3.08 and 3.11: the two means are not significantly differ-ent (F=1.03. p<.89). Also, there are no statistical differences in these two groups' rat-ings of the realism of the case (5.72 vs. 6.02; F=1.06, p<.85) and of how interesUng thecase was (5.66 vs. 5.85; F=1.33, p<.31).

166 Ho

sional duties, subjects were divided Into two groups based on a cutoffof 42 months of audit experience. This cutoff yielded two groups ofauditors; 93 "less experienced" and 63 "experienced"; the experiencelevel of the two groups was significantly different (25.4 versus 95.8mean months of audit experience, t=9.89, p<.001, one-tailed).^

Judgment Tasks and MaterialsEach subject received a comprehensive case study. The "American

Computers" case was designed to be representative of the company(and industry) data and analytical procedure results that auditors usewhen analyzing a financially-troubled borrower in a client's loan port-folio. ̂ ° A detailed narrative on Americein Computers' business, man-agement and history, complete income, balance sheet, sund cash flowstatements, and a listing of finemcial ratios were provided. Initially,the materials were based on actual company and industry data; then,they were adapted to a financial scenario of rapidly declining profit-ability' and deteriorating future prospects.

The case is designed to suggest a questionable going-concern sta-tus. First, while a mix of positive and negative information was pro-vided, the case included events identified in SAS 59 as indicators ofpotenticd inability to continue as a going concern. The indicators,grouped according to their nature, are as follows;

- diminishing earnings,- a net loss in the most recent year;

Other financial indications;- a relatively large note due very soon with no opportunity for

refinancing from the lending bank;Internal matters;

- loss of key fingincing and marketing executives,- new management has limited experience in microcomputer in-

dustry,- a late response to rapidly declining computer prices,- termination of personnel in engineering and marketing that

may inhibit the firm's ability to produce new products emdintensify marketing activities; and

External matters;- loss of dealers,- uncertainty inherent in microcomputer industry.

*rhis result (and each of the other results) is not sensitive to selection of the cutoffpoint of 42 months.'"The original case was designed by Wright [1991] to investigate the effect of presenta-tion format (i.e., tabular vs. tabular + graphics), but was revised to meet the specificneeds and purposes of the current study.


Second, the ratios in the case present a financial profile somewhatsimilar to that of bankrupt firms in the Altman and McGough [1974]sample. Based on their discriminant model, the z-score for AmericanComputers suggests questionable viability of the firm. The z-score thatminimized classification errors in the Altman and McGough study is2.675. Firms with scores below 2.675 have (at least some of the) cheir-acteristics of failed firms, although for z-scores between 1.81 £ind2.99there is the possibility of misclassification. The z-score for AmericanComputers is 1.94 which is below the cutoff point and within the rangeof uncertain firm viability.

Focusing on the borrower's financial viability, auditors were askedto provide both a going-concern judgment and a loan coUectibility as-sessment. The going-concern judgment was; "Given the informationyou have reviewed, what is your evaluation of American's ability tocontinue as agoing concern between August 1, 1988, and October 1,1989?" [itcilics in the original]. The horizon for the going-concern judg-ment (and the loan) was approximately one year from the date of thematerials. Subjects were also asked to make several intermediate judg-ments of financial dimensions, such as short-term liquidity, expectedprofitability, and financial leverage, before reaching a going-concernconclusion (see Appendix One). The elicitation of these intermediate.coftasuarf.s r.aia /irovide insight into how auditors use these cues inmaking going-concern judgments. The going-concern and intermedi-ate judgments were made on a continuous seven-point Likert scale,with 7 labeled "very strong," 4 labeled "average," and 1 labeled "veryweak." The loan judgment was the percentage of the loan to be col-lected on the due date one year hence.

After subjects completed the judgment tasks, they were asked toindicate whether they considered the case interesting and realistic.For both, responses were indicated on a seven-point scale with a maxi-mum of "strongly agree" (coded 7) and a minimum of "strongly dis-agree" (coded 1). The mean (standard deviation) of the "interest" re-sponses was 5.74 (.81) and the "realism" mean (standard deviation)was 5.88 (.83), with both of the medians being 6.00." The auditorsIndicated that the materials were both interesting and realistic.

