Sie sind auf Seite 1von 30

-13 g

.i
avro ,,f

ACCOU

After reading this chapter. you should have an appreciation of the fo owinE:

I tle lelaviourat perep€ctile and the nature of behaviou.al accounting research


I tt'e .o nt.iUution to our undeGtanding ofthe role of accounting information within and
outside the a.counting entity provided by studyjng behaviour

EI the influence of accounting information on behaviourand on decision processes


E the fact that organ isations aie complex environments and accounting disclosures are
trade-offs between competing perspectives and inierestr

E the inh erent constrai.ts on behaviouralreseaich

l!
chapter 12 dis{us.s€d research into how capital markets react to th€ releas€ of a.counting
information. A dimculty of capital markets research is that it does Dot investigak how
information is actually processed by market panicipants because it coDc€ntraks on
only two it€ms, the release of information and the capital market readion (if any), not
what goes on between these rwo evmrs. One of the respons€s to this limitation was the
development of agency theory as described in chapter 1 1 . ABency theory €xamines why
fiIms droose panicular accounting methods ftom a set of acceptable alt€matives. Aldrough
agency theory is concemed with the actions of individual manag€rs and olh€r pani$, it
make8 the important assumption that all individuals ar€ hotivated to maximise their self-
inrciest. This assumption makes it easier to dev€lop testable predictions for research but
it does not r€ally e{prain why people act as th€y do because the seme rcason is given for
different behaviou$. For instance, when a manager choos€s the FIFO m€thod of inventory
measurem€nl, agency theory saF this is because it is in the s€lf-intarest ofihe manager to
do so. similarly, if another manager chooses rhe weighted a\'era8e method instead, the
same reason b given by the theory that is, the choice was in the manager's self interest.
As we make different choices for different reasons, 'self-interes! is a very incomplete
explanation for p€oplds behaviour. Cl€arlt ifwe are to have a betei undeFtanding about
how people use accounting information, then we need to study pmpl€t actual behaviouis
and decision processe$. This is the role of a rhird arca of accounting reseaKh, popularly
knovllr as'behaviouEl accountinS research' [BAR).
This chapt€r inuoduces you to the field of behavioural accounting research by
describin8 some of the key qu€stions it investigates and some ofth€ main research tools
used by researchers. AlonS the way w€ will indi(ate some oflhe imponaot findin$ so
far fiom this research, particularly in the area of financial accounting. Like the capital
marke$ and agen.y theory schools of r€search, behavioural accoundng also has ils
limitations and these will be m€ntioned throuShout lhe €hapter wher€ rclevanr.

@- erHnvlouRAl- ACCOUNTING
RESEARCH: DEFINITION AND SCOPE
Behavioural ?ccountin8 researqh has b€en defned as:
The srudy olth€ behaviour of accounonrs or the behaviour of non-accountants as they
are influenced by a.coundng functions and repons.I

Behavioural accounting res€arch (BAR), capital markets research and agency theory
research can all b€ called'positive'research in the sens€ that they all are concern€d
with dis{ov€ring 'facts': capital mark€$ research aslis 'hou, do securities markets react
to accounting informationl'; agelcy theory asks 't ft41 ar€ the e.onomic incentives that
determine the (hoice of a(counting methods?'; and behavioural research asks 'llop do
people a€tually use and process accounting information?' However, they are also very
different in many respects. For instance, capital markeb rcsearch look at the macro
level of aggreSate securities markets, whereas agency theory and behaviounl a.counting
foclls on the micro level of individual managers and firms. Capital markets res€arch
and agency th€ory are both deriv€d ftom the disciplin€ ofeconomics and disPense wirh
people's actual motivations by assuming (hat everyone is a rational wealth maximiser'
Behavioural accounting on the other hand, is derived from other disciplines such as
psychology, sociology and organisational theory, and 8€nerally mak€s no assumptions
aboui bow people behave; rather, ils objective is to discover utry People behave as th€y do
Consequently, each of thes€ three schools of accounhng research is desrgned to answer
very different tpes ofqu€stions about the pnctice ofaccounting.

446 PARTI Ac(ourlinS and research


Research in behavioural accounting is imnense and has covered many different
spheres of accounting activity.'? Some BAR studies have, for instance, been applied in
rhe area ofaudiling to improve audirors' decision making. For e\ample, when auditors
are planning the way they will conduct an audit ofa parlicular client, they musi assess
how much risk is associared with the client. The higher lhe dsk, the more audit work
ihar will have ro b€ done. Assessing tisk is a highly comPlextask that could have serious
consequences for the auditor (and investo$) if an incorrect assessment is made and the
auditor consequently performs a poor aud;1. BARhas been used to help analyse th€ risk
assessments ofaudiiors and imprcve them. Another maior area oIBAR has b€en in the
field of management accounting. For example, BAR has b€en used to help expiore and
understand the various incenlive and disincentive issues associated lt'ith diffe.enl LyP€s
ofbudgeting processes and ho\ organisational form and accounting systems can affect
the behaviour of individuals within firms. Ilowever, sinte this is a ten on financial
accounring, th€ main focus olthis chapter is on the information contained in financial
statements for use$ enernal 1o ft€ 6nD. The major type ofBAR in this area has become
known as human judgement theory (HJl') or human informatjon processing [HIP) and
encompasses the judgement and d€cision naking of accountants and auditors and lhe
influenc€ ofthe output ofthis function on users' iudgenlents and decision making.r The
aim of research within this model is oft€n more than that ofetplaining and p.edicling
behaviour at an individual or group ievel. Il is also concerned i{ith inlprovirg fie
quality of decision making. In the .onrext of linancial accounting, thnt ailn translates
lnto improving decision making by both producers (iDcluding audhors) and usel1 of
accounting reporls.

@=wHY rs BAR TMPoRTANT?


There are a number of very good reasons that BAR is impofiant to accounting
pra{titioners and oihers:
. we noted at th€ start of this chapter how other accounting research schools such as
capital markets and ageDcy lheory are nol equipped to ans$'er questions abour lol'
peopl€ use and process accounting information. To nll this void we need research
that specifically examines the decision making acliviii€s of the preparers, use$, and
auditors of accounling information.
. BAR can provide valuable irsights into the Nays differenr types of decision nakers
produce, process and reaci to particular irems of accounting information and
communicalion lnethods. We can use these insights to imProve decision nakhg
in a variety of h'ays as described later jn the chapter uhen we disclrss the'BrLns$'ik
Iens model' ol decision making. Improving decisior making is impoftant, ofcourse,
10 both users of firancial information (who want 10 avoid making bad decisions
which lead to losses) and to preparers and atditors of financial infornarion
(who $,ant to avoid being sued). An understanding of the information-Processing
aspects of accounting is also imPortant to you in your own career' As infomation
professionals, accountants need to develop high levels of expenis€ in jnformation
gathering, processing and comlnunication. BAR can help lead to training and
knolvledge that inprov€s rhes€ skills, drus allowing you ro perform better in the
n'orkplace and improve your chances of obtaining {'ork, Saining promotions and
achieving betier pay.a
. BAR caD potentially provide useful informaiior to accounting regulalors such
as the Australian Accounting Standards Board (A,^sB) As the main obiectiv€ of
accounting is to provide 'decision usetul' information 5 menbers of the AASB are

CHAPTLR 13 BehaliournL r.search in a.!ountirS


:!q
consranrly glappling with the problem of which accouDting m€thods and what rlpes
of disdogures will prove 'useful' to th€ us€Is of financia.l $atemen6. Behavioural
accounting resea.chers can dircctly study specifrc accountiDg options and repon to
srandard setters on which methods and disclosures improv€d u$ers' decisions.
. The fndinSs of BAR can also lead to emciencies in the work practices of accountanr
and oth€r prcfessionals. For examplq the o(Pe(ise of senior and €aperienced
membeB of an accountinS 6Im can be record€d and hamessed by BAR merhods ro
develop computeris€d ereert s,$tems for a van€ty ofdecision-making contexts.Ihes€
€xpert systems can be used to train inexperienced practitioneB and to undertak€
routine task that would orherwise tie up the luable time of experienc€d staff.
Some accounting 6rmr for instance. have used BAR methods to dev€lop expen
systems to conduct risk ass€ssments ofpotential audir clients. In the pa$, rhis time-
consumidg tast would have be€n done by s€nior membets of the accounting 6rm,
but a detailed sseeninS of pot€ntial clients can now be done by less qualified staff
using th€ computer exp€n system, subiect to a final rcview by the 6rm's panners.

Development of behavioural accounting research


The ierm 'BAR'first app€ared in the literature in 1967,6 but HIf research had its
foundations in the psychologr literature with the seminal work of Ward Edwards in
1954.7 The application of the r€s€arch to accounting and auditing can be dated to 1974
when Ashton published an €xperimental sflrdy ofthe intemal control iudgem€nts made
by auditors.s
The last 30 years have seen an explosion of BAR in general and Hl'f research in
partiorlat especially in auditin& where the importance of iudgement to the audir
proress is paramount. To some extent lhe dev€lopment ofbehavioural research in the
area of financial accountinS has been eclipsed by the dominance of contraoing theoiy
since the 1980s. Nevertheless, imponant insights irio the link between accounting
information and human behaviour hav€ been fonhcoming.
Many disciplines (e.& poliiical science, orSanisation theory, sociology and siatistics)
have played a role in BAR'S growrh, but by far the most important behavioural
science in t€rms of contribuiion has been psycholog/. The growth of HII research
in accounting owes much to the adaptation of a research method alrcady well used
in the psycholog literature, the Brunswik lens model.e This technique represented
a powerfirl new research approach which could be appli€d ro the old question of
rehich data users take notice of Ashton is credited with b€ing t}le 6rst eccounting
researcher !o use this technique, followed closely by Libby who was the frsi to ule it
in a user-oriented context. Both researchers have continued to play a dominant role
in the development of BAR.

Behavioural effects of international accounting diversity


A b€haviou ral accounting study was conducted by Choiand Levich (1990). They investigated
the impact of intemational accounting differences on capital market padcipants. Cmwlh in
int€nationa I financial markets and diversity in financial rcportinS had been noted in previous
rcsearch- Therefore the authoG asked th€ followint questions: (1) ls accountinS diversity
perceived to be a problem, and if so, why? (2) Arc attempts made to cope with national
ac€ounting difterences and are these coping mechanisms successful? (3) Do problems
associated with accounting diversity lead to capitalmarket effects?
ln posing these quesiions, lhe aulhors 5ou8hl to oblain empirical evrdence about the
behavrour of paniripanls in capilal markets that would assisl regulalors and 5tdndard

PARr 3 Accounring and research


setters involved in the international harmonisaiion of financial reporting They suNeyed
jnvestment underwriteE,
52 market paticipants (innitutional investors, corporate jssuers,
market regulators; stock exchange officials and rePresentatives of rating agencies and data
services) f;om the United States, the United Kingdom, Cermany, SwitzerLand and Japan.
The authors reported that approximately half of those surveyed leh that accounting
diversity affected th€lr capital market declsions. The other half responded that diversitv
had no effect because they used successful coPinS mechanisms, or because dlverslty was
not consldered an impo(ant issue to begin with. Some investors who restated accounling
information reported that dlversity aifected their decisions, indicating that restatement was
not always sufficient to remove the problem of diversity. The research€B concluded that
accountlng differences were important, and that they affecied the capital market decisions of
a significant number ol market participants, regardless of thelr nationality, size, exPerience,
scope of international activity and organisation structure. Thus, the study provlded evidence
that accounting diversity did affect decision making, sLrggesling that activities to reduce
diversity could be beneficial.

