Class
Action Sex Discrimination Prevention Controls of AIS Faculty Promotion- A Forensic Audit, Expert Witness
Testimony & Computer Litigation Support Case
Avi
Rushinek, &
Sara
Rushinek
Correspond=
with
Sara Rushinek, arush@miami.edu=
***
*** Some of the tables have been omitt=
ed due
to space constraints. Please contact the authors in writing for more
information.
Class Action Sex Discrimination Prevention
Controls of AIS Faculty Promotion- A Forensic Audit,
Expert Witness Testimony & Computer Litigation Support Case
I<=
span
style=3D'mso-tab-count:1'> Abstr=
act
The authors =
develop
a methodology to control for “Class Action Sex Discrimination” =
law
suits using Accounting Inform=
ation
System (AIS) faculty promotion as a case in point. This is a forensic audit, expert w=
itness
testimony & computer litigation support case study approach to minimize=
the
monetary damage, flag emerging cases and providing quick settlement of a cl=
ass
that is pending. Such systems=
can
be used by accreditation organizations such as the AACSB, who accredits
accounting programs and examined how their publication rate. Academic organizations such as the
American Accounting Association (AAA) can use it to promote equal opportuni=
ties
and compliance with the Equal Employment Opportunity Commission (EEOC). The EEOC can use it, to enforce
compliance, and it can be used by universities, as well as law firms.
Using standa=
rd
statistical methods, the authors found that universities promote AIS faculty
primarily on the basis of publications as they were statistically
significant. In contrast, the
“gender” or sex was statistically insignificant, but indicate a
bias in favor of female faculty members in the sample. This means that there is no eviden=
ce of
sex discrimination in the population, but the sample suggests that universi=
ties
may be biased against males and in favor of female faculty members in promo=
tion
decisions. The methodology ca=
n be
used as red flags to investigate possible inequities in promotion decisions=
.
II &nb=
sp; I
INTRODUCTION
The proliferation of “= ;Class Action Sex Discrimination” cases reported by the press raises the iss= ue of Prevention Controls. This = study looks at the case of AIS faculty promotion from a forensic audit, expert wi= tness testimony & computer litigation support point of view. We will review some of the more prominent cases that the press discussed more recently.
Such cases include but are = not limited to the following: http://writ.news.= findlaw.com/sebok/20040809.html reports about “The Huge Class Action S= ex Discrimination Suit against Wal-Mart: Should It Proceed as a Class Action, or be Decertified?” By ANTHONY J. SEBOK, anthony.sebok@brooklaw.edu on Monday, Aug. 09, 2004. “Earlier this summe= r, a federal district court in San Francisco certified a sex discrimination class action suit against Wal-Mart. (When a court "certifies" a class action, it allows the case to proceed with a class of plaintiffs, according to class action rules - as opposed to with a number of individual plaintiffs.) The class is claiming that Wal-Mart discriminated against women in the way it recruited and promoted managers. = The issue of whether this is true has not yet been determined. Wal-Mart has cla= imed that its experts will show that there is no evidence of discrimination agai= nst women in stores, and the court noted that Wal-Mart will have a chance to pr= ove this at trial.”
The class web site, http://gende=
rclassactionagainstcostco.com/costco94.pl
reports another similar case “Are you a current or former female empl=
oyee
of Costco who sought advancement to management? A nationwide employment discrimina=
tion
class action lawsuit has been filed against Costco Wholesale Corporation. T=
he
suit charges that Costco operates a "glass ceiling" at the
store-management level which precludes women from obtaining promotion to
assistant manager and general manager positions. On Thursday, January 11, 2=
007,
Judge Marilyn H. Patel certified the class finding that three current and
former Costco employees may represent all women employed by Costco in the <=
st1:country-region
w:st=3D"on">
http://www.genderl= awsuitagainstsmithbarney.com/ reports about another similar case= of “Gender Discrimination Lawsuit Against Smith Barney Plaintiffs alleged that Smith Barney discriminated against female Financial Advisors in compensation and business opportunities. The parties entered into a four-ye= ar settlement agreement that is subject to Court approval.”