ProcedureSubjects who were attending staff training sessions completed the

case materials in a classroom setting. The time spent for completionranged from 40 to 50 minutes. To enlist more experienced auditors toparticipate, the researcher contacted partners in local offices of fourmultinational accounting firms in southern California. These part-

' 'The mean ratings of "realism" and "interest" by experienced auditors were 5.98 and5.86. respectively, which were not significantly different from those ratings by less ex-perienced auditors (5.81 for "realism" and 5.65 for "interest").

168 Ho

ners then provided lists of volunteer participants, and the researcherarranged to meet groups of volunteers either in their office conferencerooms or in a conference room at the researcher's institution. As hadthe auditors attending the training sessions, all experienced auditorscompleted the case materials in a classroom setting. Their completiontime ranged from 35 to 50 minutes.

RESULTSBflagnitudes of the Going-Concern Judgments

Subjects confirmed that the status of American Computers as agoing concern was uncertain. The going-concern judgments averaged3.18, with a median of 3.00 (labeled to be "less than average ability tocontinue"), on the seven-point response scale.

Auditors who had reached the level of actual audit responsibilityfor going-concern judgments reported different conclusions—^more au-dit experience resulted in more positive going-concern judgments. Themean going-concern judgment for the 63 experienced auditors is3.49,^2 while the mean for the 93 less experienced auditors is 2.97;and, the means are significantly different, t=2.53, p<.01.^^ Also, theMann-Whitney U test shows that the going-concern judgments of ex-perienced and less experienced auditors are not from the same distri-bution (z=2.10, p<.036). Table 1 presents the frequency distributionof going-concern judgments of the overall group and of judgments bygroups with different levels of audit experience. Descriptive statisticsof going-concern assessments are shown in Panel A of Table 2.

The correlation for the going-concern judgments and months of auditexperience is .15, p<.06.'* Unexpectedly, the correlation between thelikelihood that American Computers will continue as a going concernand the percentage of the loan thought to be collectible is not signifi-cant (r=. 16, p<. 10). A further test shows that the insignificance of thecorrelation between the percentage of the loan thought to be collect-ible and the likelihood of American Computers continuing as a goingconcern is due to the experienced group (r=.13, p<.20). For the lessexperienced group, there is a significant correlation between the per-centage of the loan thought to be collectible and the likelihood of Ameri-can Computers continuing as a going concern (r=.26, p<.01).

'^There is no significant difference between the going-concern judgments of the 21auditors who attended training sessions and the judgments of the 42 auditors who metwith the researcher as individual groups.'^The judgment distributions are essentially symmetric and unimodal; logarithmic andsquare-root transformations did not result in different statistical inferences. This re-sult is particularly important, given that multivariate normality is assumed for themodels developed in the next subsection.'••This correlation is somewhat similar to levels of association reported by Ashton andBrown [1980, p. 274! and Ashton and Kramer (1980, p. 10! for internal control judg-ments and degrees of audit experience.


TABLE 1FREQUENCY DISTRIBUTION OF GOING-CONCERN JUDGMENTS

Going-ConcemJudgment

1.001.101.401.501.602.002.102.202.402.502.602.702.802.903.003.203.303.503.703.804.004.104.204.504.905.005.205.505.806.007.00Total

Experienced#

31010

1301000000

120071060201411180

6 3

%

4.71.601.60

20.601.6000000

19.000

11.11.609.503.201.66.31.61.61.6

12.70

100%

Less Experienced#

20111

1932242122

221243191210101021

9 3

%

2.201.11.11.1

20.43.22.22.24.32.21.12.22.2

23.71.12.24.33.21.19.71.12.21.101.101.102.21.1

100%

Overall#

51121

3233242122

3412

1141

1514115121

101

156

%

3.20.60.61.30.6

20.51.91.91.32.61.30.61.31.3

21.80.61.37.12.60.69.60.62.50.60.63.20.61.30.66.40.6

100%

Consensus of Going-Concem JudgmentsEven for the highly experienced auditors, considerable disagree-

ment is reveEiled by the dispersion of the audit judgments. The stan-dard deviation of the going-concern judgments for the experiencedauditors is 1.49 with a range of 1.00 to 6.00, indicating little agree-ment. Ilie results for the less experienced auditors are similar (i.e., astandard deviation of 1.06 and a range of 1.00 to 7.00). Theinterquartile ranges also indicate substantial disagreement (i.e., [2.00,4.90] for the experienced auditors and [2.00, 3.50] for the less experi-enced auditors). The standard deviations of these two groups differfrom each other significantly (F-2.00, p<.003). For all 156 auditors.