Choi, F, & tsich, R 1990, Tne upital ftarket 6ects ol intemadanal ao tinS diretsit!' Homewoad

An overview of approaches to understanding


information processing
The basic objective of Hl_I research is to describe the way in which people use and
process pieces of accounting (and other) informalion in a panicular decision making
conten. we call our description ofa person's decision-making processes a 'model' So,
for eumple, we may use HJT research techniques to 'mod€l (or represent) the $'ay in
which a bank loan oflicer processes vadous information items lor 'cues' as they are
called) sudr as the profit and cash flow figures to make a d€cision about whether to
approve a loan application fron a comPany. Although the Brunswik leDs model has
been the dominant nethod of developing models of d€cision makinS' there are also
two other major research approaches. one ofrhese is called 'Process tracing', which is
an attempt to build a decision tree representation of a person's iudgements, and the
other is known as the 'probabilistic judgement' Paradigm, in i 'hich decision processes
are represented as probability stat€nents based on Bayes's theorem Eaci ofthese three
approaches to describing (modelling) decision making is outlir€d belon'

The Brunswik lens model


Since the mid 1970s, the Brunslvik lens model has been used as an analytical
framework and the basis for most judgement studies involving predictjon (e8 of
t0 Researchers use the lens
bankruptcy) and/or evaluatioD (e.g of intemal conrrol)
model to investigate the relationship between multiple cues (or pieces ofinformation)
and decisions, iudBements or predictioN, by looking for regulariiies in the responses
to those cues. The decision maker (e g. bank loan of8cer) is viewed as looking through
a lens of cues [e.g. financial ratios) which are probabilistically related to an event'
in order to reach a ronclusion about that eveni (e.g. likelihood of loan default/
non-default).rr The diagammatic representation in frgure 13 1 will helP make this

aHAPTER 13 Behavio!ra research in acco!niing 449


Debt to

$
Achievemot

FICURE 13.1 Diagrahmatic r€presentation ofthe lens model


Soulcei AdapEd 1106 R Libby, A.counting and hunan infdnanan prccessinq: lEoty and arytications,
Engle@od ClitG, Ni: Prentice Hall, 1981, p. 6.

In developing a specific version of the Brunswik lens model, subiecls are asked
to make iudg€ments for a large number of cas€s which are based on the same set
of cues. For example, they may be asked to ass€ss whether several 6rms are likely to
fail, given rhe 6fms' working capftal, price-€arnings, eamings per share, quick, debt
to equity and otlrer ratios. A linear model describinS the fuDctioml rdationship
betw€en the cues (ratios) and responses (likelihood of failure), i! then constructed as
a means of representinS the way in h,hich infofilation is processed by individuals.r2
UsinS 6gure 13.1 as an example, oul bank loan officer is gv€n fimncial ratios and
other accounting information for differ€nt companies. Th€ bank loan officer is asked
to make a iudBement, on fi€ b.sis of this information, about which ofthe companies
will default on their loan paym€nts and which companies will not default. A regression
analysis is ihen conducted using the bank office/s responses (i.e. default/non-default)
as the dependent vadable and the financial ratios and other data as tb€ independent
variables that are attempting to explain the ofncels iudg€ments. As a result of the
analysis we might, for inslancq enive at a model of the bank loan offrcels judgements
in th€ form of the following simplified equation:
Uk€lihood ofd€faulvnon-default = a consrant term - 0.t5 profr + o.25.ash flow + o.50 debt
to equity ratio +... oth€r informarion cues ... + enor

Each ofth€ hodel represents the rclativ€ importance


b€ta weiShts in our regression
ofrhe information cu€s to (h€ bank loan omcer when making a judgement. so, in our
equation above, the debt to equity ratio is viewed as being th€ most imPortant cue to
the bank omcer b€caus€ it has the highes( beta weight, followed by cash flow and then
prc6i.
The Brunswik lens model is a very powerful tool for helping us understand decision
processes in very specific situations. Think about the warrs the mod€l in 6$r€ 13.1 can
be used to improve decision making. Firs! Iook at th€ left-hand side of the dia$am,
that is, the relationship between the event of interest (defauh/non-default) and the
variables being us€d to ass€ss that event- On€ set of lers studies can examine the
relationship bet!,reen the event and the information cues being used by the decision
maker to see whether the 'right' cu€s are being used; that is, whether the Pieces of
accounting information ar€ useful pr€dictors of the event. If the information cue! are

450: PART 3 AccountinS and research


not relevani to the decision, then decision makers can be informed ofthis and taught
which pieces of information are more helptut. Now consider the riSht-hand side of the
diagram, that is, the relationship between the decision mal€r and lhe information cues
A Brunswik lens model study can be conducied to learn how decision makers actually
use ihe accounting information cues and whal weighting (imponance) they assign
to each cue. On the basis of our first set of studies w€ might find that th€ decision
maker is not lveighling the cues corlectly- Perhaps, for instance, too much emphasis
is plac€d on the profit 6gure when our anatysis of the information cues suggesG that
more €mphasis sho ld be Siven to the debt lo equity ratio and oPeraiing cash fiow
information instead. Armed with this informarion, we could then tmin decision
makers to chalge the weighlings rhey give to the differenl information cues to improve
th€ accuracy of iheir judgements. In addition, we mighi discover that there are Pieces
of accounting information useful for making decisions that the decision maker is not
pres€dly using. Again, dris informatior could be used in training sessions to improve
the decision makels performance. ln general, rhe use of th€ Brunswik lens merhod has
led to the discovery ofvaLuable insiShts regarding:
. pattems of cue use evident in various taskl
. weishts fiat decision mak€rs impli(itly Place on a variety of information cues
. the relative accuracy ofdecision makers of different exPertise levels in predicting and
evaluatlng a variety of tasks
. the circumstances under which an e\Pen syst€m and/or lnodel of human behaviour'
outperforms humans
. the srabiliiy (consistency) ofhuman jud8ement over time
. the degree of insight decision makers possess reSarding their pattern of use of data
. ih€ degree of consensus displayed in a variety of group decision tasks.
Because this information is usetul in undersranding decision making processes,
researchers have been (and still are) 1rl'ing to detennine the €ntire decision mod€l or
the decision processes used by various classes of users

Process tracing methods


As described later iD this chapter, models of decision making derived from using the
Brunswik lells model usually have beelr found ro have very good Prediciive Powers'
Indeed, a conparison of lhe predicliors of a lens model equation and the decisions
of a human decision maker usually shows thal the l€Ds model is a b€iter Predictor of
ihe e!'ent of interest than the person from whom ihe nodel was derived. one of rhe
reasons for this is that the slatisrical l€ns model removes nuch of the random error
thai creeps into human iudgemenl olvin8 to such things as tiredn€ss, illness or lack of
concentation. Howevet one impofiant limitation of the Brunswik lens approach is
that it is not a good descriptor of horv people actually make decisions. Th€ use of an
equation format implicitly assumes that the decision maker is able lo simultaneousll
process all the informalion items, but the maiority of decision makers rePo( that rhey
analrne problems in a step-by'step Process, looking at one Piece of information firsi,
assessing that, moving on to the ne,\1 piece of infotmation and so on unril a decision
is reached.
Although having a model that is a good prediclor is very impofiant, researchers
and practilioners also $'ant to have an er'?lanation about how a de(isior is made An
eriplanation for.he d€cision can help reveal a'ealoesses in the decision process that can
then be removed with iraining and other means of impro\€ment. Thes€ improvements
$'ill intum lead ro better predictions thar before.In an attempt to captur€ the st€P{ise
approach 10 decision makinS, some HIT researdrers have used a diff€rent approach to

CHAPIER 13 Beha!io!ral resea(h in a.counl nB 4!1


:q
modellinS decision makinS caued 'proaess racing' or 'verbal prorccol' methods.r3 In
process facin& the decision maker may atain, for instance, be tiven a series of case
studies to analyse but rhis time is asked to verbally desdibe each st€p gone through
wh€n maHng the decision, These verbal descdptions are recorded by th€ res€archer and
then alalysed to produce a 'decision tlee' diegram to represent lhe decision processes of
the decision maker. Figure 13.2 presents a h)?oth€tical decision tree model of a bank
loan officert ddautvnon-default iudgemmts. Each .ircle (or'node') of lhe decision
u€€ contains a question related to a step in dre decisiod process- Depending on the
answ€r to the qu€sdon in a node of th€ tre€, a decision miSht be reached (the square
boxes in fgure 13.2) or a further step is made towards reachhg a decision. So, for
exampl€, in figrue 13.2 the frrst step of the bank offcer's decision proc€ss is to ask
whether the debt to equity mtio is $eater than 3. Ifthe answer k 'yes', then tre officer
immediately condudes that the loaD applicant will defauh. If the answer, howev€r. is
'no', tlren rhe offcer Soes to the next staSe of the decision process and asks wh€ther
the company's size is Sreater than $10 million. The answer to this question then leads
to funher questions and so on down the tree until a iudgement is reached about the
likelihood of the loan applicant defauhing.
ln geneml terms, the d€cision t€€s derived from prcc€ss rracing methods are
intuitively Sood descriptoB of people's decision processes. Howevet, rclative to
Brunswik lens models, process uacin8 methods are not always good predictors of
the ev€nt of interesr. On€ reason for this is that decision makeB often have dif6culty
eplaining all the st€ps they 80 thro€h- This is particulady true of talks lhat decision
makers do routinely and often because the task becomes so familiar tha! rhe decision
process€s ar€ impliqt and unconscious in the mind ofthe decision maker.r{

rlCURt 13.2 Hypothelical decision tree fora bank loan ofiicer

some researchers have tried to overcome th€ g€nenl limitations ofboth the lens and
prccess tracing methods by combining the predictive and descdptive powen of the two
approach€s. On€ such ah€mative is a statisiical technique known as 'classification and
regression tr€es' (CAFI) thar uses stalstical methods to partitioo (or split) the output of a
decision mak€1s ,udgements into decision 'nodes' that maximise th€ power of the model
to correcdy predict th€ dassificatior of different cases into th€ dght t?e of d€cision.

452 PART 3 Accounling and research


Howiesonls used the CART merhod to model the share action recommendations
('buy', 'buy/hold', 'hold, 'hold/sell', teu') of three Australian investm€nt analysts,
using accounting and oth€r information taken from rhe company reportswritten by rhe
anal'sts. Figure 13.3 shows an exiract ofthe CART decision tree derived for one ofthe
anal)sts.

ls NET
< 17.4?

f- o.oJY'"n!o --__l f-i,?',1---_l


f- s45r/
ls EPs -- r- ls SIZE r- -r h NET
,t.*,-
tlsSIZE-
I
.l<NFT.
| I
sELL
=
I

NETr
[r,):1-] I =ruo,
s BK/MKT,
I

t|t
r

t|t
le
.r rsr i:%",Ti,'t tl:',i"',t
l<rsztl lca',l
i
J
i
i
HOLD/ HOLO
SELL
I <LB?
tt
tt
HOLD/
BUY
I

BUY
tI
|.]',o,tLl t'

iii
I

i
i IHOLD 8UY

l(ct: NET = 0crcentrBc chanse in netprof t


DIVYTD
5rzt fih sizc ftarucdas markct capilalisation
cu(enl price to enimaled earn nBs rat o

B(/MKT

FICURt 13.3 Extra.t of CART decision tree for an Australian share analyst
sour.€: 8A Howieson, 'A s(urny.nnlyst s,ction recommendirionr: an ippli.ation oi recuEive
pnrr tion fg to nrodeL .g tldSeren r', / uir/a/,.ai /o!"a/ ol M 8€mear, vol I ij, fo. 2, De.emb€r

'Ih€ diagram shou's how va$ous accoundng variables were used by the aDalyst to
affi\.e at bq', hold or sell de.isions. The CAKI Dodels combired stong power to
colr€clly classify analysts'recommendations with an inruitively appealing descriptor of
their decision proc€sses. Ilo$,€\€r itwas lound that the mo.r data available foraDalysis,
the more complex the r€sult;ng decision trees, makiDg it more dificult to deril.e simpl€
rules lor training other analysts. lhe relativc complexiry oi decision ir€es has been a
common problenr no matter whether process iracing or CART methods have be€lr used
to deri!,e them.

Probabilistic judgement
The probabiiis(ic judgement model is useful for looking at situations in ac(oundDg
*'here initial beliefs aboul a prediction or evaluarion need ro be revised once funher
evid€nce becoDres availabl€. An in\'€sror's revision of an invesnrent decision in the
li8ht of ne$, eviden.e concerning rhe ourcom€ ofa laNsuit againsr a firm is an exarnple
of this t).pe ofsituation.
l-his model a.gues that ahe 'normatively' correct way to revise initial beli€fs, stated
as subjecrive probabiliries, is by applying Bayes's rheoreDr, a basic renet ofconditional
probabilitv tlreory. Bay€s's theorem states that rhe levised (posrerior) probability in the
light ofadditional evidence is equal to the original belief (base ra(e) muhiplied by the

CHAPTTR 1l Sehnrioural r.tin.ch in dccounling


amount try which p or oeectations should be r€vise4 that is, by the informativeness
or diagnosticity of the new data.r6
Thus:

Posterior odds = Likelihoodratio x Priorodds


G€vised (amount by which (initi.l probability or
p.obability) prior exPetlations base rate)
should be r*ised)

Thus acrual, intuitive assessments of probabilities, revisions of those probabilities, arrd


choices among altematives are each compared with those prescribed by a formal model
or'optimal answer'.17
The model has been extensively res€arched in rhe area of psychologt while the
model has a certain logi.al appeal, the body of research suggests that human d€cision
maken are not good intuitiv€ stausticians,ls Research specifically involving accountants
and auditors gen€rally corcurs with this finding.te The evidence from numerous studies
suggests that accounknts and auditors invoke a series of 'rules of thumb', because ofthe
complexity ofthe types ofiudgements th€y need to make and their own information-
processing limitations. Probabilistic judgement can be used in a range of decision
settrngs. A discussion of the evidence ftom studies involving probabilistic iudSements
in accounting and auditint follows later in this chapter.
A non-accounting €xample, drawn fiom Libbt'?o might helP clariry the normative
applicatioD of Eayes's theorem-
SuppGe rhat you are in charSe ofsecurity for a large depanment store. A r€cent audit
has indicated that losses owing to €mploy€€ dleft have riseo ro 10 per cent of sales ln
response to your supe.ior's .oncem, you institute a mandabry lie detedor screering
progran for the emptoyees. Pr€vious res€a.ch has shown that:
. 2 p€r €ent of employees steal
. the probability that the employe€ will produce a 'deteptive' respons€ on lhe lie
detector ifh€ or she did itr fact li€ (tru€ positive) is 0.9
. th€ probability of producing th€ 'de(€ptive' r€sponse ifthe enplovee did not lie
(false positive) is 0.12.

