This article was downloaded by: [The Aga Khan University]On: 16 October 2014, At: 07:00Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

The Journal of Experimental EducationPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/vjxe20

Linking Academic Entitlement andStudent Incivility Using Latent MeansModelingJason P. Kopp a & Sara J. Finney aa James Madison UniversityPublished online: 02 May 2013.

To cite this article: Jason P. Kopp & Sara J. Finney (2013) Linking Academic Entitlement and StudentIncivility Using Latent Means Modeling, The Journal of Experimental Education, 81:3, 322-336, DOI:10.1080/00220973.2012.727887

To link to this article: http://dx.doi.org/10.1080/00220973.2012.727887

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

THE JOURNAL OF EXPERIMENTAL EDUCATION, 81(3), 322–336, 2013Copyright C© Taylor & Francis Group, LLCISSN: 0022-0973 print/1940-0683 onlineDOI: 10.1080/00220973.2012.727887

MOTIVATION AND SOCIAL PROCESSES

Linking Academic Entitlement and Student Incivility UsingLatent Means Modeling

Jason P. Kopp and Sara J. FinneyJames Madison University

Academic entitlement has been theoretically linked with uncivil student behavior; however, this re-lationship has not been tested. To address this gap in the literature, the authors used latent meansmodeling to estimate the relationship between the Academic Entitlement Questionnaire and un-civil student behavior. The authors gathered scores on the questionnaire from 2 samples of students:civil/compliant and uncivil/noncompliant. Measurement invariance was established for the AcademicEntitlement Questionnaire, providing additional validity evidence for the scale and allowing for theestimation of the latent mean difference in academic entitlement. As predicted, noncompliant studentswere significantly higher in academic entitlement than were compliant students. These results empir-ically link academic entitlement and student incivility, further supporting that academic entitlementis an important construct in academia and should receive increased attention from educators.

Keywords behavior problems, college students, factor analysis, higher education, structural equationmodeling

EDUCATORS ARE BECOMING INCREASINGLY CONCERNED that students have a sense ofacademic entitlement (AE), defined as the expectation that one should receive positive academicoutcomes, often independent of performance. Academically entitled students often expect highgrades without reciprocal performance (Achacoso, 2002), or expect the professor to rearrange theclass structure to meet student desires (Greenberger, Lessard, Chen, & Farruggia, 2008). Whenacademically entitled students feel that their demands are not met, they may become hostile(Dubovsky, 1986). This hostility can lead to a breakdown in student-faculty relations, hinderingeffective education (Hirschy & Braxton, 2004).

The increased attention on AE may be a product of shifting cultural norms. Students enteringcollege today were born during the self-esteem movement (Branden, 1969). High self-esteemwas believed to lead to success, thus it was the responsibility of parents to do whatever possible

Address correspondence to Jason P. Kopp, Center for Assessment and Research Studies, James Madison University,821 S. Main Street, MSC 6806, Harrisonburg, VA 22807, USA. E-mail: koppjp@dukes.jmu.edu

Dow

nloa

ded

by [

The

Aga

Kha

n U

nive

rsity

] at

07:

00 1

6 O

ctob

er 2

014

ACADEMIC ENTITLEMENT AND INCIVILITY 323

to increase their child’s self-esteem. To bolster self-esteem, parents and teachers began givingrewards for minor achievements. Children were awarded trophies for merely participating inathletic competitions and some schools stopped publishing traditional honor rolls, for fear studentsnot achieving this criterion may experience diminished self-esteem. Moses and Moses-Hrushovski(1990) hypothesized that meeting all of an individual’s needs for relatively little reciprocal effortwould result in an exaggerated sense of entitlement. Thus, it is possible that taking part in theself-esteem movement caused parents to unintentionally rear increasingly entitled children.

How Does the Entitled Student Behave in College? The Rise of Collegiate Incivility

It has been theorized that entitlement has led to an increase in disruptive and uncivil studentbehaviors (Achacoso, 2002; Chowning & Campbell, 2009). Uncivil student behaviors generallyencompass behaviors that violate the social norms present in academics, such as “sending wirelessmessages [during class], arriving late to class, leaving class early, and inappropriate use of laptopcomputers in class” (Chowning & Campbell, 2009, p. 982). Uncivil behaviors can also bepresent outside the classroom in the form of rude and demanding e-mails (Lippman, Bulanda,& Wagenaar, 2009). When Twenge and Campbell (2009) asked faculty and staff from variousuniversities to send stories of entitled students, they were met with a flood of responses describinguncivil behavior. One professor lamented the amount of time she had to spend dealing with gradedisputes, especially when the students would recruit their parents to help argue their case. Someprofessors reported that students were threatening, making statements such as “I’m not leavingyour office until you change my grade to an A!” (Twenge & Campbell, 2009, p. 231). A survey offaculty on experiences with student incivility found that uncivil student behaviors were common,and ranged from relatively minor (e.g., not paying attention in class) to major confrontations withprofessors (e.g., angry yelling, threats; Goodyear, Reynolds, & Gragg, 2010). Student incivilitiescan create a negative classroom environment by reducing classroom enthusiasm and commitmentfrom other students (Hirschy & Braxton, 2004).

It follows logically that AE may lead to student incivility. Entitled students feel they deservepositive outcomes without needing to reciprocate; university faculty and staff exist to serve them.If entitled students do not receive positive academic outcomes, they see this as a failure on thepart of the university faculty and administrators. Accordingly, Achacoso (2002) theorized thatacademically entitled individuals will assert themselves when they feel that they are receivingless than they deserve. Providing evidence for this theory, the most egregious instances of studentincivility typically occur after students receive unfavorable assessments (Goodyear et al., 2010).In addition, students who score higher in AE tended to rate vignettes describing inappropriatestudent behavior as more appropriate than did less academically entitled students (Chowning& Campbell, 2009). Although this research suggests a link between AE and student incivility,no research prior to the present study has empirically linked AE to instances of uncivil studentbehaviors.