This study uses the number = of refereed publications as a measure of “research productivity.”<= span style=3D'mso-spacerun:yes'> The authors develop a model = to forecast the faculty rank (1 assistant, 2 associate and 3 full professor ) = as a function of the number of publications and the faculty gender (Sex=3D1 for = males, and 0 for females). Likewise,= this model offers a methodology of testing for Sex discrimination by the significance of the Sex variable on rank promotion decisions. If the Sex is statistically signif= icant, then discrimination may exist in the population. In contrast, if Sex is statistical= ly insignificant, then discrimination is less likely in the population. If it appears like discrimination exists, then the plaintiff may need to set up a class-action server and let people join the class. We are developing such a server with a survey to support such class-action lawsuit= s.
The purpose of this study i=
s to
develop a regression analysis connected to a continuous online survey that =
will
raise red flags for the existence of sex discrimination. This is a tool for compliance fore=
nsic
audit with regulations of government agencies, such as the EEOC in the
We are using data collected= by Simkin (2007) but applying it in a different way, and suggesting another data collection methodology for the future.&nbs= p; Unlike Simkin et. al who focus on the “profile of the publicat= ion rates of accounting faculty, and gender” for academic purposes, we are looking at it more from the forensic audit, expert witness testimony, and litigation support point of view.
Thus our goal is to create benchmarks that might be useful to decide whether gender or sex discriminat= ion exists on a continuous basis in an automated cost effective method of an internet WWW server online survey methodology. For this purpose we recommend usin= g Microsoft SharePoint software that= lets a lay person such as an administrator that is not an IT (Information Technology) professional to design and implement online interactive continu= ous survey, quickly and effectively.
III &n=
bsp; REVIEW
OF THE LITERATURE OF RESEARCH PRODUCTIVITY
This is base on earlier studies by Rushinek and Rushinek (1985) who
“… st=
udy of
micro, mini & mainframe computers using an interactive artificial
intelligence expert” dealing with using computer software as an ex=
pert
or to assist experts in expert witness testimony and computer litigation su=
pport
in such cases as a labor law employeement “class action law suit.-
… “<=
span
style=3D'mso-spacerun:yes'> As Rushinek and Rushinek (2000) po=
int
out in an “Internet F=
raud
Auditing: A Simulated Health Care Industry Case StudyR=
21;
in the Journal
of Forensic Accounting, 2000 http://w=
ww.rtedwards.com
The forens=
ic
auditor can provide effective expert witness testimony and computer
litigation support regard“ AIS Sex discrimination class action law
suits.
What we are using here is a discriminant model, a model that
discriminates between significant and insignificant sex discrimination. Much like earlier work by Rushinek=
and Rushinek
(1984) dealing with “Mini/Micro
Computer Evaluation of System Features: An Empirical Discriminant Model of
Software and ” published by “Managerial And
Decision Economics, 1984 describing “Compatibility, Cost-efficiency,
Installation and Delivery” of information, which is the thrust of the=
current
study. This is an
empirical study like Rushinek & Rushinek (1984) dealing with “Accounting And Auditing Software Evaluation With
Knowledge Based Expert Systems: An Empirical” discussed at the
“Fourth
Annual International Conference: Phoenix Conference, 1984.” We have reviewed numerous applicat=
ions
and selected the best for evaluation of regulatory compliance such as the
compliance with the US EEOC. =
Much like “Using Expe=
rts
for Detecting and Litigating Computer Crime,” Rushinek and Rushinek “Managerial
Auditing Journal,” 1993=
, the
current study is detecting the potential of a class action law suit. “Health Mar=
keting
and Sales Distribution Software Related to Computer User Satisfaction”
is similar to the current case as Rushinek and Rushin=
ek
report in “Health Marketing Quarterly”, 1985 fast system
response time is the main determinator of overall user satisfaction.” If filling out the survey takes to=
o much
time, users will fail to fill it and the data will not be accurate. This study deals with “… software related to compu=
ter
user satisfaction: an interactive online expert system using diagnos=
tic
…” anti-discrimination compliance software, as Rushinek & Rush=
inek
discuss in “International Journal of Mathematical Education in
Science …, 1986, it is “an interactive online expert system using diagnos=
tic
audit trails through telecommunication networks.” The audit trail is an important fa=
cet of
the current study, due to the fact that we may need it to support the
litigation.