Qustion
If rhe €nployee produced a decePtive resPonse to the question, 'Did you steal flom
$e storel'what is the probabilitythat the €mployee did in fad steal?
- NorEative Bays's colution
Posterior odds = Likelihood ratio x Prior odds
= 0.90/0.12 x 0.02/0.98
= 0.153/l
Probability =0.i53/1.r53
= r3tk
Libby argues 1hat most people suSSest that the Probability is quite high, around 80 p€r
cent.?l
The findin8 that human declsion makers do not apPly Bayes's multiplicative rule when
revising prior beliefs is pervasive iD th€ HIT research lit€nture. This led researchers in
psychology to investigate what rules, ifany, are used. They found that the explanation
seems 10 lie in the use of n-lles of rhumb or biases, which humans reson to in order to
simplify complex tudgement task.

',454 PART 3 Accounti.g and research


Three categories ofrules ofthumb (knorvn in the lilerature as'heuristics') have been
identified in the psychology literature representativeness, a ilability and anchoring."
As previously mentioned, financial accounling researchers have accumulated much
evidence that these rules of thumb or biases are aiso peffasive among accountants,
auditors and users of financial statements. That is not to imply that these rules of
rhumb are necessarily d,$functional or that they automalically lead to poorjud8ements.
Rather, it may be that they represent an efficient and effeciive method of dealing with
compla\iry and the limiradons of human cognitive Proc€sses. A later section of this
chapter explains how each ofthe three identified rules ofthumb operates, and reviews
evidence reSarding the exislence and effecb of€ach in an accounting context.

Lens model studies the evidence


-
Many srudies have used the lens model framework to examine the accuracy ofhumans'
predictio$ of business faiLure. This task is important and realistic for People such
as i estors, bank loan officers, other lenders and auditors. It has generally been
researched by providing the subject with a number of numerical cues across repeti(ive
cases o{actual business success and failure, drawn from archival data. Thus, in this task
[as opposed to others described later), a',:orrect solution exists as a bendrmark against
which to compare human performance.
Using the lens model as a researci lool in this way allo s analysis of the consisteucy
of judgements, of whether a'model ofhuman behaviour'can predicr more accurately
than a human. lt also enables analysis ofthe ability ofthe cues to predict the event in
question (the 'environmental predictability' usinS an ideal cue weiShtingl ln addition,
it can give insights regarding the degree ofconsensus between decision makers
The 'model of human behaviour' is developed using a mathematical representation
of an individual's pattern of use of the cues. This model is then aPPlied to the cases
in question. Th€ evidence consistently shows that humans are reasonably prolicient at
developing principles or models to solve the success/failure task using fiDancial ratios,
but are outp€rformed wher] their own models (inferred from the pattern ofuse of cues)
are applied mathematically for t\vo reasons: they misweigh cBes, and th€y inconsistenily
appli' th€ir decision rules due to factors such as fatigue and boredom Math€natical
application of €ither the €n'ironmental model (using ideal cue weightinSt or the
'model of human behaviour' is perfectly consistent over 1ime, eliminating random

Libby was th€ first toresearch the lask of assessing busin€ss failure' and several
studies have followed.'?r what has emerged in the literature is the question ofwhether
disclosing to the subjects rhe actual rate of failure is necessary to achieve realism in
rhe task. Th€ actual rate of business failure is very low, less than 5 per cent Therefore
subjects. unless they are told otherwise, bring to the judgement task the expectation
that the number offailure cases will be marginal.
on the other hand, research€rs cannot hope to Sain evidence Fom this task setting
unless a 'reasonable' number of adual failure cases arc included in the materials
provided. Researchers hav€ generally used a failure rate of berl\'een 33 per cent and
50 per cen! and experiNents manipulating the rate of failure and ils Prior disclosure/
non'disclosure have been carded out.:r lhe results have been inconclusive25 ir lerms
of rhe e{ten1 to which prior disclosure of the sample failure rate matters, bul it seems
that rask predictability'z6 and information rePresentativeness play a role'?7
other variations of this line of research include obseNing the effect of allowing
subiects to choose their own ratios, exanining the impact of information overload,

CHAPTER 13 Behavioura research in accounl ng 455


and analysing the confdence level rhar decision maken place on rheir iudAement and
whether confidence influences accuracy. AbdelKhalik and El-Shesbai concluded thar it
was the subject's choice ofinformalion, ralher than his or her processing of ihe chosen
cu€s, that limired accurag.,s Simnett and Trotman found tha! although subiects
w€re able to use all the information from ratios they themselves selected, they were
unable to improve perfoimance when asked to apply an ideal cue-weighdng model.
These aurhors conduded rhat wh€n subiec(s are uDable to choos€ their own ratios th€ir
informalion-processing performance declines.2e
The information overload literature has impliGtions for rhe presentarion and
disclosure issse in financial accounting. It provides evidence of lower cons€nsus
and lower decision-making consistency for individuals experiencint overload.3o h is
thought that, as the amount ofinfotmarion ircreas€s, initially the use and integration
of the informatioD iDcreases- However, beyond some point, additional infomation
results in a desease in the amount of information integrated into the decision-making
task.3t Che$ning and Haftell, in a financial distr€ss prediction task, found evidence of
ihe above theory once subieds were given more rhan 8 cues (financial ratios).3, Libby
noted that *rc addition of less lid cues to a set containing more valid cues decreased
performancei33 however, other studies have detected no such relationship,3a
Ovemll, rhe literature on infordation overload has produced inconclusive results.
One reason for the lack of dear outcomes across these differ€nt studi€s is that most
researchers have not sought to der€rmine whether the additional data provid€d were
actually 'informativ€' (i.e. r€levant to the decision ar hand) in the 6rst place. Furthet
there hae been litde attempt to see whether decision makers acrually used lhe additional
dala provided by the researcher.rs
The jud8emeDt confiden@ lirerature has consistendy found that both eJee( and
non-expert subjects are overconfident oftheir ability in specific judgement rasl$.36 This
overconfidence seems to siem fiom three factors:
. the tendency for humans to seek out and overweight positive feedback
. the limiled nature off€edback in many in$ances (e.9. in failure or disrress prediction
the corectness of a decision not to lend is rarely evaluated)
. fie interdependency of actions and outcomes (e-g. the act of lendin&hot lending
hself influences success or failure).37
Libbya and Zimmer3e found that the ac.uracy ofiudgem€nb iDcreased wirh inceasing
.onfrdence, bur other studies have shown dlat confidence isnotrelated to accuracy.a0
After examining th€ lens model evid€nce collected ftom many decision task,
including the prediction of failure. Libby sums up the findings of tbis category of
research, as follows:

In many importam deosion-making situations, the environmental predictability


of availabl€ information is low Howev€r. even in situations where €nvironmental
predidability is .elatively high, poor judgemental achievemenr is the norm.
Both human inconsistency and miswei8htin8 of cues conrribute 10 rhe poor
acltievelnent. CombininS quantitative information in repetitive task does not app@r to
be a function that peoplep€rform well TIus, in $€se situations, replacingpeople ith
models (e.9. environmenhl regression nodels, models of man, and €qual weighting
models) sholcs promis€ for increasing predictive a.curacy.at

Although thi$ statem€nt rvas made in 1981, the evidence gained since then has not
violated the thrust ofLibby's observation-
Theory in acaion l3.t provides an exampl€ ofhow accountin8 in0uences behaviour
by investors. This media rclease illustrates the Srowing influenc€ of the AustraliaD
Shareholdels' Associarion.

i PARTS Ac.ounting and eiearch


The power o{ inter€st group perceptions

ASA seeks to remove NAB director


The Aus$alian Shareholders'Association (ASA) today caLled for sharehoLder suppo'1 for a
resolutjon to remove Mr Paul R zzo, Chair of Nationa! Austra ia Bank (NAB) Risk Committee,

iohn Curry, Chairman olth€ ASA'S Company Monitoring Commjttee in Victoria, sa d "The
NAB Risk Committee has failed ln its oversight ot the bank's rlsk managemeni, as evldenced
by the si8nificant provGions and hedging costs $,hich the bank has incurred in relation to its
pottolio of collateralized debt obllgations".
"Mr Rizzo, as chairman ot NAB'5 Risk Committee, should be held accolntable Ior th s

The ASA has commenced a campaign lo obtain the 100 signatures necessary to movc a
reso ution atthe NAB ACM on 18 December2008forthe removalofNlr Rizzo. MrCurry sald,
"We have emailed our mernbers today and wl I be giving NAB shareholders the oppodunity
to sign the reque( at ASA meetings-"
"Civen the coniact we have been receiving from disgruntled NAB sharehold€rs, w€ do not
expect to have any difficulty mising the required 100 signaturcs."
Upon obtainin8 the 100 signatures the ASA will request that NAB place a resollrtion
before the ACM seeking the removal of Mr Rizzo, and circulate a statement ln support of
the resolution to all shareholders with lhe notice of meeting. The statement the ASA seeks to
circu are in accordance with Secilon 249P oflhe Corporations Act 2001, is below.
A previous rcsoLution moved by ASA membeB in 2005 to dismiss a NAB director, Cco{frey
Tomlinson, received 39% ofthe vote.
National Australia Bank Limited (NAB)
Statement pursuant to Se.tion 249P ofthe Corporations A.t
National Australia Bank's Rlsk Committee has fail€d to adequately consider the underlylng
value ot or risks invo ved in, the significant portfoiio ol collateralised debt ob lgatiorrs (CDOs)
held by the bank. These asseis have resulted in la€e provlsions that have significant y reduced

whi enlire board i5 responsible, Mr Paul Rizzo, as a director and chajrmai ot NAB's
st the
Risk Commiltee, should be heldaccountablelortheweak riskmanagenrentculturewh ch has
led to the loss of shareholder vi ue and bank credlbility in re atjon to thcse assets.
when ofnolncin8 the $l8l nrillion charge against lhe valuc of its port{olio ol CDos in
March 2008, NAB claimed to be avoiding high rlsk bu5lness.
ln N.lay, the chajrman of NAB, Mr Mich.el Chaney, wrote ro shnreholders stati g "1he
volatiliq,inworldfinancialmark€tshashighightedtheimporranceofrobustriskmanaSement
syslenB and prudent capital management in financlal institutions. The efforts of our people
to strenBthen o!r risk framework dnd balance sheet in recent year has been validated by the
companyrs resil ence durinB fie recent p€riod oi sign ficant iinancial di,rupt on".
ln Julv the bank announced a iLfthet provis on of $8309 mil ion against the va ue of this

ln September, NAB announced .dditional co5t5 of $100 mi llon in 200B, $60 million per year
for s tears afd a lesser ycarl) amount thereafter to hedg€ againstdefaultolthe cDo portfol o.
ln addition NAB still has a substantial poldo io of CDO5, corporate bonds and comnrercial
mo{8age backed securities that are neither pro!Gioned nor hedged.
Shareholderu will have an opportunity to hold Mr Rizzo accountabLe fo. this failLrre of risk
manag€mentby supponinSa resolution to remove him as a directorofNAB at thefo hcomlng
annua general m€eting.
lf you are unable to attend and wlsh to nominate the ASA as your rePresentative at that
meetlng p ease inse'1 lhe name "Australian Shareholders' Associat on" as your proxy of the
proxv Iorm and lodge it in accordance with the instructions shown on the torm itse 1..
solr.er Nredia Re ease posled b,v r\urt'a an Sharcho.lcr' A$o.iation, I O.robcr 2008, r'ww xa sr ar IZ

CHAPTTR 13 Behav oural resear.h i. accotnl iE 457


Qu€3lions
1. On what typ€ of information would the ASA decide that a company board or chai.person
should be censured or removed for poor performance?
2. What is motivating shareholdeB to operate as a group to censure or remove company
boards which do not appear to be managinS the company in an appropriate fashion?
3. k the ASA concerned abut corporate Sovernance or raisinS its own profile to incr€ase
membership and r€lative power? What type of value judSements did you make to answer
this question?