Most examples of incivility focus on a student engaging in behaviors that are uncivil. However,a student can also be uncivil by failing to act. A type of student incivility of importance to bothfaculty and administrators is noncompliance with university policies. Faculty and administratorsregularly require students to perform certain tasks as a part of their enrollment in the university.These tasks can include attending class, submitting assignments in a timely manner, paying tuition,or registering for classes by a certain date. However, the entitled student believes that education

Dow

nloa

ded

by [

The

Aga

Kha

n U

nive

rsity

] at

07:

00 1

6 O

ctob

er 2

014

324 KOPP AND FINNEY

should be delivered without having to give anything in return. Thus, the entitled student mayview classroom or university policies as unjustly attempting to require something of the student.This attitude may cause the student to act in an uncivil manner by refusing to comply with thesepolicies. This noncompliance can lead to additional tension between students and educators,as educators struggle to elicit compliance from the wayward students. Given the importanceof noncompliance to educators, noncompliant behavior may be an appropriate starting point toempirically investigate the relationship between AE and incivility.

The Measurement of AE

Proper measurement of AE is needed in order to empirically investigate the link between AEand uncivil student behavior. Although interest in AE and student incivility has spawned severalmeasures of AE (e.g., Achacoso, 2002; Chowning & Campbell, 2009; Greenberger et al., 2008),scores derived from many of these measures are lacking in validity evidence (for a review ofexisting measures, see Kopp, Zinn, Finney, & Jurich, 2011).

Of the existing AE measures, scores derived from the eight-item Academic Entitlement Ques-tionnaire (AEQ; Kopp et al., 2011) possess strong validity evidence, as assessed using the threestages of Benson’s (1998) strong program of construct validity. This program involves ensuringsubstantive, structural, and external aspects of validity. Kopp and colleagues (2011) built theAEQ on a strong theoretical foundation, explored and confirmed the structure of the measureusing multiple samples, and empirically supported hypothesized relationships with psychologi-cal variables. By contrast, other AE measures (Achacoso, 2002; Chowning & Campbell, 2009;Greenberger et al., 2008) lack important components of the validation process.

The Present Study: Investigating the Relation Between AE and Uncivil Behavior

This study aimed to empirically test the link between AE and uncivil student behaviors. TheAEQ was chosen to investigate this relationship because of the strong validity evidence gatheredpreviously (Kopp et al., 2011). For the present study, student incivility was operationalized asnoncompliance with a university policy to attend a universitywide testing session. As mentionedpreviously, noncompliance can have a detrimental effect on the education process and is ofparticular concern to university administrators and faculty.

Establishing measurement invariance

To study the relationship between noncompliance and AE, measurement invariance of theAEQ was examined across two groups of students: those who complied with university mandatesto attend a low-stakes testing session (compliant students) and those who did not comply withuniversity mandates and were compelled by the university administration to attend a makeuptesting session in order to remove a hold on their university record (noncompliant students).Measurement invariance was investigated to assess if the AEQ functioned equivalently across thetwo student groups. If measurement invariance did not hold, this would indicate that the AEQwas measuring AE differentially for the two student groups. Thus, establishing measurementinvariance was essential in order to compare AEQ scores across compliant and noncompliantstudents.

Dow

nloa

ded

by [

The

Aga

Kha

n U

nive

rsity

] at

07:

00 1

6 O

ctob

er 2

014

ACADEMIC ENTITLEMENT AND INCIVILITY 325

The low-stakes context of the administration of the AEQ could create violations of measure-ment invariance. That is, the noncompliant students may manifest their noncompliance by notputting forth adequate effort when completing the AEQ at the mandatory makeup testing session.Brown and Finney (2011) were similarly concerned that noncompliant, uncivil students maybe unwilling to fulfill low-stakes testing requests, resulting in invalid, thoughtless responding.However, this concern was unfounded in their study; the researchers established measurementinvariance of a psychological reactance scale across compliant and noncompliant students. Sim-ilarly, Swerdzewski, Harmes, and Finney (2009) found that noncompliant students typically putforth sufficient effort when completing low-stakes developmental or attitudinal measures, suchas the AEQ, but not cognitively taxing measures, such as a science test. As such, we predictedthe noncompliant students would provide thoughtful and valid responses to the AEQ and, in turn,the AEQ would function similarly across the two types of students. However, it was imperativeto empirically test this assumption of measurement invariance before estimating the relationshipbetween the AEQ and uncivil behaviors.

Estimating the latent mean difference

If measurement invariance was supported across compliant and noncompliant students, wecould then estimate the latent mean difference in AE between the two student groups. The latentmean difference is corrected for measurement error, and thus represents the unbiased differencein AE between the two groups. As prior theoretical work supported a relationship between AEand student incivility, we hypothesized that AE would be higher for the noncompliant students(Achacoso, 2002; Chowning & Campbell, 2009; Dubovsky, 1986). Thus, given equivalent func-tioning of the AEQ across compliant and noncompliant students, the goal of this study was toempirically assess the relationship between AE and actual uncivil student behavior, addressingthis gap in the literature.