&nbs= p; We focus on providing expert witness testimony and computer litigation support= for employment and labor law discrimination cases, in general and class actions= in particular. There are many ot= her reasons why measures of research productivity are important to accounting faculty. As Simkin et. al. (2= 007). suggest that it is important “in creating benchmarks for the quantity= and quality of publications needed to obtain tenure, be promoted, or receive an= nual merit awards (Hasselback, et. al, 2000; Hult, et. al., 1997, Pierce and Gar= ner, 1995; Hagerman and Hagerman, 1989) Then too, administrators have a similar interest in these norms (Maranto and Streuly, 1994; Campbell, et. al., 1983; Schultz, et. al., 1989).
&nbs= p; Faculty research productivity is also important to those universities and colleges seeking accreditation (Bublitz and Kie, 1984; Hasselback, 2000). One application in the accounting = area is for comparisons among similar institutions (Lucertini, 1995; Posy and Parker, 1989; Jacobs, et. al., 1986), comparisons between Ph.D. and non-Ph.= D. granting institutions (Schultz, et. al, 1989), or studies seeking to identi= fy the accounting profession’s most prolific researchers (Hech and Brems= yn, 1986; Hech, et. al, 1991; Gopolakrishan and Chandy, 1991).” Some research correlates publicati= ons to the area of specialization. (Campbell and Morgan ,1987; Englebrecht, et. al.,1994).
Gender studies of research productivity have been common, but none of them focus on sex discrimination and compliance testing. As Simkin et. al. suggest “Cole and Zuckerman (1984) n= ote, for example, that over 50 such studies of academics outside the accounting discipline have found that female faculty members publish significantly few= er articles than their male counterparts. In a later study, Zuckerman (1987) f= ound that women publish between 50 and 60 percent as many papers of what men publish.” However, for = the purpose of class action employment litigation is it significant?
&nbs=
p; Gender
differences apply to the accounting areas as Streuly and Maranto [1994] found th=
at
men published slightly more articles, but that these differences were not
statistically significant. Dw=
yer
[1994] confirmed the finding academics (per Hasselback’s Accounting
Faculty Directory) that women had published fewer times in academic
journals, but is it statistically significant and up-to-date?
&= nbsp; As schools increase the publications requirement for promotion (Read, Rama, and Raghunandan, 1998; Zivney, 1995; Milne and Vent, 1987, Englebrecht, et. al, 1994), testing for sex discrimination may be more important. This study foc= uses on “premier” journals articles (Van Fleet, et. al., 2000, Hasselback, et. al., 2000). However, it is possible that the other “non premier” jou= rnals may exhibit similar patterns of less publications for females. (Hasselback,= et. al., 2000; Heck, et. al, 1990 and 1991).
The quality of a journal is= a subjective matter and varies among researchers (see, for example, Zivney, e= t. al., 1995). AIS faculty member often publish in computer publications, which may not be considered as Accounting Journals. As Daigl= e and Arnold (2000) found that the “top 25 AIS researchers” listed non-accounting publication outlets as the top journals. Likewise, Christiansen, et. al., (= 2002) suggests that the tendency applies to accounting Ph.D.’s in that study’s researcher cohort.
Adjusting for co-authors (Hasselback, et. al., 2000) or accounting specialty (Campbell and Morgan, 1= 987), may further confound the results (Alexander and Mabry, 1994; Brown and Gard= ner, 1985). “Leveling the playing field” among the research faciliti= es of organizations may further confound the finding (Englebrecht, et. al, 200= 0, Read, et. al., 1998, Dwyer, 1994, Schultz, et. al., 1989
IV &n=
bsp; =
; MICROSOFT
SHAREPOINT ONLINE SURVEY- ADVANTAGES OF AN ONLINE COMPARED TO A
TRADITIONAL SURVEY
&nbs=
p; As
Simkin et. al. (2007) note “many of the studies cited above are now at
least ten years old or used statistical methods that some scholars suggest =
hide
more than they show.” T=
he
trouble with paper surveys is that they are so expensive, and time
consuming. Traditional survey=
also
involves many un-needed intermediaries that slow down the process. Such intermediaries as chairperson of the accounting
department at each school are “requesting that this individual in turn
distribute them to the faculty members in their departments,” which is
yet another intermediary in the process of data collection. Furthermore, such surveys are typi=
cally
based on printed directories, such as “faculty and school listings in
Prentice Hall’s Accounting Faculty Directory 2001-2002
(Hasselback’s).,” which may be outdated as soon as they are
published. In contrast, schoo=
ls
update their web sites much more frequently than any printed publication.