Process tracing studies the evidence


Brunswik lens models and process
-tacing style studi€s are different technologi€s
wirh rhe same objective of modelling decision processes as completely as possible.
MenuoD has already been made ofthe main differences between thes€ two modelling
methods. Brunswik lens models implici(ly reat the decision prcc€ss as a simple
linear combination of th€ informadon cues whereas the decision trees d€rived from
prccess tracing acknowledge the step-by-step nature of decision makio& in hich
the information content of one piec€ of data interacts wirh that of other piec€s
of data. The maiority of studies which have inveati8ated the linearity of decision
makers' iudgementsa, conclude that the assumption of simple linear combinations
of informarion cues is iustified but some studies in busin€ss contexts43 have found
evidence of statistically significant interactions between information items suggesting
that process iracing methods are advantageous mod€lling techniques for r€presenting
decision making in some conrexts.
The relative predictive power of lens and process tracing models has been studied
in business conlo.ls by Larcker and l€ssi8aa and Selling and Shank.as In a share
selection scenario, hrcker and trssig found that proc$s tracing mod€ls outperformed
the statistical linear models, but Selling and Shank found the opposite when the two
approaches were compared in a task iovolving the prediction of bankruptcy- These
differcnces in the studies may rellect our earli€r comment that the typ€ of decision
tesk requires different styles of decision processin8. As usual. the compl€xity of human
decbion making means that more in-depth research is ne€ded to understand what t ?es
of decision task charactedstics determine the most appropriate style of information
Processing.

Format and presentation of financial statements


In 1976 Libby observ€d thal three basic options existed for improving decision
making:
. changrng the pres€ntation and amount of infomalion
. €ducatin8 decision makers
. replacing decision makers either with a model of thems€lves or with an ideal cue-
weishting model.a6
civen the imponance ofthe 6rst slrggestion to accountants, auditors, regulators and
standard s€tte6, surprisingly littl€ research has been undenaken in as.ertaining ideal
accountinS pres€nta(ion formals. '[he $udies that do exist have tend€d to examine
radical changes to financial statement presentation in lhe form of multidimensional
graphics (9€e figure 13.4)- Howevet in response to calls in the litemture, resear€hers
have retumed to the issue of ideal pr€sentation of the more traditional table format.

458 PARI3 Accounting and research


The information load literature is also p€rtineni ro the question ofimProving decision-
making capabilities of users of fi nancial statements-
The lens model is useful in e{amining financial statement presentation issues as
well as in anatysis of predictive judgements. h permits analysis of human iudgement
accuracy in t€rins of determining ihe extenl to which the individual detecls ess€ntial
properties ofthe judgement task and consistentiy applies judgement policy. If changing
the repo( tormal ofinformation results in improving either of the above charact€ristics,
human iudgement accuract should increase.:lhe decision-usetulness objective, adopted
in the conceptual framework, d€pends Panly on the user's ability to interPret the data
for a given in'estm€nt or credil decision. The imPact of a change in the repoft format
on subiects' ability to detect a change in th€ financial status ofthe lirm can similarly be
e{amined lvithin a lens model framework.
The multidimensional graphics most rcsearched have been in the forn of schemati.
or Chemoff faces as iLlustraled in fisure 13.4

1969 1970 1971

'1972
1973 1974

FIGURE 13.4 Chernoff face. tepres€nting changes in financial .ondition


sourcer D Sr.ck a.d CJ WaGon, llum judgenrent accurac,v: md td nrens onn SrJphi.s.nd Irlnun!
lerjrs modc s', /olDal oiA..ounri.g AeJea..h, Spri.81931, p 202

The faces are construct€d by mapPing transformed financial variables onto facial
featLres. Mathematical precision in terms of nose length brow angle and mootir
curvature isusedto iepr€sent changes in financia I posltion from one period to the nej':t
Interest in this mode of prcsentation arose io 1979 wben Moriarity reporl€d that
subiects usins financial information represented by such graPhics outperformed a
$,ell'accepled financial distress model.aT As previously noted, it has been a consisreni
research finding that models generally oulperform humans and so lhis form of
presentatior offered promise. Stock and Watson found only weak confirmation of
lvtoriarity's r€sults in a predicrion of bond ratings task aB Howevet they did lind that
accuracy performance using mulddimensional graphics was superior to thal artained
with a conventional tabular presentadon. This latter result held even when the subiecr
had minimal accounting training. Sto& and Watson concluded:
... a muhidimensional Sraphi.s approach would be usetul when€ver cost or dau
availability nakes good staristical models impossible to build, especiallv if the resulls
ae
using multidimensional graphics are at least as good as the rcsults ofa model
To date, financial stalement pleparen have not been prepared to pLrblish graphics as
ndical as the Chernoff fac€s. Howev€r, the us€ ofcolour and more convenlional graphs

CHAPTTF ll BehJ! ouril r.\ear.h jn ...ounr r8


is common. R€search€rs ilr the fields ofstatistics, psycholoSy, information syster$, and
educatiod have investigated the rclative advantages of!'arious Sraphic and tabular fonDs
ofvisual presentation for displaying both fnancial and non-financial informadon. To
datq results are contlicting and queBtionables0 but a common fioding in accounting
contexts is that pres€ntation does influence decision mal'jng.
For elample, Blo.h€r Mofne and zmud5r investiSated ihe effects of different forms
of prcsentation (tables and colour graphics) on accuracy and bias of intemal auditom'
d€cisions. They found that the relativ€ efTectiven€ss of different forms of presentation
is a function of the amount of information that is presented to, and mu:it be processed
by, the decision mak€r. Graphic reports seem better for low Ievels of.omplexity ard
tabular reports for high levels of complexity.
Davis, using MBA studems, invBtigated rhe impact of three Braphical f;orma$ of
financial statements (line graph, bar cha( pie chart) and the conventional table. That
srudy found thal the qu€stion the decision maker souSht to answer from the statements
and th€ foms of presentation interachvely afected performance. No one form of
prcsenlation was best in all situations.52
Firancial forecasting was the task Desanctis and Jarvenpaa chose for assessing the
impact of bar graphs comparcd with tables. They found only a modest improv€m€nt in
the accqracy of forecast judgements associated with graphical formats and then only after
pnctice in using ihese formats was provided to subjects. This is a somewhat surprising
finding given the alleg€d vaiue of graphs in detecting trends and r€latiomhips.s3 The
auihors wam that when accounting data are present€d in a graphical format, users
may to througb an adrustment or leiuniBt process before the gaphical information
becomes meaningful.
In an auditing coDtexL Ricchiutd4 found that iudgements r€garding necessary
adtustments to the accounts may be affected by the mode of presentation of informalion
to the auditor: visual and/or auditory. Since most audit res€arch presents vwitten
maredals ro subjects, h€ warned that this nnding may threaten the generalisability of
results (applicability beyond each research setting).
Recefl rcsea(h confirms that the impact ofdiferent prcsentation formats and styles
remains a complex area lhat needs furth€r context-specific investiSation. In two studies
based on a bankuptcy prediction scenado, So and smith investiSated the impact of
colour graphics, gendet rask complexity, and different presentation formats on the
predictive accuracy of a sample composed mainly of undergaduate business students.
In one of the studies5s the decision makers used either colour or black-and-white bar
chans of financial ratios as rh€ basis for their iudgemenb in different tasks that were
either high or low in complexity (as measured by the level of inconsistency in the
financial ratios). The results sugge$ed rhat coloured graphics wete not effeciive when
the lask was complex and thar, interestinSly, females made more eff€clive use of colour
than males. So and Smith concluded that more resear.h needs to be conducted on the
relationship between colour and information proc$sing to aid in understanding the
resul$ they found.
In anolher srudy,56 So and smith obtained results confirming the call by Desan.tis
and tarv€npaa mentioned earlier fot more education on the use of graphics. So and
Smith ask€d their d€cision mak€rs to work with one of the following sets of data: a
combination of tables and bar.hans, or tables and chemoff faces, or tables only.
They found ihat in situations where the complexity of the information was hi8h, the
us€ of lables alone led to the highest accuracy, suggesting thar lhe use of EraPhics and
pictorial representations of data actually led to a decrease in the effecliveoess of us€rs'
deosion making. One of the reasons for this might be the desire of decision makers

460 PART] Accounring and research


to.hoose the easi€r option when situations are complex, but graphical and pictorial
representations of data are often more abstract and less detailed rian the infomalion
found in tables, These findings are a concern in an age $/hen many lirms are turning to
the Internet and multimedia-slyl€ presentalions to comnunicale wirh stakeholders.
These examples demonstrate that Wainer and Thiessen's claim, tha( ihere is no
well developed and tested lheort thal can be used to specify the rircumslances under
which different forms of presentation are most appropriate, remains rrue.e7 That this
silualioD still holds is probably, at least in par! attribulable to the (onceruarion of HJI'
research in an auditing cont€xt over rhe last decade. [luch ofrhat auditiDs research has
taken place $'iihin the probabilislic judg€ment framei{od! the evid€nce from which is
disossed in the next seclion.

Probabilistic judgement studies the evidence


-
in many accounting contexts and especially in auditiDg, there is no'(offect'solution
with Nhich iudgemen{s can be compared ilr order to assess their acc racy. One rvay ol
coping wi(h this lack of a b€nchmark criterior against which to judge performance is
to examin€ the degree ofconsensus regarding a particular decision across a number of
d€cision makers. Another rvay is to use a mathemalical orstatislical model. As discussed
in an earlier s€clion, probabilistic i dgement research is based on analysis ofvhether
humans revis€ their beliefs in line wifi Bayes's theorem once new evidence becomes
available. HII research within ihis model has consistently demonstrated that humans
possessing a i?riety ofskill levels and, obsen'ed ovrr a variety oftask, revise their prior
probabilities to a lesser enent than Bayes's lheorem prescribes. this conservadsm has
been attributed to fie use of rules of thunrb and biases which nre adopted as a meaus
of simplifyhg complo{ iudgemems in order for irumars to copc.
Three rules ofthumb ar€ d€fined in (he literaturc as folloi{s:
. Rqresentatiwness. This rule of thunb slares ahar wh€n judging rhe probabilir). rhat a
pafticular item comes from a panicular populaiion of items, people's jridgement will
be determjned by tbe e\tent to Nhich the iten is representative of the population.
Items or ev€nls that are \.ierved by the decision maker as beiiS more representative
\.vill be assessed as having a higher probabiliry of occurrence rhan those that are
less represenlative. For instance, a bank loan oflicer nay jt-rdge the likelihood that
a company $'ill default on it$ loan by horv similar h is to a stereotlpical failed firrn.
As descdbed larer. research has sho\\/n that the use ofthis rul€ ofthumb can lead ro
poor decisions because decision makers who use it ofteD ignore other relevant dam
that are not part ofth€ representatrve stereoilTe. On€ exanrple ofthese effors relales
to rhe rate at which the evcrrt ofinteresr actuall!'o(curs in the total population (rhis
is called $e'base rate'). 1}€ populat;on rat€ ofloan d€fault might be, say, 5 per cent
but rhe decision maker might ignore this important informalion and consequeotly
overctale lhe rate ofdefaull M,hen evaluating: sanlple of coI)panies mainly because
he or sh€ is looking onl.v for companies that have characteristics thal corr€spord to

. lraiiariliry. The availabilhy rul€ ofrhunlb refcrs ro the assessment ofthe probability
ofan event based on the ease with $hich instances ofthat event come 10 mind 53 1he
consequence ofuse ofthis rule ofthull)b is that probabilities rclated to sensational'
even!s are likely lo be overestimat€d.
. Atlchotin| and adjustmrrr. 'l h is rule o f to a general judgelnent proccas
rhunb rc iers
in which aD initiall_v gen€rated or Siven responsc anchol and o(her

(lHAPTER 1:J lJ€r.!iouri r'nrr.h nicco!fr nB 461


,q
information is used to adiust that rcsponse.se'Ih€ consequ€nc€ of this rule of thumb
is th€ possibility of insufnci€nt adiustment in the light of changing circumstances.
As previously stated, most of the research into probabilistic reasoning has used
auditors as subiects. Auditors are ideal subiects for the study of human iudgement. and
prcbabilistic reasoning in particular, since many audit iudgements entail the need to
revise a$essments in lh€ light of additioflal evidence. In addition, in the United Slates
the larger public accounting firms have be€n ortrcmely cooperative in making available
both funds and practising audilo$ for research.60 A brief oveNiew of research related to
all three rules of thumb follows.