170 Ho

CO

CO h

V

I9 s^q ^ 00 10 ^ o

S CD 05 00in

••. n c» CO c s CO

o O o o o o o

S _ IE...,o lONO

S2

O

.-, 5> _. ocncooiM

CO

ooCN

08

oCO

70

0)

sCO

52

00

8(N

17

CO

73

8cs

21

8CO

00

CO C S C O C O ! N C O ( N C O

CO

G

S'o''o o o o op OOOOCO £1 £2 21 ci 2 , cs CON -îfiCOCSÔOlOCI3 -;(NN-<cq(N02(N ( N C O C O M ' C O C S C S

G

CO

05 . . - . in , . ^^ . . ^CS 3r H t s . . c o i n i n c o c o c

l i

M C O l O N - H - ^ - ^ C S &rt ! O ^ 0 5 C J > ^ 0 5 ^ t,

o q q o q q q q q '^CO c j CO CO £i £2 s i £2 t j00 o c s i o i n t - . c - j o c " o S^ r H l O C p ' ^ N I N O J c d oCO I M ' C O C O M ' C O C - J I N n 'S

IIll

CO "O

iMi

I(0

CO

in

qV

dcs

oqV

CO

cs

d

:.28

inoV

a.

632.

:.i9

oV

a

543.

OCO

OCO CO CJ) .i n o> CO M

CO OO

o „cs

oo

d o -q 8 qV V V

a a aCD lO' CQ COcs CO (N O

I

11ft. .zi

X 'o V

CBiS CR

S I

O. O

in


the standard deviation of the judgments is 1.27, with use of the entirerange of the response scale (i.e., 1.00 to 7.00).

The dispersion of the going-concern judgments at different levels ofaudit experience is indicated in Table 1. If increasing audit experi-ence leads to increasing judgment consensus, the expectation is greateragreement at higher levels of audit experience. Such an effect is notindicated. As can be seen from Table 1, both the experienced and lessexperienced auditors made judgments in the substantial range of 2.0to 6.0. The general conclusion is that considerable disagreement ex-ists at all levels of audit experience. Therefore, HI is not supported.

Models of Going-Concem JudgmentsThe going-concern judgments differ for the experienced and less

experienced auditors. To explain these differences, models of mentalintegration of information were built for both groups. The intermedi-ate judgments of current and expected liquidity, expected future prof-itability and current financial leverage, as well as evaluations of re-cent trends in these variables, were possible predictors in each judg-ment model. Descriptive statistics for the intermediate judgments areshown in Panel A of Table 2. Each intermediate judgment is a mentalaggregation of data pertaining to that financial dimension. Having thesubjects mentally aggregate many data items into intermediate judg-ments facilitates model building; that is, the judgment process can bemodeled more psirsimoniously and reliably.

Separate models were estimated for the less experienced and expe-rienced auditors (see Panel B of Table 2). The seven intermediate judg-ments (see Appendix 1) were made avEiilable in a forward, stepwlsemanner (with a .05 significance cutoff). A clear pattern of results isrevealed. Two intermediate judgments were important for the going-concern judgment in both models: LIQUID, which is the auditor's evalu-ation of the current liquidity position of American Computers, £indPROFIT, which is the expected relative profitability of American Com-puters over the upcoming 24 months. LIQUID was always the first toenter the stepwlse regressions in the two models. LIQUID and PROFITare, respectively, aggregations of balance sheet and income data thatare suggestive of the ability of the firm to survive as a going concernover the upcoming year. None of the other five intermediate judgmentsis significant in either of the models. An overall regression analysisthat combined the two subsamples also shows that LIQUID andPROFIT are the two most important intermediate judgments for thegoing-concern judgment; their respective weights Eire .391 (t=3.775,p<.0002) and .500 (t=3.450, p<.0007).

As can be seen from Table 2, one potential explanation for differ-ences In the going-concern judgments of the experienced and less