METHOD

Participants and Procedure

Data for the present study were collected at a midsized, southeastern public university. Asnoted earlier, compliance and noncompliance were operationalized by whether a college studentattended a mandatory universitywide testing session. For the university to assess educationaleffectiveness (Spellings, 2006), all students are required to complete a set of assessments twiceduring their academic careers: once before they begin classes as first-year students, and onceafter they have accumulated between 45 and 70 credit hours. Classes are cancelled for both ofthese “assessment days.” These two assessment days last approximately 3 hr and consist of abattery of affective and cognitive measures. For the most part, students are administered the samemeasures during the first and second assessment days, thus facilitating conclusions regardingstudent growth over time (i.e., value-added). There are no consequences for performance for theindividual student (i.e., low-stakes testing).

Despite mandatory attendance, every year there are a number of students who do not attend thetesting session. If a student is absent from the scheduled assessment day, the student must attend

Dow

nloa

ded

by [

The

Aga

Kha

n U

nive

rsity

] at

07:

00 1

6 O

ctob

er 2

014

326 KOPP AND FINNEY

a makeup session in order to remove the hold placed on his or her academic record and registerfor next semester classes. It is highly unlikely that the absent students were simply unaware of theassessment requirement; fliers are posted around campus, students are notified by e-mail, classesare canceled, and the date is clearly marked on academic calendars. It is also unlikely that themajority of absent students have legitimate excuses for their absences. Assessment specialistswho facilitate assessment day note that very few noncompliant students (less than 1%) givegenuine excuses for missing their assigned assessment session. Given that the majority of theabsent students were made aware of the assessment requirement, and that few offer legitimateexcuses for their absences, it can be inferred that their nonattendance is a blatant instance ofstudent incivility and noncompliance with university policy. Lending evidence to this conclusion,proctors of these makeup testing sessions often report the students exhibit higher levels of uncivilstudent behaviors (e.g., sending text messages, talking, ignoring instructions) than do students inthe standard assessment session. Thus, for the purpose of this study, failure to attend the initialassessment session was conceptualized as an instance of student incivility.

Compliant sample

The data used in this study to examine the relationship between AE and compliance werecollected from sophomores and juniors assigned to complete their second assessment day in thespring of 2008. A total of 3622 students complied with the university policy to attend the testingsession. Of that total, the AEQ was administered to a random subsample of 381 students. Onemultivariate outlier was identified using Mahalanobis distance. This individual seemed to respondrandomly, justifying removal. The final sample of 380 compliant students had an average ageof 20.1, and was comprised of 66.6% women, 81.8% Caucasian students, 3.4% Asian students,3.2% Hispanic students, 2.6% Black students, 1.1% Pacific Islander students, 0.5% AmericanIndian Students, and 7.4% of students who did not specify their ethnic background.

Noncompliant sample

AEQ responses were collected from all 366 noncompliant students participating in the uni-versity makeup assessment sessions. One multivariate outlier was identified using Mahalanobisdistance. This individual seemed to respond randomly, justifying removal. The final sample of 365noncompliant students had an average age of 20.6, and was comprised of 49.5% women, 79.5%Caucasian students, 5.7% Asian students, 3.6% Hispanic students, 4.9% Black students, and 6.3%of students who did not specify their ethnic background. The noncompliant group consisted ofmore men and was slightly older, which has been found in previous research (Swerdzewski et al.,2009).

Measure

Academic entitlement questionnaire (AEQ; Kopp et al., 2011)

The AEQ is an eight-item self-report measure of AE. Participants were asked to respondto the items using a Likert response scale of 1 (“Strongly Disagree”) to 7 (“Strongly Agree”).

Dow

nloa

ded

by [

The

Aga

Kha

n U

nive

rsity

] at

07:

00 1

6 O

ctob

er 2

014

ACADEMIC ENTITLEMENT AND INCIVILITY 327

Previous research supported a unidimensional structure, with internal consistency (coefficientomega) estimates of .81 and .84 for two student samples (Kopp et al., 2011).

Data Analysis

Measurement invariance

The establishment of measurement invariance was necessary in order to estimate the latentmean difference in AE between the two student samples. Testing measurement invariance con-sisted of three steps: establishing configural invariance, then metric invariance, and finally scalarinvariance (Steenkamp & Baumgartner, 1998). First, the fit of the empirically supported one-factormodel (Kopp et al., 2011) was assessed for each of the two samples (compliant and noncom-pliant). Although establishing model-data fit for each sample separately establishes configuralinvariance, we also estimated a combined-sample configural model to examine the combinedmisfit associated with fitting a one-factor model to both samples simultaneously. This combined-sample model served as a baseline model that was then compared to the fit of the metric-invariantmodel. This metric-invariant model was estimated by constraining the unstandardized pattern co-efficients to be equal across groups. If there was a significant and practical decline in fit betweenthe configural and metric models, this would signal a lack metric invariance, indicating the itemsdo not have equivalent saliency to the latent AE factor across samples. Lack of metric invariancewould prevent meaningful comparisons of the two groups with respect to AE. Given metricinvariance, the scalar invariant model was estimated by constraining the item intercepts to beequal across groups. If there was a significant and practical decline in fit between the metric- andscalar-invariant models, this would signal a lack of scalar invariance, which indicates differencesin observed AE scores across samples are not accurate reflections of latent AE differences.