@-" nrenrsENTATtvENEss: THE EVIDENCE


Kahneman and TversMr 6rst reported the existence of rePrcsentativeness and the
tend€ncy to ignore base rates. Since theo, research in both the psychological and
accounting fields has iN€stigatd the phenomenon. The evidence is incondusive
in that it shorvs base-nte informatioo is sometimes neelected and sometimes used
appropriately in assessing the probability of an event.62 Th€ use of base-rate information
seems to be highly sensitive to a \€ri€ty of tasks and contexts, and this has led to the
hlpothesis that probabilistic reasonilg involves contingent Processin8.63
Joyce and Biddle6a used an accounting adaptation of the employee theft/lie detector
ev.ample previously given in this chapter to illustrate th€ aPPlication ofBayes's theorem.
In this exampl€ relat€d to managem€nt ftaud, again a v€ry low bas€ rate was used. 11
was €xpecled that, as in the previous examplg subi€cl$ would pay iNumcient att€Dtion
to the low base rate and, therefore, have too high an eQectatron of ftaudulent activity-
Practising auditors were the subiects, and results indicated that' although th€ auditoB
did not ignor€ the base mte given. they also did not consider it sufnciently. Since audit
judgements often involve low-base'rate, high-impact evenrs (e.g. ftaud), the aulhors
were concemed about the imPlications of this lesult. However, Holt6s cast doubl on
the fndings of ,oyce and Biddle, arguing that it was the wording of the Problems,
leading to a ftaming effec! which had driven their results. Framing effects are defined
as a cogniriv€ persp€ctiv€ elicired by tbe task characieristics.66 They are often illustrated
by use of the half-empty/half-tutl Slass analoay where a pessimist r€gads a glass as half
empty and an optimist rcgards the same Slass as half full

Availability: the evidence


The basis of this rule of thumb is that likelihood iudS€ments are based on retrieval
ftom m€mory of relevant instances or construction of Plausible sc€narios The more
instances recalled or th€ great€r the ease with which one can recall instances or
generate plausible explanatrons for an even( the higher will be the iudged prcbability
of occurrence of the even!.6? However, this requires a large sample of probabilities to
improve prediction a(uracy.
Moser6s examined ihe availability rul€ of thumb in connectron with investors'
predictiv€ iudgements. He asked half of his 58 subjects to list rcasons the target
company's profits would increase and then reasons the comPany's profits might decrease
.Ihe
iemaining subiects listed reasons in the oPposite order. Moser found that the 6rst
group ofsubiects made predictions of higher probability re8arding an in.rease in profits
for th€ compan, despite th€re tteing no obiective basis for their oPiimism The outcome
for which the subjects were able to generate the most supPoning leasons lras iudged
more probable- He concluded thai some envrronmental €vent reSarding a particular

:...462 PART 3 Accounting and research


company that caused disproponionale news coverage might systematically influence
predictive iudgements. Individual investors might either overP ce or underprice shares
because ofa tendency for all to think abour the comPany from the same perspective
optimistically or pessilnistically.
-

Anchoring and adjustment: the evidence


loyce and Biddle6'again Lrsed practising auditors as subjects in examining t}le effect of
changes in internal control systems on the ertensiveness ofsubstantive test (audit tests
desigled to search for the existence of dollar erron in the accounts) li was expected that
the subiects would adiust for changes in internal controls by adjusting audit scope but
that th€ adjustment would be insufficieni as anchoring on the initial internal control
rvould occur. No evidelce of anchoring and adjustment was found. I-lowever, Kinnev
and UeckerTo did find evidence ofarchoing and adjustment in analytical r€view (ratio
analysis) and compliance test (audit rest of internal conrrols) tasks.

Expert judgement and rules of thumb


The res€arch involving exp€( iudgement is concemed with examining the thoughi
processes of erpens and the determinants of erPerise. New€ll and SimonT' provided dre
analldcal framework $'iti theirlheoryofbound€d rationality.They susg€sted thathuma$
possess a short'term memory with very Iimited capacity (a-7 chunks) and a vinually
unlimited long'term memory. The structure ofthese memories and tbe (haracterisrics of
task combine to determine thewaydifferent t}?es ofProblems are rePresented in memory
(cognitive representaiion) which in turn deiennines the way Problems are solved.
Much of the early work on alpen memory involved medical practilionersT'? aDd
chess masters-7r The ability of e\pens to effectively expand rheir memory capacitv in
s uations related to their er?ertise has been a consisient finding in the literaiur€ lt
seems 1hat a small se1 ofcue values brings to mind a larg€r sea ofcues associated wilh a
prototp€. The explanalion li€s in the use of repres€ntativeness to recognise a fan]iliar
patrern and bring a prolot)?€ into shon-tenn memory out of long tem memory.
BouwmanT! found evidence of representativeness in an accounting task when
he ask€d 15 students (novices) and 3 accountants (expe(s) to analyse four cases
contai ng exiensive finan.ial information to derermire any underlving probiem areas
H€ found that the experts follorved a dire.ted strat€gy based on standard check lists,
complex trends and stereoq?es, and developed an o\€rall piclure of the company
The erpeft, when encountering a violalion of the slereoo?e, attenpted to uncover
rhe caus€, searching for borh confirming and non-confirminS evjdence The students,
on the other hand, folio$,ed a simple undirected sequendal strategy, evaluating
information in tie order presented, sealching for a significant fact $'hi.h could explain
the situation. The-v searched only for contuming evidelce and did not atlempt io find
(ausal e{planations.
Numerous studies in auditing have coffirmed that audii experts have better
recall,:s integrad\€ abilities76 and error frequency learning abili1y77 than do novices
Audii €xperts e\hibit eedence of all rluee ru]es of fiumb and it is not clear that illis
?3 The double-€ntry nature of
necessarily results ir inferior quality in decision maknrg
bookkeeping means that audil tests often overlap and mechanisms exist (eg panner
review) which try to assure qLlality.T' Process tra{ing nethods mav be a good wav of
iearning more about the differen.es betweel the decision Processes of e\perts and
nolices. This knor"'ledge $'ould be valuable for trairing purposes

CHAPTER ll B.hr, ouraLrc'tdr'l


'n
r.counr nE 463
a

@- nccouNTrNc AND BEHAVIoUR


of individuals or groups of
Accounting erists as a dircct function of tbe activities
individuals (defined as accounting entiries). There are different viewPoints of
accountin& indicating that there are a number of accounting persPectives possible.
Even in a period of c€ntralised government regulation of accounting disclosures by
firms, there are lhousands of choices and assumptions requir€d between altemative
accounting te.hniques in the prcparation of financial statem€nts fol corPorate entities.
Even under the more stringent regulations ofthe Australian taxation legislation, there
is considerabl€ disoetion in the techniques lhat can b€ applied to the calculation of
to(able income. Th€ main issue is that the kchniques adopied, and the interPretation
of reported information, are matt€rs of perspective. Ttere are many competin8
interests a€ross the %rious PeoPle who interpret hnancial information reponed by
organisations. Ba6ically, lhe us€rs of accounting information rePresent a variety of
perspectives and obj€ctives, ranging from employee groups (unions), individual
sha.€holders and investor groups to the management ofan organisation- Accounting
standard setters have ofien spent considerable tim€ debating the technical validity
of panicular techniqu€s proposed. Howevet even technical validity is a matter of
perspective.
The aim of this section is to reinforce an underlying theme throughout this and
a number of other .hapt€rs of this text: that accounting is a funclion of human
behaviour and activity. As such, atcounting information will influenc€ behaviour,
both in the methods adopled to measure and repo( information, and in r€sponse
to the information disclosed. Th€ r€sponses to information are a function ofhuman
pe$p€ctives aod are therefore not divorc€d from th€ PeBonal aims and interests of
users, whether actinS as individuals or as a SrouP of like interests. Consequently,
accounting operates in a complex envitonment, Accounlants should be (onscious
of thb environment and appreciat€ the impact of accounting information on
b€haviour.
Bu(helt et al. summarised the si8li6cant role of accounting in a broad economic
conl€xt:
Accoundng dala are nolv used in the d€rivation and implementation of policies for
economicstabilisation, pdceand wag€ control, forthe regulation ofparticular industrial
and comm€ftial sectors and the planninSofnauonal economic resources in conditions
ofwar and peace and prosperity and depression.
No lon8er seen as a mere assembly of talculative routines, it now tunctions as a
s0
coh€sive and infiuential mechanism foreconomic and sooal manaSement

In addition, il is important to consider the factors \"/hich influence changes in


accounting systems and tle nature of reponed infolmation. Accounting is not a one-way
proc€ss, with accounting information influ€ncint b€haviour or promprinS a response
on th€ part of some user group. Accounling information and systems are affected
and changed by faciors, most of which are beyond the direct control of accountants.
Ac€ordin8 to Zimmerman, ih€ accoundnS system is a fundamental component of
an organisation's architecture, $'i(h senior managers constantly seeking to adaPt the
architecture to ensure re best structure ofthe firm Zimmerman provides two important
ob$rvations about the faclors affecting the accounti ng system:8t
i. Chang€s in the accounting system rarely occlr in a vacuum Accounting ststem
chenges generally occur at the same time as chanSes in the firm'6 business strategy
and ;ther organisational chan86, paniftIarly with regard to (he Panilioning of
decbion riShis and the pe ormance evaluation and r€ward syslems

464 PARr 3 Accounting and rcearch


2. Alterations in fie 6rm's organisalional architecture, including changes in the
a.counting sysrem, are likely to occu in response to changes iD the firm's
business sl'ategy caused by e{temal shocks from technologv and shifting market

Accounting information derefor€ siSnificantly affects th€ behaviour of individuals


both within an entity and extemal to it. However, the influence is tl\.o-wa, with
individuals [or groups of individuals) directiy and indire.tly affecting the structure of
accounting syst€ms and information disclosures. If behavioural research is to progress
and provide valuable insights into these complex relationships, it will need to diverge
from capiral markets research hich operates under assumptions that represenl the
individual or shareholder group as a set of jnterchangeable blanks (each sharing the
same powers of interPreration, motives and beliefs) and lvhich accepts that economies,
societies and the entilies within ihem are living complex organisms thal do not act in
isolation bu! affect eadr other in complex ways.32

@-.utrlrrnnoNs oF BAR
This oveNiew of BAR has shown that we have learnt a 8rcal deal about holv different
d€cision makers use accounting information. However, it has also revealed that there
is significantly much more for us 1() learn in this ar€a The frequenr (and frustrating)
cortradictions between th€ findings of similar studies simply mean that human
information processing is far more complex than the developnent of current research
fteo es and methods.
Maines has argued:

Unfodunately, three criticisms leveLled against this r€search have limit€d iL, impact
Firs! studies on the same topichave produ.ed conflicting lesults, Preventjng conclusive
guidance for policy decisions. In addition, the o{perim€ntal subiecls and settings
;s€d in these studies often differ from dose found in real iudgment sertings. finally,
accountingresearchers have ql€stioned"h€theI policyshould be inl'luenced bvresearch
on individual decision makers.33
These limitations of BAR have meanl that it is yet to achieve the same level of
dominance in the academic literature Presendy enjoyed by rhe capital markets and
agency theory research schools. O\€rall, the maior limiration of BAR is the lack of a
single und€rlying theorywhich heLps in the uniication olthe diverse r€search questions
and findings of BAR. Unlike the capital markets and agency theory research schools
which have grounded their research activities and theoreiicai development within
a particdlar 6eld of economics, BAR rcsearchers have borrow€d from a multilude of
disciplines and contants and have no common framework with;n wllich to develop
usetul generalisations for poliq' nakers. There is also no signal that the developmenr
ot su l- d rheor) is lilel\ rr Lhe fore{eable furure
Nelertheless. it remains trlre that BAR is a valuabl€ and practical research school
BAR methods have b€€n used bv many groups of decision makers 1o develop eryeft
systems and other practicai tools for information processing aDd training purposes
in the workplace. BAR also offers the promise of revealing systematic errors (such as
rhe {lifferent rules of lhumb dncussed previouslv) made by all decision makers within
specific (ontexts thar have implications for improvemeni at the macro lev€l For
lnstance, research is stardng to show ho$' the q?es of inceDtives share anal''sts fac€ in
the workplace bias their decisioDs.sa For example, they are less inclin€d to make selL
r€commendations than buy recommendalions for (ompanies with {'hich ihey enioy
close business relationshiPs.

cHAPTER l:l tsehaviorral research in acco! ntlnB 465


]t
Crealer intercst is now being shown in rhe d€velopment of non-financial measures
of company p€rfomance such as the environmental and social p€rformance indicators
proposed in kiple bottomline reportinS'. As lh€re is litd€ existing knowledge about
\ rhat these nolr-finaocial measures should be or how they should b€ teported, BAR
has a significant role to play in h€lping accountinS practitione$ and policy makers
in developing €merging areas such as this. Funher continuing ad nces in research
ieclmologies and melhods will enable future BAR researchers ro develop rich€r
hlpotheses and tests of how decision makers process accounting informalion.