We have used the data that = Simkin et. al. (2007) collected as follows: “The areas of research specialization were (1) accounting information systems (AIS), (2) auditing, (Audit), (3) financial accounting (Financial), (4) managerial accounting (Management), (5) tax accounting (Tax), and (6) other (Other). The Other category captured all responses that did not list the first five areas abov= e, and included governmental and not-for-profit accounting, international accounting, accounting education, and ethics. Our academic rank categories = were (1) Adjunct Professor, (2) Assistant Professor, (3) Associate Professor, (4) Full Professor, and (5) Other.”
We have focused on AIS (Acc= ounting Information Systems) faculty member’s ranks, Assistant, Associate, and Full professors, their number of publications, and whether they were males = (1) or females (0). This way we had = a much more homogeneous data, which would result in a higher R-Square in a multivariate linear regression analysis.
V &nbs=
p; DATA
ANALYSIS AND RESULTS:
We ente=
red the
information from the Simkin et. al. (2007) into Microsoft SharePoint server
survey.
The Publication Rates Of Ac= counting Faculty: A Multidimensional Profile Publication Rates. Of Accounting Facult= y:. A Multidimensional Profile divided the&nbs= p; “survey instruments into a computerized data set. Auditing of = the entered data resulted in an approximate error rate of .2% for the 15% of the data subjected to random review. The two data-entry errors found were corrected.” Their data “provides mean publication rates for all faculty combined, separated = by area of research specialization and gender. These results reflect the repor= ted number of peer-reviewed publications divided by the number of respondents reporting that specialization. Thus, we grouped the data into 6 groups of o= nly AIS faculty by gender, and rank of assistant, associate, and full professor= s.
Table A
<= o:p> |
<= o:p> |
T=
able A |
A=
IS
Rank=3Df(Pubs,Sex) |
|
|
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
|
Regression Statistics |
<= o:p> |
<= o:p> |
<= o:p> |
|
|
M=
ultiple
R |
0.971985 |
<= o:p> |
<= o:p> |
<= o:p> |
|
|
0.944755 |
<= o:p> |
<= o:p> |
<= o:p> |
|
A=
djusted |
0.907925 |
<= o:p> |
<= o:p> |
<= o:p> |
|
S=
tandard
Error |
0.271403 |
<= o:p> |
<= o:p> |
<= o:p> |
|
O=
bservations |
6 |
<= o:p> |
<= o:p> |
<= o:p> |
The SUMMARY OUTPUT in Table= A describes the AIS rank as a function of Pubs (Publications) and S= ex (Male or Female), AIS Rank=3Df(Pubs,Sex).
&nbs= p; &= nbsp; &nbs= p; &= nbsp;
Regression Statistics inclu= de the Multiple R 0.971985, R Square 0.944755, and Adjusted R Square 0.907925, w= hich shows that the Publications and Sex explain about 90% of the variations in = rank promotion decisions for AIS faculty. Likewise, the table shows the Standard Error of 0.271403, and the Observations 6.<= span style=3D'mso-tab-count:2'> &= nbsp;
The Regression Statistics s=
how the
Multiple R with a value of 0.971985, and the
Following is a description = of the faculty rank values and their position descriptions:
Rank Legend = ;
Rank = Description of Rank
1 &nb= sp; Assistant Professors
2 &nb= sp; Associate Professors
3 &nb= sp; Full Professors
Table B
|
A=
NOVA |
<= o:p> |
T=
able B |
A=
IS
Rank=3Df(Pubs,Sex) |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
|
|
|
df |
SS |
MS |
F |
Significance
F |
<= o:p> |
<= o:p> |
<= o:p> |
|
R=
egression |
2 |
3.779021 |
1.88951 |
25.65186 |
0.012985 |
<= o:p> |
<= o:p> |
<= o:p> |
|
R=
esidual |
3 |
0.220979 |
0.07366 |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
|
T=
otal |
5 |
4 |
&=
nbsp; |
&=
nbsp; |
&=
nbsp; |
<= o:p> |
<= o:p> |
<= o:p> |
|
|
|
|
|
|
|
|
|
|
|
|
Coefficients |
Standard
Error |
t Stat |
P-value |
Lower
95% |
Upper
95% |
Lower
95.0% |
Upper
95.0% |
|
In=
tercept |
1.034536 |
0.206693 |
5.005184 |
0.015349* |
0.376747 |
1.6923249 |
0.376746971 |
1.69232486 |
|
Pu=
bs |
0.085692 |
0.011964 |
7.162661 |
0.005605* |
0.047618 |
0.123766 |
0.047618181 |
0.123765976 |
|
Se=
x |
-0.359907 |
0.227225 |
-1.58392 |
0.211378** |
-1.083039 |
0.3632256 |
-1.08303909 |
0.363225632 |
|
|
|
|
|
|
|
|
|
|
· * Significant at the .o5 level of statistical significance
· ** Not significant at the .o5 level of statistical significance – Less likely to have a Class Action Sex Discrimination Case!