Latent means modeling

The latent mean difference in AE between the two student groups can be estimated afterestablishing configural, metric, and scalar invariance (Thompson & Green, 2006). Establishingmeasurement invariance ensures that the estimated latent mean difference is reflective of an actualdifference in AE and not differential item functioning. Moreover, by estimating the latent meandifference in AE between the two student groups instead of the observed difference in compositeAE scores, we can account for measurement error, which attenuates the observed effect size. Inorder to estimate the latent mean difference, we constrained the AE factor mean of the compliantgroup to zero, and freely estimated this value for the noncompliant group. The unstandardizedlatent difference value is difficult to interpret; thus, we calculated a standardized latent effectsize that is analogous to Cohen’s d (i.e., estimates the standard deviation difference in latent AEbetween the groups; Hancock, 2001).

RESULTS

Data analyses were conducted in three stages. First, the data were screened for univariate andmultivariate nonnormality. In addition, descriptive statistics and correlations were examined

Dow

nloa

ded

by [

The

Aga

Kha

n U

nive

rsity

] at

07:

00 1

6 O

ctob

er 2

014

328 KOPP AND FINNEY

before conducting any structural equation modeling. Second, measurement invariance (configural,metric, and scalar) was tested. Third, given measurement invariance, the latent mean differencebetween the compliant and noncompliant samples was estimated.

Data Screening

Before conducting the structural equation modeling analyses, the data were screened for univariateand multivariate nonnormality. Depending on the severity, nonnormality can bias standard errorsand fit indices (Finney & DiStefano, 2013). Absolute values greater than 2 for skewness andgreater than 7 for kurtosis were considered indicative of univariate nonnormality (e.g., West,Finch, & Curran, 1995). The data appeared to be univariate normal (see Table 1). To assessmultivariate normality, we used the macro that DeCarlo (1997) provided to compute Mardia’snormalized kurtosis coefficient. There is no universal cutoff value for this coefficient (Finney& DiStefano, 2013), but it has been suggested that utilizing maximum likelihood estimationwith data having a standardized Mardia’s value greater than 3 could result in bias (Bentler &Wu, 2002; Ullman, 2006). Mardia’s coefficients for both samples suggested the data deviatedfrom multivariate normality (compliant sample = 15.11, noncompliant sample = 15.27). Todetermine the effect of nonnormality, models were estimated using both unadjusted maximumlikelihood estimation, as well as using the Satorra-Bentler adjustments to χ2 values, fit indices,and standard errors (Satorra & Bentler, 1994). The unadjusted maximum likelihood results didnot differ substantially from the Satorra-Bentler adjusted values and all substantive conclusionswere consistent across estimators. Thus, unadjusted maximum likelihood results are reported.

TABLE 1Correlation Matrices and Descriptive Statistics for AEQ Scores for Compliant and Noncompliant Samples

Noncompliant

Item 1 2 3 4 5 6 7 8 M SD Skew Kurtosis

1 — 0.39 0.47 0.15 0.39 0.31 0.34 0.41 4.04 1.49 0.095 −0.5802 0.44 — 0.40 0.19 0.35 0.32 0.30 0.33 3.42 1.62 0.423 −0.5333 0.42 0.43 — 0.33 0.61 0.49 0.61 0.54 2.74 1.44 0.778 0.2384 0.29 0.26 0.35 — 0.42 0.40 0.34 0.29 3.76 1.59 0.178 −0.8025 0.41 0.38 0.50 0.35 — 0.46 0.62 0.52 2.72 1.34 0.926 0.7406 0.33 0.28 0.40 0.36 0.37 — 0.51 0.39 3.25 1.59 0.537 −0.2667 0.33 0.35 0.47 0.34 0.52 0.38 — 0.42 2.52 1.31 1.050 1.2168 0.44 0.38 0.55 0.34 0.48 0.39 0.45 — 3.10 1.42 0.437 −0.232Compliant

M 3.83 3.24 2.42 3.49 2.39 2.70 2.18 2.77SD 1.40 1.57 1.27 1.63 1.19 1.42 1.05 1.31Skew −0.085 0.643 0.914 0.261 0.859 0.833 1.111 0.693Kurtosis −0.717 −0.244 0.689 −0.894 0.352 0.308 1.803 0.068

Note. For the compliant sample, n = 380; for the noncompliant sample, n = 365. Values above the diagonal representthe correlation matrix for the noncompliant sample; values below the diagonal represent the correlation matrix for thecompliant sample. All correlation values were significant at the p < .001 level.

Dow

nloa

ded

by [

The

Aga

Kha

n U

nive

rsity

] at

07:

00 1

6 O

ctob

er 2

014

ACADEMIC ENTITLEMENT AND INCIVILITY 329

Covariance matrices were derived for each sample using PRELIS 2.72, and the various modelswere estimated using LISREL 8.72 (Joreskog & Sorbom, 2005).

Measurement Invariance

Assessing model-data fit and specifying models

Multiple indicators of fit were used to evaluate the absolute and relative fit of each model.The χ2 provides a dichotomous decision regarding fit. Given the present study is focused on theapproximate fit of models, we also examined the comparative fit index (CFI) and the root meansquare error of approximation (RMSEA) values (Hu & Bentler, 1998). As rough guidelines, Huand Bentler (1999) considered values of CFI greater than .95 and RMSEA less than .06 to be in-dicative of adequate model-data fit. To evaluate local misfit, we examined standardized covarianceresiduals, which are the standardized difference between the actual and model-implied covari-ances for a pair of items. Standardized covariance residuals indicate how well the relationshipbetween item pairs is being reproduced by the model. Thus, a positive standardized covarianceresidual value indicates that the model is underestimating the relationship between an item pair,and a negative value indicates that the model is overestimating the relationship. For the scalarmodel, we also examined standardized mean residuals to assess local misfit. The standardizedmean residuals provide a standardized measure of the discrepancy between the actual and model-implied item means. Both standardized covariance and mean residuals are on a z score metric. Itis unfortunate that standardized residuals are rarely reported, that there are no clear cutoffs thatindicate misfit, and that their values are affected by sample size. Joreskog (1993) noted, “A goodmodel is characterized by a stem-leaf plot in which the residuals are symmetrical around zeroand with most residuals in the middle and fewer in the tails” (pp. 311–312). Thus, for each modelwe assessed if this pattern of residuals emerged.