@' rssurs FoR AUDrroRs


lust as behavioural accounting research can address questions about how people use
and process accounting information, behavioural auditing research can investiBate
how auditors p€rform their audit tasks and make theL judgemmts. fuchival studies
of auditor choic€, examined in earlier chapterg often treat the process of auditing as
a 'black box. Results that Big or specialist auditors are associated with higher audit
fees and a lower cost of capital arc interpreted as €vidence lhat these audito$ arc
higher quality, but it is not direct evidence of belter auditing performance. Behavioural
research att€mpts 10 go inside this black box to €xamin€ the characteistics of better
performing auditors and investigate hcto$ that affect auditoff' pedormance.
Early r€search staned with fairly obvious quesiions. For example, does 8r€ater audit
€r.peri€nce improve the quality of auditors' iudgements?3s Despite the apParent simplicity
of this questior! researchers soon found that the answ€Is were not conclusive.36 Auditors'
performance raried between settinSs in ways which suggested that auditors had both
general knowledS€, which was common ro all auditors, and sp€cialised knowledSe that is
gained through pmctice and feedba& in particular domains or contexts.3T
One particular cont€xt that has be€n researched intensiv€ly is indusuy sPecific
experie$ce.83 Auditors with experience in industry specific contercts appear to have higher
quality auditor judgemeot when working within those contexts. This eff€ct appears
to occur because auditors acquire specialised knowledge when given the opportunity
to receive indusuy sp€cifc traiDina eithet formally or informally through on-dre-iob
experi€nce within the industry audit team environment.ae Owhoso, Messier and Lynch
show that when auditors work rtithin their industry specialisation teams th€y are more
e0 If
effective than oiher auditors at deiecting both conceptual and mechanical erro6
auditors are rcquircd to work oulside their area of specialisatioa they do nol €xhibit
$eater levels of performance. This research inding is suppon€d by obsewations that
large audil6rms organise their practices along indusrry lines.
Owhoso, M€ssier and Lynch's results show that auditoru' specialised knowledge is
not transferable to other contexts. Hammersley suggests that this is becaus€ industry
specialist auditors are able to use multiple cues more effciently than other auditors
within the industy specialist context. t t Cues are piece of infomation, and Hammersley
was inter€*ed in the relative performance of sPe.ialist and non-specialist auditors
in using apparenrly irnocuols cues to male Patt€ms that represent the €xistence of
financial misstat€ments. Her results show differences in the wa)s that sp€cialist and
non-specialist auditors make sense of the cues when they are sprinkl€d throughout the
informauon available to the auditors. She shor'is that specialist auditors tend to us€
more efncient and effective plocedures to detect the prcsence ofthe misstatement, bul
only $'ithin their o$n field of specialisation.
The industry specialisation research focr.Fes on auditor competence, which is one
pan of audit quality." Behavioural researcherc have also investigated the issu€s

A.counlin8 and resear.h


surrounding the other component of audit quality; auditor independ€nce. (och and
schmidte3 er€mined the effects ofdisclosing conflicts of interest on auditors' reporting
decisions. Initia y, and consisr€nt with Cain, Loewenstein and Moore,ea th€y found
that auditors are more likely to misrepo( wh€n they disclose conflicts of interest
because such disclosure mitigates rheir moral concerns. They feel th€y have a licence
to misrepofl because i 'esto6 should realjse the existence ofthis possibility. Ilowever,
koch and Schmidt found that more o.?erienced auditors are tess likely to misrepon
in these circumstances. They also found that when audho$ do not disclose lheir fees,
rhey work lo build a reputation for rruthful repodng. Koch and Schmidt shou' thal
in,reasing the comple\ity of the €xperiment by incorporalin8 mor€ faclors, such as
experience aDd reputation, can produce richer results and a deeper understandin8 of
auditors' behaviour-
Another way ofinvesrigatinS auditor irdependence is to e{amine invesiors' reaciions
to informarion abour auditors- For example, Davis and Holliet' and Dopuch, King and
Sch$,anze6 investiga(ed investor percepiions of auditor independence lvhen audilors
were receivinS non'audi! seNice fee revenue from their audit dient. Bolh studies
found that disclosure ofnon-audit fees reduces the accuracy of investor perc€ptions of
auditor independence. l'hese results are important because iheir evid€nc€ suggests thai
even when the auditor is independent in fact, the appearance of independenc€ can [']e
impaired and shar€ prices ad\€rs€ly affected. Regulations such as the Sarbanes Oxley
Acr (2002) were introduced to prcvent auditor independence problems by reslricting
auditors' provision of non-audit seric€s to their clients. Even if there is no aclual
ind€pend€nc€ impairm€ni to prevenl, these regulations could help audirors and rheir
clienB ayoid problems caused by pcrceiv€d independence impainnents.
Exp€rimental research has the polenrial to conrplement research using archival dai;r
by focusinS on the behaviour of auditors and investors. Holvever, the medodological
(hall€nges faced by behalioural researdrers are also significant. Ther€ is oftell a tension
b€tw?en making the experiment suificiently realisti( by incorporating nrany contellu;l
fadols and makine it simpl€ enough so that the researcher can be sure that the obseffed
outcom€ is carlsed by the aniplrlation of th€ spe.ific factor onder invesligation. Ih€
r€searcher has 1() consider all 1he design issu€s before assembling the subjects because,
unlike archival data research, there is no opponunity ro collect addhional data al a
later date. Research continues to progress increr)entally as researchers try differem case
instrumenls, contexts, and subjects.

CHAPTER l3 Bflri! !!ri €scirch I arcr,!nrii8 467


It l* n" U"huuiouralperspective and the nature oI behaviourat accounting research
Behavioural research in accounting can be interprct€d widely to include financial
accountin& audiling and managerial accounting. The focrus of ihis revi€w has been
human tudgem€nt theory (HIT) and (he use of that theory in explainin& predicting
and improving decision making. Research in this area has relied heavit, but not
exclusivel, oir three models of human information processing: the Brunswik lens
model, the pro.ess tracirg model, and rhe probabilistic judtement model. A review
of the research evidence demonstrates that prediction of company failure or disuess
has been heavily research€d using the lens model. The consistent finding has been
that failure is easier to predi€t than success and that the issue of disclosure of rhe prior
probability of failure is important to realism in the task. To a lesser extent, the
prcsentation and format of accounting information has also been researched using the
Ie model ftamework. It is, as yet, no! possible to formulate an underlying theory as
to which formats or modes of presentation are best ilr panicular ctcumstances. This
question awaits fur$er research,

The contribution to our understanding of the role o{accounting information within


and outside the accounting entity provided by studying behaviolr
Process uacing methods have revealed that th€re may be contexts in which decision
tree representations of human information proc€ssing may be more appropriate than
the simple linear combinations of data implied by Brunswik lens models. Proress
ta.ing meihods arc pa4icularty helpful when there are significant interactions between

ItllF- th" influ.nce oI accounting information on behaviour and on decision processes


within the probabilistic iudgeme model, auditor judgement has been o.tensively
research€d. Firdings reveal that auditors and others use the same rules of thumb as
have beelt preval€nt in researd in psychology, although oflen to a lesser €nem- The
rules of *rumb of representativeness, availabiliry, and anchoring and adiustment seem
to be adopted in auditing and other business contexts in ord€I to simplify complex
iudgement tasks and alleviate limhatrons in the cognilive processes ofhumans. whether
rhe pres€nc€ ofthese rules ofthumb leduces the quality of decision makrng in auditing or
accounting r€mains in dispute. Auditin& in panicular, is an adivity where compensating
m€chanisms, such as panner review, €xist as a means ofassuring quality. More tesearch
is needed in a hotistic sense in ord€r to assess whether deleterious effects occu( ftom the
use of rules ofthumb.

46A PART 3 Accountjng and research


The fact that organisations are complex environments and accounting disclosures are
trade-offs between competing perspectives and interests
Accounting e{ists as a dircct function of the activities of individuals or groups of
individuals (de6ned as accounting entiries). Consideration needs to be given to the
complex environm€nt in $'hich ac(ountants operate and the competing d€mands fot
and ofaccounti g information
For example, disclosinS environm€ntal information and the likely cost of an
environmenlal clean up by an organisation may l€ad to suppon ftom environm€ntal
iobby grotrps; however, it may also altracl a very different r€sponse ftom shareholder
Sroups, who may take the vie$' that lh€ company is expending ex(essive resources to
appease marginal inter€st groups.

@. The inherent constraints on behavioural research


Behavioural researchers and accountirrS practitioners need to recognise the complex
nalure oflh€ users and uses ofaccountirr8 information, and th€ influence oflhes€ users
on accounting practice and disclosure,

@-. lssues for auditors


Behavioural auditing res€arch investigates characreristics of high-performing auditors
and dre factors tl]at affect auditor iudgenrent. Results have shorvn rhat industry
spedalisr audirors outperform other auditors when they are in (heir sp€cialisr indusrry
environment. lhey appear to process multiple pieces of information nrore efficiently
and effectively to determine the existeDce of financial misstatements. EyperiD€ntal
research also sholvs that rhere are complex intemctions between experi€nce and context
in auditors' reponing decis;ons Additionally, tbis research shows that investors react
as though th€y perc€ive auditor independence is impaired when auditors re(eive non
audii sewice revcnue from ih€ir audit clients even if actual auditor independence is
not affected. However, experimemal researchers face challenges when they aitempt to
bahncc reali\m and simpli.iry in rescarcl design.

Questions
l. Consid€ra decision task, other dran a bankruptc.v prediction lask, that uses
accounting informarion (e g. nraking recommendations for shar€ investors).
Assume thar -vou are intending to conduct a Brur$rvik lens model experiment
on your select€d decision rask. Lisl seven information cues you think would be
imporlant vn riables to use in naking your decision. \Mry did-vou choose th€se
cues? Compare and contrast)'our list of information cu€s with a coll€ague. Dis.uss
wiLh each orher the similadties and differences in your choices.
2. Nlost humaD judgement research is undenaken in an €Ipenmental setting. Holv
(ould yoLr respond to the assertion drat exp€rinrenis cannot be generalised to
ihe real rvorld? \dhat are the rLeaknesses and strengths ofthis research methodl
{Hinti See R, Swieringa and KE weick, 'An assessment of laboratory exp€riments
in accounring, .iorrrndl of A.Iountins Reseafth, lal. 20, Supplement, 1982,
pp. s6-93.)
3. Use the probabilistic iudgement fraDrework io describe an accounling or auditing
related decision task.
4. Lxplain the implicarions for a.counting ifd€cision makers in an accounring
context display any or all of the represeDtativeness, availability, or anchoring and
adiustm€nt rules of thumb.