The ANOVA (Analysis of Var= iance) in Table B support the summary in Table A. It shows the statistical signifi= cance of the model of AIS Rank=3Df(Pubs,Sex).&nb= sp; Table B includes the following parameters: df (Degrees of Freedom), = SS (Sum of Squares), MS (Means Squares), and the F value and the Significance F. For the Regression, these paramete= rs have the following values of: 2, 3.779020662, 1.889510331, 25.65185978, and 0.012984867, respectively= .
Most importantly, the last= 2 values reflect the test of hypothesis, rejecting the null hypothesis that entire m= odel have zero correlation with rank with an F-value of 25.65185978 and a statistical level of confidence of 0.012984867, which is better than at the= .05 level of statistical significance.
In contrast with the F and=
the Significanc=
e F
for the entire model, only some of the individual variables are
significant. The first 2 name=
ly the
intercept and the Pubs are significant, but the Sex is not significant at t=
he .05
level. The second part of this
ANOVA test shows the Coefficients, Standard Error, t Stat, and P-value,
for the Intercept 1.034536,
0.206693, 5.005184, 0.015349 and the Pubs 0.085692, 0.011964, 7.162661,
0.005605 respectively.
Most important are the pairs of t Stat and
P-value for each of the variables.
The P-values show the statistical significance being lower than the =
.o5
required to reject the null hypothesis, saying that in the population these
variables are greater than zero, and therefore, statistically significant in
making rank promotional decisions.
It suggests that the number of publications is a significant factor =
in
promoting assistant to associate, and associate to full professor of AIS.
=
=
However, for the last variable Sex, the
opposite is true, it is insignificant at the .05 level. The Sex, a dummy variable (1 for male a=
nd 0
for female) has the following information after the Pubs in the table, it s=
hows
the: Coefficients, Standard Error, t Stat, and P-value: Sex=3D-0.359907,
Another important piece of information is the Coefficients of Publications and Sex. These coefficients reveal that on = the average the Pubs with a value of 0.085692079, implies that it takes 1 publication per rank. Or, in = other words, as an Assistant professor would need another publication over and ab= ove the Intercept of 1, or about 2 publications to be ranked as an Associate professor and an Associate Professor may need about 1 additional publicatio= n to be ranked as a Full Professor.
The Sex coefficient with a= value of Sex=3D-0.35990673, is the most interesting from the EEOC compliance stand point. The negativity of Sex = shows that a male with a data value of 1 incurs a penalty compared to a female wi= th a value of 0, in terms of the contribution of their publications to the promo= tion. Or in other words, on the average = male AIS faculty members may need more publications to be promoted compared to t= heir female counterparts. It may r= eflect the fact that in the sample female faculty member published less then male faculty members of the same rank. However, it is not sufficiently statistically significant to general= ize it to the entire population, and therefore, we can conclude that there is no evident of sex discrimination in promotion decisions for AIS professors, in= the population.
Table C
|
F=
-Test
Two-Sample for Variances |
T=
able
C Pubs Differs Between Sexe=
s |
|||||
|
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
|
|
Pubs |
Sex |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
|
M=
ean |
13.36667 |
0.5 |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
|
V=
ariance |
108.2187 |
0.3 |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
|
O=
bservations |
6 |
6 |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
|
d=
f |
5 |
5 |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
|
F=
|
360.7289 |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
|
P=
(F<=3Df)
one-tail |
2.18E-06 |
*=
|
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
|
F
Critical one-tail |
5.050329 |
&=
nbsp; |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
|
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
<= o:p> |
· * Significant at the .o5 level of statistical significance