We examined the �χ2 and �CFI values to assess the relative fit of the invariance models.The �χ2 significance test is an exact test of the additional misfit associated with constraining amodel. For example, when comparing the metric model to the configural model, we are interestedin answering if the additional misfit associated with constraining the factor pattern coefficientsto be equivalent across the two groups is statistically significant. Relative approximate fit can beassessed by examining changes in approximate fit indices between two nested models (Quintana& Maxwell, 1999; Steenkamp & Baumgartner, 1998). Cheung and Rensvold (2002) suggested�CFI greater than .01 indicates a significant decline in fit for the constrained model relative to theunconstrained model. Because the �χ2 values and �CFI values can result in different substantiveconclusions (French & Finch, 2006), we also examined changes in standardized covariance andmean residuals. If the residuals associated with the constrained model were much larger thanthose associated with the unconstrained model, this was an indication that the unconstrainedmodel was more appropriate to represent the data.

To estimate the models, the metric of the factor was established using a referent indicator. Thisscaling of the factor was done by constraining the unstandardized pattern coefficient between thelatent variable and Item 1 to a value of one. Before using Item 1 as the referent indicator, weensured its metric invariance using the method proposed by Rensvold and Cheung (2001). Thatis, we freely estimated the unstandardized pattern coefficient for Item 1 and then constrained itto be equal across the two groups, using each of the other items as referent indicators. Model fit

Dow

nloa

ded

by [

The

Aga

Kha

n U

nive

rsity

] at

07:

00 1

6 O

ctob

er 2

014

330 KOPP AND FINNEY

TABLE 2Tests of Measurement Invariance Across Compliant and Noncompliant Samples

Model χ2 df �χ2 �df CFI �CFI RMSEA

Configural 92.75∗ 40 0.98 0.06Metric 113.56∗ 47 20.81∗ 7 0.98 <0.01 0.06Scalar 123.39∗ 54 9.83 7 0.98 <0.01 0.06

Note. ∗p < .01.

did not decrease statistically or practically when comparing the constrained models to the freelyestimated model, indicating metric invariance for Item 1 across groups.

Configural invariance

A one-factor model was estimated separately for the two student samples. The model fitthe compliant sample data reasonably well, χ2(20) = 31.54, p = .04, CFI = 0.99, RMSEA =0.04, with standardized residuals symmetric around zero and no larger than 3.53. Similarly,the one-factor model fit the noncompliant sample data adequately. The χ2 and RMSEA valuessuggested some misfit, χ2(20) = 61.21, p < .01, CFI = 0.98, RMSEA = 0.08. However, thestandardized residuals were symmetric around zero and no larger than 3.51. Consequently, the fitof the one-factor model appeared to be sufficient for both samples.

Next, fit indices from a combined-sample configural model were estimated in order to serveas a baseline for the metric invariant model (see Table 2). As expected from examining the factormodels separately, the combined-sample configural model fit the data well. Parameter estimatesfrom this model are provided in Figure 1. Across both samples, the items functioned fairly well. R2

values ranged from .25 to .53 for the compliant sample, and from .22 to .64 for the noncompliantsample, indicating approximately a quarter to a half of the items’ variance was accounted forby the latent AE factor. Moreover, coefficient omega reliability estimates were .83 and .84 forthe compliant and noncompliant samples, respectively. Because the configural model fit the datawell, we examined metric invariance.

Metric invariance

The metric-invariant model fit the data well overall. Although the �χ2 value indicated aviolation of metric invariance, the �CFI value was negligible (�CFI < .001). To thoroughlyassess this slight misfit, invariance of each item was assessed individually by comparing theconfigural model to models where each item was constrained to have equivalent factor patterncoefficients across samples. Item 2, �χ2(1) = 4.63, p = .03; and Item 7, �χ2(1) = 5.30, p = .02,were found to have statistically significantly different factor pattern coefficients across the twostudent groups. Practically, the unstandardized factor loading for Item 2 was not very differentacross the two samples (1.07, 0.98). Moreover, the model constraining Item 2 to be metricinvariant across samples did not have a large increase in any standardized covariance residualscompared to the configural model, with the largest being 3.55. However, the unstandardizedfactor loading associated with Item 7 had a larger difference between the two groups when freely

Dow

nloa

ded

by [

The

Aga

Kha

n U

nive

rsity

] at

07:

00 1

6 O

ctob

er 2

014

ACADEMIC ENTITLEMENT AND INCIVILITY 331

FIGURE 1 Parameter estimates associated with the one-factor model. Values above the arrows are parameter estimatesfor the compliant sample (n = 380), and values below the arrows are parameter estimates for the noncompliant sample(n = 365). Standardized estimates are presented in parentheses. The unstandardized pattern coefficient representing therelation between the academic entitlement factor and Item 1 was fixed to 1.00 for both samples in order to set the metricof the factor. All estimated unstandardized pattern coefficients were statistically significant (p < .05). The latent varianceof the academic entitlement factor for each group is reported within the oval, with the compliant sample’s variance abovethe noncompliant sample’s variance.

estimated (0.83, 1.22), and constraining Item 7 to be metric invariant across samples increasedthe standardized covariance residual between Items 1 and 2 (3.53 to 4.30). In order to provide athorough investigation of the invariance of Item 7, we conducted scalar invariance and estimatedthe latent mean difference twice: once with the unstandardized pattern coefficient and interceptassociated with Item 7 constrained to be equal across groups, and once with this item’s patterncoefficient and item intercept freely estimated across groups. These analyses resulted in the samesubstantive conclusions and a nearly identical latent mean difference. Thus, although there wasstatistically significant misfit associated with the metric-invariant model, the lack of change inthe substantive conclusions suggests that this misfit was not practically significant. Therefore,we concluded that this rigorous assessment of metric invariance supported the claim of invariantpattern coefficients across the two student groups. As the metric invariance assumption wasupheld, scalar invariance was assessed.