CHAPTaR ll B.havlolral resenrch if a..ou.llt! 469


aq

5. D€scdbe how (and why) ihe informatron ptocessing s''stems of exP€ri accouqtants
might be different ftom those ofaccounting students. How might the exp€nise
of experienced accounting practitione$ be effectively Passed on to accounting
students?
6. 'Most people are not good intuitive statisticiam.' Discuss this statement in an
accounong context, dra\ting on research using the Probabilistic model
7. Crmpare and contrast the etncient markets h)?othesis and human judgement
theory. fue they inconsistent with each other? Explain
8. Does consensus always imply accuracy in studies of accounting decrsion makingl
Justi& your answer.
9. why is a 'model ofhuman belavioul generally superio. to human iudgem€nts?
10. what ahernativ€s exist for improving the format and presentation of accounting
information? what is the research evidence regarding the merit of the various

11. Human tudgement rheory does nol puryon to peneuate the 'black box of
cogniuve processing. verbal protocol research, however, has that ability. What are
th€ $reng$s and weak'nesses ofthis r€search methodology? (Hittt S€e CF Klersey
and Tj Mock rr'€rbal protocol research in auditin g: , AccoufitinS, Oryanizatiofls a
Sociery. vol. 14, no. l l2, 1989, pp. 133-5r-)
12. Re.oncile normative accounting studies and human judg€ment th€ory.
13. why do individuals form lobby grouPs to influence companies' behaviour and
informadon disclosures? what g?es of information would shareholder Sroups
require to determine whether company directon should be censured for poor
perfotmance?
14. What is mor€ imponant for an organisation the 'right' Physical assets or the
'righf peopl€? Explain-
-
15. Accouming is a function ofhuman behaviour and activity. As such, is not atl
accounting research behaviouml? lustiry' your answer'
16. List nine or more factors that will influence the accounting system adopted by a
6Im and lhe information disdosed- Which of these factors is a direct function of
human behaviour!
17. What is ao industry specialist auditor?Will an indusary specialist auditor always
perform beter lhan a non-specralist auditorl Explain.
18. wbat is the difference betlveen independence in fac! and independenc€ in
appearance? Which is more imPo(ant?

Additional readings
Burgslahler, D & Sundem, CL 1989, "fhe €volution ofbehavioral accounting research in
the unit€d states, 1968- 1987', Behauiorul Research in Acco nlins, 'tol l, pp 75 108'
Caplan, EH 1989, 'Behavioral accounting a peGonal view'Behavioral Research in
-
AccountinS, vol.7, PP rO9-43-
Gibbint M 1977,'Human inf€rence, heuristi€s and auditor's iudgmenl process" CICA
Audit Research SYmPosium, CICA
Hogarti, RM &Einhom, HI 1992, 'Order effects in beliefupdating: the belief-adiustment
nr.odel,' Cognitiue Psvcholog" vol 24' pp l-55'
Hoprvood, Aa 1989, 'Behavioral accounting reuospecr and ProsPect" Bel auioral Research
in Ac.o ntinS,'rol. 1, 1'989' pp 1'22
' A66orntitl&
Johnson. Pg ,amal, K & Berrfnan, RG 1989, 'Audit iudgement tesearch
OrS^nizarions and Societv. vol 14, no r/2, pp 83-99-

470 PART 3 AccountinS and re5ear.h


Murrat D & Regel, Rw r 9 92, Accuracy and consensus in accounting studies of decision
IrnaU,t'trg, Behaviorul Reaearch in Atmunti11*, vol 4, Pp 127-30
Swieringa, Rl &Weick, KE 1982,'An assessment oflaboratorye\periments in accounting',
Iaurnal of Accounting Research, vol.20, SuPplement, pp 56 93

one area of research;n BAR is to discover the differences in the iudgements ofvarious
/€xperts'and to discover what characteristics and skills lead to expertise in aparticular

Telstra opts for David Thodey as rePlacement for Sol Truiillo


bl lohn Dutie and Jenniiet He||ett
An embaftled Telstra board willtry to improve relations with th€ Rudd Covernment and
the market by today announcing iis business head, David Thodey, as the replacement
for chief executive Sol Trujillo.
Mr Thodey was one of three interna candidates for the position, with one British
executiv€ al;o making the shortList . other internal candidates .. were finance chief
John Stanhope and Se;sis boss Bruce Akhurst. Neiiher is expected to
quit butjust what
management changes Mr Thodey has in mind are not known.
Mr Thodey, 54, is exp€cled to nrove quickly to nrend fences with ihe Covernment
and end the acrimonldrs relarionship that had developed between Canberra and
Telstra.
The Telstra chairman, Donald Mccauchie, has sotrnded much more conciliatory in
recent months, and the Government has publicLy and privately welcomed what it see,
as a far more conslructive tone.
But Canberra will betrying to drive a hard balgain in its negotiations over the building
of jts proposed $43 billion natiotral broadband network, and thjs will severely test the
n€w chiet execLtive and the board.
Mr Thodey, an Alrstralian, does not have a close relatjonship with key government
ministers, but is regarded as re atively calm. And he has the poliiicai advantage of
not being closely identlfied wiih Mr Trujillo's aggressive stlategy againn government
requlaton over the pastfew )e. s.
-e
,1. .o" , nie e\e.Jtrve \il nlco Lo o^r.uddc in.F'rors .P sI rol n"lc roo
"
n a-\ 'or. F.,iol. ro rnberia r'rJ- $ould a'rd I p 'd n' -B l-c p ' d 'hdleho d'
lrp'o. o bLrildirs l. nFr$o " d116.l''r '|p.to'o \Lm'r\ * lb'o',:m' 'dl"
"fip cdb) rl cr"rir', le.l'a $"n". "'d i-.."
)d lerSrh' n"gorr'o1s' 'n bP

Jourcer ExceeLs lrom IneAutt/a/nn, I Au8!st 2009, sw\{ theau(ralin1.news com au

Questions
'L Why would the Te stra board appolnt an iniernal candldat€ to the position of chief
€xecutive, ratherlhan an externa candidate?
2. \rvhy miSht Davl.lThodey have been preferred to the othcr int€rnal candidates one
jndustry
oi whori had flnance experience and the secord of whom had other telco

3. The aticle repods that David Thodey had internal exPerience. How might his skill
set have changed during his iime wilh Telstra and what miSht have been his main
means of learning new skills?
4, Descrlbe the complex environment in which TeLstra is operating and identify the
paftic!lar skills th;t the new chief execut ve might be able to use to inflLrence that

cHAPTtF lr BchJ ou'a rc.PJtu[ | rr.!unLinS


Will a focus on cash rather lhan debt chango decision process€s? The followint exlract
consideF tlle shifting focus lowads the importance of cash flow in congi&.ing

Reporting season's moment of truth


by Barban Dtury
This month's rcportinS season h the momenr of truth for corporate Australia when it
reveals the extentofany damage sustained during the global tinancial storm.
'This is probably the tou8hest reporting season in decade6," says Fat Prophets founder
Angus Ceddes.
"lt's time to pick the wheat from the chafi . . . the ones lo focus on are coping well and
will be in a strong porition in future-"
John Price, ofConsciour lnvestor, encourages people to ask it all the indicatols point
to a business being successful in five years' time ... key indicators to identify the top
share investments in these turbulent times include. -.

Lowdebt
Too much debt was a feature behind last yea/s corporate casuakies such as ABC
Leaming Centres, Allco Finance and Babcock & 8rown.
"Companies with little or no debtwillcome out{ofthe cufient crisis) stronger," Price
says. He look for a debt-to-€quity ratio of under 50 per cent.
Ceddes says cash-flow problems are an important sign that a company's financial
posilion is deteriorating. H€ cites Qantat which sprung a nasty surpris€ on investors
earlierthie month, announcinga $500 million capitalrais;ng to pay down debt. "Qantas
is one of the best-run airlines in the world bur given the stare of the global€conomy the
airline market will be among the last to recover," he says-
solrce: Exceryls frcm t e A6e 25 february 2009, h(pr'lthebigchan.com.a!.

Qu€stion$
l. lf cash i5 the be5t meesure of success, why have the accounting profession and
reSulalors persevercd with analysing debt-to-equity ratios?
2. What do analysts means by the term'nasty surprise'?
3. Why does the market so rcadily interpret a cashjlow p.oblem as a sign that a
company's tinancial poeition mighl be deteriorating but was appar€ntly prepaled to
focus on debt-to-equity ratios for a number of years?
4. The anicle refers to the need to'sort the wheat lrom the chafi'by focusing on
companies lhat are copingwell. How are such judgernents made?

Endnotes
r. T Hofst€dland i Kinard, A state8y 4. S€e pp. 3a6-7 ofSE Bonrer, adudes and pedoman.d. Iormal
for behavioral ac.ounrin8 rekarch', 'tud8m€nt and d€ck'on-nal.jng oJ AccorntinS Rardfth, vol s, t967 ,
A.c@nnn8 Revieu', lantary 1970, Iesea(h in arcounting', Accountir8 pp. 22s-8.
p- 43. Honzonr, vol. 13, tro. 4, December 7. W Edrvads, The tbory ofdecision
2. c Sieael and H iamanauskas 1999, pp.38s-98 makind, Pr/.loldSicdl Bulenn, tuly
M^rcni, Behaviotul Acctu in$, Ohi,o: 5. See paiagnph 43 ofSAC 2 Objecdve 1954, pp. 380-417
South'W6reh Publishjry Co., I 989, of CeneBl Purpose Financial 8. RH Ashon, 'An experimenial srudy
p. 4 dBcribes hos'wjdely BAx.an Reponin& Melbouhe: Publi. Sector of int€mal odrol judsmenls,
Accounring Sted.rds Board and lor 4l of A. r tins Rcserrch, spnnA
3. EM Bambd, 'Opponmides itr AccountinS Standards Rdiry Boa.d, 1t74, pp. 143-57.
behdioBl accounting researcb , Au8usr, 1990, 9. E Br.ttswik, me Con.ett@l Ftu'ne@ft
B.h4tbtul R.eotdt i^ Ac.ouht' 8, 6- Sw Becker, 'Discussiod ofthe of Prtr,DloSl, ChicaSo: UniveBity of
eol- 5, 19t3. pp. 1-29. effed of ftequenry of feedback on Chicago, 1952

472 PART 3 Accouoline and .ese:rch


ro. AI tnrd, 'D*elopment ofbeharioml of the impa.f, .id/ftdl df A..o/ntin8 36. See, for qample i Solomon, A Aiyo
n\o!8ln', Behaiotul R^earch ifl R6ru r.ir Spring 1983, pp.300-6; and LA Toma$ini, 'Comextual €trects
, c@u,ir8, vol- 1, 1989, pp 124 49. and I(A Horghton and R Sengulta, on the calibration of probabilistic
11. R Libby and B Lewis, 'Human 'Ibe etrect of pnor probability indgnenrs', launat ol Appiied
infomalion p.ocssing iesea(h in disclosur€ and infomaiion set Psy.ioloSl, Augusl1985, pp.528 32
accounting: The state of Lhe an, consnuction oD baDkets' ablliry 10 for a .evielv of this literarure.
Accatnti S, OtsdniatiaB dfld so.iety, predict failur€',.ro!ftal o/,{dornri,g 17. Hl Einhom and RM HoSarrh,
vol. 2, no.3, 1t77, pp. 245-68. Seiedrdl, Auiumn 1984, pp. 76S-7s. 'con6dence in iudg€mem:
12. Ni Ashton, H hdn lhfonation 25. S.e, for emple, (A Houghbn and Pe6isten.e of ihe illusion ofvalidiry',
PraasinS tu dc.auflti 8, studies ifl D woodliff, 'Finan.ial .a1ios: The P\chalagical Redtu, \ol. 85, i\o. 5,
accountins Bearh no. 17, Arnetian pEdiction of corPomle succes 1978, pp.395 416.
Accountir8 Associadon, 1982. and fai lu!e', .lo!flal 0/6 rinar 38. R Libby,'Man v*sus model ol
13. See CF KleBey and TJ Mo.k.r'erbal Finaa.e dnd AccauntiflS, \lilIet 19a7, man: Sone conflicti.g eridence',
resear.h in alditinS',
lroto.ol !p 537-54. Oryannaioflal Behariow unrl Hun1n
Arco ntin& OryMizationt and Sa.ietf, 26. c caser, ir andTsellnrg, The effecl Peforndh.., lnne 1976, pp. t-12
lol
i4, nos. 1/2, 1989, pp- 133-51. ol rask prediciabilitl aod prior 39 I zimmer A lens siudy olrhe
14. Bonner op. €it., p.394. probability discLosur€ on judg€m€nt prcdi.tion of corFrate failu.e
15. See BA Howieson, ? seority quality and confiden.e', Aaoratinj b. bank loans oflice$', Iolmdi o/
ana\Gf s adion r<ommenda(ions: Rcuier4 April 1t86, pp. 302 17. Acauntins Resear.h,lol. tA, no 2,
An alplicarion ot reoEive 27. M van Br€da and K Feris, A nole 1980.
paniiionibg to modellinA judEemenr, on ihe effec! of prio. probabiUry 40. Fo! dample, seeT Kida,'An
Astali.n loundl af MdndxenmL disclosure and infomation invesrigation into audnos' contnujty
vol- no. 2, Dec.mber 199 r,
16, r€presentativene$ on subject and related qualiicarion iudgemenrs',
PP. 165-35. predicrive acoracy , ae,rd,tordl larrfldl of Acco flinS R5earch,
16. R Libby, Hbndn iflJadbtion Re\earh h ArcountinE, rol 4, 1992, Aulumn 1980, pp.506 23.
proce$ing 'theory aad afplicatioB, Pp. 140 51. 41. Libby op. cir., p. 27.
EnSlewood Cliffs, Nl: Prentice Hall, 28. AR Abd€l-(halik and (M Dl-sh.shai, 42 See Maines op. cit., pp. 79-80 lor a
l tBr, p s5. 'lnfornaton choice aDd utilizadon rdiew ofdrese nndings
17 R Ashton. 'ltuman infomation io an dpe.imenl on default 43. Se€, for emplq P slovic, 'Analyzing
processing r.search in auditin8r -{ ptedi.lion , Iaumal of A@tntinS fie egen judger A descriptive
r$'ierv and synihesis', in DRNichols R.idr.l, Aulumn 1980, pp. 325 42. srudy of a stockbroker's jnr4tment
and HFSreder {edt, Aldrt1r 29. R Sinnett and (Trotmad, Audilor de.ision prc.es6',/ouftal olApplizd
Slnpdrt tr VL Kansas, 1942, p. 80. v€6us model: infomalion .hoice A),.nolql, vol 53, pp,255 63; and
18. JG BinberS and JFshieldr Tlxee and infomadon prcce$ing, ,{ Schepanski, Tes$ of theories of
decades of behavioral a..ornting ,A.rolrtfl8 Ra,iax,, July r989, info!mation pro.e$ing behavior in
reseddr:A sear.h lor order'. pP. sr4 23. oedit judsmenf,,{..omrins x",i.tr',
Behalb l R5eneh in A aunting, 30. S.e, tor.smple. D Srowball,'son. vol. 58, 1983, pp. 58r-99.
vol. 1, 1989, p.25. ef..$ ola.colndrg dperrise a d 44. D Lar.k€r and \P Lessig 'An
19. See, for dample, Libby, op. cir., nrlonnalion Ioad: An empirical dadination ofdr linear and
strdi, A.caanlinS aryahiatians dntl retrosle.live Pro.cs tracing
20 ibid, p. 54. Sd.i4),, ir80, pp.32:r 8. approaches to iudSnem modellnr8,
2r.lbid.,p. 81. 31. HNi S.hroder, NrJ Ddver and ,4r.or in8 R?rt.r,, vol. 58, 1983,
22. A Tvebla and D (anneman. s sldenten, H!ftdn inJantuIion Pp.5a-77.
'iudgment under c€ftainty: Htu.isliG pm.€$i"8, N€w York Holr, Pjn€haft 45 TsellinE and l shank,'Linear
and biases', Srt n.d, Seplember 1974, and U'i.ston, 1967 veBus proces tn.in8 apProa.hes
Pp. il24 31. 32. Abd€l Khalik and El Sheshai op cir ro j!'dgmeDl modelli'rg:A Dew
23. lor emple. C Casey,lr The 33. Libby op cn, p.37. p.sPective on are nnPotuncei,
$etuhes of accoudl8 ratios lor 34. For ebmple, CJ CasE, Jr, ladation Accouflti g, Ortariatiofls dfld Sociery,
a subje.(s predidions of corporale nr lcco(nting nrformalion load: The rol 14, 1989, pt.65 77.
failur€: Replication and e\lensions', .ff.d on loan ofices' p€d i.tions 46. R Libby, 'Prediction adrlevement s
lo mal af A..onting Raearh, ofbanktuptcy', A.tor ttS Relrs, an slension ofthe p.edictive ability
Au mn 1980, pp.603-11; and January 1980, pp. 36 49. critenonr A reply', n..!!nlina Reria4
I Zimmer, n lens srudy ofthe 35 S€e L4 lvlaines, 'Jud8.menl and luly 1976, pp. 672 6
piediclo! of coryorare lailure decision makinE rcsearch in fnancial 47 SMoriantt, 'Conmunicatl.g
by bank loan otrc€N', /o!rul ol accounring: ,^ r€view and analysis', information throrglr
fi nancial
nmddng Reieu..i, Autumn r980, in RH r$hton and AH Ashron (edt, multdnn€nsional grapbics', /ornrdl
pP.629-36. ludSekent and decision-nahitg Esearh aJ Accounnry Resedr.h, Sptinq 1979,
24 For qample, C Case_v.lr'ldor in e.ornrinS anl auritinS, Ne$ York: pp.205-25.
probabilitt disclosure and loan Cambridge Uoiv€rsity Pre$, 1995, ,18. D Srock and Cl watson,
of6.eu' judsedcnt!: sohe eridence P. 92. 'thman juds€m€nt accuaq',

CHAPTER 13 Behrv o!ral .esearch in accounl rg 473.


muhidlneNional gnphics and 56.Iw Payn€, I B€uman and E, ,ohnron, led.), vi]''al inlomation ptuceslinS,
hunang veNus nodels', .loutui ot 'Behavioral decision rerear.h: A New York Asdemic Press, 1973,
Afuunting Raedfth, Spti^E t9a4, .oraf uctiv€ Prccelsing Pespedive" pp.215-81.
pP. 192-205. Anhual R@i@ of Pstcholog,, aol. a3. 7a. M, EosnaD, Th€ u* ofa.counling
{9. ibid., p.201. 1992, p. r03_ infomation: Elpen !6us novi(e
50. For €*ample I B€nbasat and 59- ibid. b€havioul in C Ung6on and D
R Schroeder, 'An experim€ntal 60. for ir$ance, ttle Peat Marwick BraunsreiD {eds), De.6ioi An
invetiSation ofsome MIS design 'R.s€arch Oppodunni6 in Auditin8 'rd}inJi
inll lieitiitury inquiry, Bdton:Kmt"
vdiables', MIS Q@rfnly, March 1982, pp.234-67.
t977, pp,37-50; H Wainet 61. D r.ahnemar and A TveBlry, 'On 75. for oomple, SF Biggs and Tl Modr
M lono add c crov6, On de dirpla), the psycholo$/ of prediction . 'An inv6ti8ation of audilor decision
of d^u: sone efttincal fndinss, Pty.hologicol Retieu, \ol. aO, t
73, prca$es in €liluarion of inlernal
washinSlon Dc: Ihe 8ure.u of PP.237-5r. contrcb ad audit s€ope decisions,
S(ial Science RMrch, 1962; and 62. RM Hosanh (ed.), lrutrhB ir drc'rion Iou al of Atuu tint Res.irch, Sqine
E, Lusk and M KeBnick, The eFe.r nohing: A iibute to Hillel J- Einhom, 1983, pp.234-55' SF Big8s,1! Mo&
of cegf,itiE style aDd repon tomat chicaSo: UniveBity of Chi.ago Press, and PRwa*ins, 'Auditorb ure of
on lask p.rformaD<e The r.,tls d6itn 1990). analtri.al @iew in audn pro8mm
.o sequenc6'. M1na&hdt Scitu . 63. PaFe, E.nmar and,ohNo., op. ci1., desieJr', A@nti'r( R@ieu, lan\aty
Augusr l97t pp.787-98, P. 104, 1988, pp. 148-67; F choo and KT
51. E Blocher RP Moffie and Rw Zmud, 64- E roy.e and Cc Biddle, Are auditoa' Trotman, The relationship be$een
'Repon fornat and task compldity: iudS€m€nts sutrci€ndy rcsresrive?', knowledse strudure and iudgemen$
lmemction in risk iudtem6rs', lounal ol Ac@uating R5tutch, for €rperienced and inqpenened
AccoMtin& OryMizations ad So.ietf, Autumn 1981, pp. 3.23-rt9. a\tditors', Accuntin8 R@itu, fuly
t9a6 pD.457-69. 65. DL Holr 'AlldnoB.nd be nt€s 1991, pp. 464-65; and DM FRded.k
52. LRDavir 'Repon fomat and taisited', A@ounrin& Oryanitutioft ?udi(oK' repeentation and relliel
the d€cision makels task An ond Soci.ty, r9A7, pp,5?\-A. of inl€mal.ontol knowledE',
e'+trimenul invesdSadon', 66. D khneman and A ft€rsky, A.@untiag R@itu, Apr:n 19t1,
A@ tin& Oqaniatifu and Soci6r, 'choi<€s, r€ld6 and fiam€s', pp. 240-58-
El. 14, no.5/6, 1989, pp.495-508. An*icat Psycholqist, 19a4, 76. CL M@dd, 'Th€ €ff€ci of el(perience
53. C D€ran.tb and S Jarvmpaa, pp. 341-50. on ardiioE' memory eFoB', ,olmal
'Craphlcal presnGtion of accorntinS 67. 4 Tverslq/ and D Kabneman, ol Accounting Research, A&umr. 1990,
data for fioancial for.castirg 'Availabilir). A heurisric for judSing pp. 368-87.
An qperimental inwsri8adon', ftequ€n.y ad p.obability', Crftilip 77. lL Bu('Frequen<y judg.ment! in
A@untin& Oryatiun sa d Societf. Arrtoloy, S€pbmbe. 1973, an auditing-rclared msr, .rolnal o/
vol- 14, no.5/6. 1989, pp 509-25j pP.2Q7-32. Acrolrtn8 Xarrrrcll Auiumn 1988,
wF wriBhr, 'Cmphical displals 68. DV Mo*r, The €ffftts otoulpur PP. 315-30; and R Libby and
and improrrd derision 6aking', interfe.en@, availability aod DM l'rderick, 'Experien e and th€
in CD Garen and s NaFl {eds). a@untina idfomation on investoa' ability to dplaio audit findin8s'.
Canpu@r and tt locial ei.Eet, pr€dictive tudSemerls', Iotnal of Ac@untinS RMr.h,
Crenwici, CT: tAl Press, 1989. ^..o!nlir8
Reuar, Iuly 1989, pp.433-48. Artunn 1990, pp. 348-67.
54. DNPj.chiuk, An €mpid.al 69. EI loyce and GC liddle, 'Anchoring 78. See, for sampl€,, Shanteau,
as*ment of the impad of and adilstdent iD pobabilist'c 'cognitive heuritia ard bias6
allemative lark pB.nlation mod6 inferen.e id auditing', /o/ndl o/ in behavioral audirin& Rdi€w,
on dd3ion-making 6€arch in A@!ntin8 xdqr.ft, Spring r98t, commenis and obseMtions"
anditi,nd , Iautial ol Aeounting pp. 120-4s. Accoud^x, O4dniuiotu and
Ra?dr.ir, Spring 1984, pp.341-50, 70. wR Kinney, lr and WC Uecket So.ieq, vol. 14, no. r/2. 1989,
55- S So aDd M Sniih 'Colour BEpbi<s 'Miti8alinE rh€ conkquencd of PP. 165-77.
and lask complodty in muhi!2riale &choring io auditor iud8n.nf, 79. Se€, fo. exmple, R Hoganh, A
dd3ion makin8 , ,a..r!rrif,& Aulning Accounring Reuiew, l^N^ry t942, persp6tir'€ on cognitire .es€arch in
\n.l Accoun&biliq loumal,$| t5, Pp.5s-69. accounling" R&i4u, APril
no. 4, 2003. 565-93. 71. A Nsell and llA Simon. Huhan ^rroldfin8
1991,pp.27?-9o.
56. S 50 and M Smith, 'Pre*ntation ltoblo Solri!& EnSlsood Cliffs, Nl: 80. S Aurdell, C Clubb, A Hopwood,
fomar and infoma.ion <omplaity Prentic€'Ha[ 1972, , Hu8h6 and , Nahapie( Th€ roler
in mulrivariare decbion making, 72. see, for *ample, AS Ehl€in, LE of accounting in oBanizations and
UnpDblish€d working paper, shulman and sA Spralka, Medi.al society , Acountin& Ory, iutio$
Univeciry of Sou6 Aust6lia, probl.tn An a^ob6is oJ ctrtual dtrd S@e,,, vol. 5, no. 1, 1980,
2001. realonin&'a1,in8:
Camb.idge, Mas' Harvard pP. 5 2?
57. E waio€rrDd D Thie$en, CrapNcal Universiry P6, 1978). 8r . t Zimmffiar, Aceotnnnt lot d.cisbr
Data Ad lysit, Aanual Retid! ot 73. WC Ch ..nd HA Simon, 'fh€ ahinE dnd u^nol, Chi.aAo: lwio
Psfrholw, 19a1. pp. 19 l-247. mind's ey€ in ch$'. inwC Chase Publishin& 19t7, pp. 633-89.

474 PART 3 Arcoumin8 and research

Das könnte Ihnen auch gefallen