Scalar invariance

The scalar model fit the data well. The model did not fit statistically or practically worsethan the metric invariant model. Moreover, the standardized mean residuals were small, with2.78 being the largest value. Thus, item intercepts were approximately equivalent across the twostudent groups. Given that both item intercepts and pattern coefficients were equivalent across

Dow

nloa

ded

by [

The

Aga

Kha

n U

nive

rsity

] at

07:

00 1

6 O

ctob

er 2

014

332 KOPP AND FINNEY

groups, a significant difference in latent means across groups would indicate an actual meandifference in AE.

Latent Means Modeling

Given the equivalence of factor structures, unstandardized pattern coefficients, and item intercepts,the latent mean difference on AE across the two student groups was examined. The latent meandifference was statistically significant and positive (κ = .29, p < .01) indicating that noncompliantstudents were significantly higher on AE than were compliant students. The latent effect size (.36)suggested a small to moderate effect. Specifically, the noncompliant sample was .36 standarddeviation higher in latent AE than was the compliant sample. As expected, this latent effectsize was greater than the effect size computed from the observed composite AEQ scores (d =.33; for noncompliant, M = 3.19, SD = 1.02; for compliant, M = 2.88, SD = 0.92), as thelatent estimate is adjusted for random measurement error. The discrepancy between the latentand observed effect sizes was small because AE was measured reliably in both samples (ω =.83 and .84), indicating little measurement error was present in the observed scores. The highreliability and small discrepancy between latent and observed-level conclusions indicates thatlatent variable techniques do not need to be employed to model AE with sufficient accuracy,which is encouraging for those who want to use the AEQ in research or practice.

DISCUSSION

This study was the first to empirically link AE to uncivil student behavior. Further, we determinedthat the AEQ functioned equivalently across compliant and noncompliant students, indicatingscores were comparable across groups. The empirical link between AEQ scores and uncivilstudent behaviors, as well as the utility the AEQ possesses for measuring AE in both behaviorallycompliant and noncompliant students, provide further validity evidence that AEQ scores representAE as it has been theoretically defined (e.g., Chowning & Campbell, 2009). The implications ofthese findings are subsequently discussed.

Relationship Between AE and Noncompliance

This study established a relationship between AE and uncivil student behaviors. It should benoted, however, that the effect (one third of a standard deviation) was relatively modest. Thiseffect corresponds to about 3% of the variance being shared between compliance status and AE.Thus, there are likely a number of other variables that explain noncompliance with universitypolicy. In addition, excluded variables may mediate or moderate the relationship between AEand noncompliance. For example, Brown and Finney (2011) found that psychological reactance(an aversive motivational state directed toward restoring a threatened freedom; Brehm & Brehm,1981) was related to compliance behavior, but it is unknown whether reactance is related toAE. In addition, research has found that mastery goal orientation and work avoidance relate tononattendance at a mandatory testing session (Swerdzewski et al., 2009) and these two motivationvariables are related to AE (Kopp et al., 2011). Given these issues, it is premature to make anycausal claims (e.g., high AE causes noncompliance) before conducting further research. Future

Dow

nloa

ded

by [

The

Aga

Kha

n U

nive

rsity

] at

07:

00 1

6 O

ctob

er 2

014

ACADEMIC ENTITLEMENT AND INCIVILITY 333

research should compare the usefulness of the AEQ for predicting university noncomplianceversus other developmental or cognitive measures, as well as investigate possible mediation ormoderation effects.

Future research should also explore other instances of noncompliance and incivility. Ournoncompliant sample consisted of students who did not attend a scheduled assessment session,and instead had to attend a makeup session. There are a number of reasons besides student incivilitythat an individual may miss a scheduled assessment session, despite numerous reminders (e.g.,extreme forgetfulness). Thus, the observed relationship between AE and noncompliance behaviormay be biased by these other variables. Moreover, this specific type of noncompliance is justone of many possible ways to represent uncivil student behavior. This study provided the initialempirical investigation of the relationship between AE and student incivility, and we hope futureresearch can expand the nomological net of AE with respect to student behavior. For example,future research could examine the relationship between AE and referrals to judicial affairs foruncivil conduct.

Last, it is also important to acknowledge that the mean entitlement scores for both compliant(M = 2.88, SD = 0.92) and noncompliant (M = 3.19, SD = 1.02) students were below the midpointof the scale. The majority of the students in both samples were not extraordinarily entitled, asthey largely disagreed with the items on the scale. The mean AE scores for both samples inthis study were slightly higher than was the mean AE score for freshmen college students(M = 2.76, SD = 0.84) found by Kopp and colleagues (2011). The difference in mean AElevels may suggest that sophomores are more academically entitled than are freshmen, althoughinvariance testing or longitudinal modeling between the two groups should be conducted beforemaking this claim. As was the case in the study by Kopp and colleagues (2011), AE scores hada slight positive skew. That is, although the majority of the participants disagreed with the items,some individuals agreed or strongly agreed with many of the items. The slight skew may suggestthat AE is a deviant trait, and only a small percentage of individuals are high in AE. Furtherresearch should determine if there is a level of AE that is especially problematic, and possiblyfocus interventions on those students.

Measurement Invariance

Configural, metric, and scalar invariance of the AEQ were established across compliant andnoncompliant student samples. Establishing measurement invariance across these two types ofstudents was important as it provided evidence that the AEQ functions similarly for studentswho behave differently and who have different levels of AE. As noted earlier, we had concernsthat noncompliant students could manifest their incivility through invalid responding; this studyprovided some empirical evidence that this did not occur.

Future research should focus on establishing measurement invariance across additional studentgroups. For example, different academic cultures may affect students’ conceptualization of AE,leading to a violation of measurement invariance. It is also unknown whether the AEQ willfunction equivalently for different types of college students (e.g., graduate students, communitycollege students). Last, the measurement invariance findings from this study, combined withfindings by Kopp and colleagues (2011), suggest that AEQ scores could potentially be comparedacross cohorts. That is, this study consisted of upperclassmen participants and supported theone-factor model uncovered by Kopp and colleagues (2011), who utilized a freshman sample.

Dow

nloa

ded

by [

The

Aga

Kha

n U

nive

rsity

] at

07:

00 1

6 O

ctob

er 2

014

334 KOPP AND FINNEY

Cohort invariance studies could be conducted to determine the comparability of AEQ scoresacross different cohorts, and longitudinal invariance studies could be conducted to determine thecomparability of AEQ scores across time. Given longitudinal invariance is established, researcherscould model change in AE over time and identify predictors of change in AE, which would bean essential step in constructing interventions to reduce AE.

Implications and Conclusions

Given the results of the present study and the research conducted by Kopp and colleagues(2011), AE could cause major problems for higher education. AE has been associated with ahost of maladaptive traits, such as external locus of control, work avoidance, putting forth lesseffort on tests, and being less focused on mastering concepts (Kopp et al., 2011). Further, thisstudy indicated that AE is related to student incivility. As a result, university administrators andprofessors may spend an inordinate amount of time and resources coping with students highin AE. For example, the makeup assessment sessions administered for noncompliant studentsrequire a great deal of additional funds and staff time. Educators may have to spend valuable timeand effort dealing with students who refuse to attend class or complete assignments in a timelymanner. Thus, reducing uncivil student behavior is important for university functioning.

If future research establishes a causal relationship between AE and these uncivil behaviors,university administrators and educators should attempt to reduce academic entitlement. Twengeand Campbell (2009) suggested that inducing gratitude in students may be an effective wayto reduce entitlement. University volunteer programs may engender gratitude in students byexposing them to those who are less fortunate, which may lower AE. Empirical research shouldbe conducted to determine which types of programs effectively reduce AE. The AEQ can then beused to assess these programs. For example, the community service programs at our universityare beginning to use the AEQ to assess whether their programs effectively reduce AE.

If programming is able to reduce AE, we believe the AEQ scores could also be used to identifystudents high in AE and specifically target those students for these interventions. Students couldcomplete the AEQ as entering freshman, and those scoring high on the measure could then beprovided with additional resources to lower their levels of AE. The short length of the AEQ allowsit to be easily integrated into existing university assessment, as it takes little time to administer.

Last, the individual educator can also attempt to curb AE in the classroom. If unconditionalrewards have led to an increase in AE, then it is possible that withholding these rewards mayreduce AE. Avoiding grade inflation and having high standards for students may reduce AEand student incivility over time. However, it should be noted that the most flagrant instances ofstudent incivility come when the student does not receive favorable evaluations (Achacoso, 2002;Dubovsky, 1986; Goodyear et al., 2010); thus, pursuing these tactics may inadvertently increasestudent incivility over the short term. As such, changes in educational standards would be mostconstructive if they occurred systemwide.

AUTHOR NOTES

Jason P. Kopp is a graduate student in the Assessment and Measurement Doctoral Programat James Madison University. His research focuses on utilizing structural equation modeling

Dow

nloa

ded

by [

The

Aga

Kha

n U

nive

rsity

] at

07:

00 1

6 O

ctob

er 2

014

ACADEMIC ENTITLEMENT AND INCIVILITY 335

techniques to evaluate different theories of student development. Sara J. Finney has a dualappointment at James Madison University as Associate Professor in the Department of GraduatePsychology and as Associate Assessment Specialist in the Center for Assessment and ResearchStudies. Much of her research involves the application of structural equation modeling to betterunderstand the functioning of self-report instruments.

REFERENCES

Achacoso, M. V. (2002). “What do you mean my grade is not an A?” An investigation of academic entitlement, causalattributions, and self-regulation in college students (Doctoral dissertation). Retrieved from University of TexasLibraries, http://repositories.lib.utexas.edu/handle/2152/2486

Benson, J. (1998). Developing a strong program of construct validation: A test anxiety example. Educational Measure-ment: Issues and Practice, 17(1), 10–17.

Bentler, P. M., & Wu, E. J. C. (2002). EQS for Windows user’s guide. Encino, CA: Multivariate Software, Inc.Branden, N. (1969). The psychology of self-esteem: A revolutionary approach to self-understanding that launched a new

era in modern psychology. New York, NY: Nash.Brehm, J. W., & Brehm, S. S. (1981). Psychological reactance: A theory of freedom and control. New York, NY: Academic

Press.Brown, A. R., & Finney, S. J. (2011). Low-stakes testing and psychological reactance: Using the Hong Psychological

Reactance Scale to better understand compliant and noncompliant examinees. International Journal of Testing, 11,248–270.

Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit indexes for testing measurement invariance.Structural Equation Modeling, 9, 233–255.

Chowning, K., & Campbell, N. J. (2009). Development and validation of a measure of academic entitlement: Individualdifferences in students’ externalized responsibility and entitled expectations. Journal of Educational Psychology, 101,982–997.

DeCarlo, L. T. (1997). On the meaning and use of kurtosis. Psychological Methods, 2, 292–307.Dubovsky, S. L. (1986). Coping with entitlement in medical education. New England Journal of Medicine, 315,

1672–1674.Finney, S. J., & DiStefano, C. (2013). Nonnormal and categorical data in structural equation models. In G. R. Hancock

& R. O. Mueller (Eds.), A second course in structural equation modeling (2nd Edition; pp. 439–492). Charlotte, NC:Information Age.

French, B. F., & Finch, W. H. (2006). Confirmatory factor analytic procedures for the determination of measurementinvariance. Structural Equation Modeling, 13, 378–402.

Goodyear, R., Reynolds, P., & Gragg, J. B. (2010, May). University faculty experiences of classroom incivilities: A criticalincident study. Paper presented at the annual meeting of the American Educational Research Association, Denver,CO.

Greenberger, E., Lessard, J., Chen, C., & Farruggia, S. P. (2008). Self-entitled college students: Contributions of person-ality, parenting, and motivational factors. Journal of Youth and Adolescence, 37, 1193–1204.

Hancock, G. R. (2001). Effect size, power, and sample size determinants for structured means modeling and MIMICapproaches to between-groups hypothesis testing of means on a single latent construct. Psychometrika, 66, 373–388.

Hirschy, A. S., & Braxton, J. M. (2004). Effects of student classroom incivilities on students. New Directions for Teachingand Learning, 99, 67–76.

Hu, L., & Bentler, P. M. (1998). Fit indices in covariance structure modeling: Sensitivity to underparameterized modelmisspecification. Psychological Methods, 3, 424–453.

Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteriaversus new alternatives. Structural Equation Modeling, 6(1), 1–55.

Joreskog, K. G. (1993). Testing structural equation models. In K. A. Bollen & J. Scott Long (Eds.), Testing structuralequation models (pp. 294–316). Newbury Park, CA: Sage.

Joreskog, K. G., & Sorbom, D. G. (2005). LISREL 8.72 [Computer software]. Lincolnwood, IL: Scientific Software.

Dow

nloa

ded

by [

The

Aga

Kha

n U

nive

rsity

] at

07:

00 1

6 O

ctob

er 2

014

336 KOPP AND FINNEY

Kopp, J. P., Zinn, T. E., Finney, S. J., & Jurich, D. P. (2011). The development and evaluation of the Academic EntitlementQuestionnaire. Measurement and Evaluation in Counseling and Development, 44, 105–129.

Lippmann, S., Bulanda, R. E., & Wagenaar, T. C. (2009). Student entitlement: Issues and strategies for confrontingentitlement in the classroom and beyond. College Teaching, 57, 197–204.

Moses, R., & Moses-Hrushovski, R. (1990). Reflections on the sense of entitlement. Psychoanalytic Study of the Child,45, 61–78.

Quintana, S. M., & Maxwell, S. E. (1999). Implications of recent developments in structural equation modeling forcounseling psychology. Counseling Psychologist, 27, 485–527.

Rensvold, R. B., & Cheung, G. W. (2001). Testing for metric invariance using structural equation models: Solving thestandardization problem. In C. A. Schriesheim & L. L. Neider (Eds.), Research in management (Vol. 1): Equivalencein measurement (pp. 21–50). Greenwich, CT: Information Age.

Satorra, A., & Bentler, P. M. (1994). Corrections to test statistics and standard errors on covariance structure analysis. InA. von Eye & C. C. Clogg (Eds.). Latent variable analysis (pp. 299–419). Thousand Oaks, CA: Sage.

Spellings, M. (2006). A test of leadership: Charting the future of U. S. higher education. Washington, DC: U.S. Departmentof Education.

Steenkamp, J. E., & Baumgartner, H. (1998). Assessing measurement invariance in cross-national consumer research.Journal of Consumer Research, 25(1), 78–90.

Swerdzewski, P. J., Harmes, J. C., & Finney, S. J. (2009). Skipping the test: Using empirical evidence to inform policyrelated to students who avoid taking low-stakes assessments in college. Journal of General Education, 58, 167–195.

Thompson, M. S., & Green, S. B. (2006). Evaluating between-group differences in latent variable means. In G. Hancock& R. Mueller (Eds.), Structural equation modeling: A second course (pp. 119–169). Greenwich, CT: InformationAge.

Twenge, J. M., & Campbell, W. K. (2009). The narcissism epidemic: Living in the age of entitlement. New York, NY:Simon & Schuster.

Ullman, J. B. (2006). Structural equation modeling: Reviewing the basics and moving forward. Journal of PersonalityAssessment, 87(1), 35–50.

West, S. G., Finch, J. F., & Curran, P. J. (1995). Structural equation models with non-normal variables: Problems andremedies. In R. H. Hoyle (Ed.), Structural equation modeling: Concepts, issues, and applications (pp. 56–75).Thousand Oaks, CA: Sage.

Dow

nloa

ded

by [

The

Aga

Kha

n U

nive

rsity

] at

07:

00 1

6 O

ctob

er 2

014