Research DesignThe research design for this study uses structural equation modeling (SEM) to examine structural relations between variables representing employee satisfaction, income, and lifestyle with to determine the attitude and motivation of public servants to commit fraud. SEM is useful in determining relationships between attitudes and behaviors, among others, that is useful in research regarding what people will or would do in given scenarios (Bakshi & Mishra, 2016). Multivariate analysis can be used to conduct t-tests and other multivariate determinations of difference and correlation; however, this can be very technical and time consuming when there are more than two variables (Kline, 2015).

You're lucky! Use promo "samples20"
and get a custom paper on
"Structural Equation Model of Employee Satisfaction, Income, Lifestyle, and Fraud Intentions"
with 20% discount!
Order Now

Research Respondents and Sampling Technique
A purposive sample of public servants was used to recruit survey respondents. While convenience samples can limit the representational validity of the results, because of the narrow criteria for inclusion it was not possible to use random sampling techniques (Etikan, Musa, & Alkassim, 2016). The determination of sample size for SEM is more complex than the determination for other statistical approaches due to the use of multiple variables (Bakshi & Mishra, 2016). It is commonly repeated in the literature that samples of over 200 respondents are needed for SEM (Snoj et al., 2004). Wolf, Harrington, Clark, and Miller (2013) disagree, based on their study to determine optimum sample sizes in SEM data for applications. After examining impacts of changes to variables and other parameters, they determined that the optimal sample size varied greatly based on the extent of the relationships between the variables, bias and missing data, leading to a range of 30 to 450 being the ideal sample, contingent on what was being measured (Wolf et al., 2013).

Research Instrument
The research instrument to be used is a survey questionnaire which builds on previously validated measurement instruments in the independent variable areas of public service employee satisfaction, respondent reported data regarding income, and lifestyle. The survey is based on other instruments, but for the most part attempts were made to use the precise same question to ensure validity and comparability with the original resource. The survey questionnaire included Likert scale questions, multiple choice questions and true of false questions.

Employee satisfaction is measured in the Federal Employee Viewpoint Survey (FEVS), which has been in use by the Office of Personnel Management for 15 years (Fernandez, Resh, Moldogaziev & Oberfield, 2015). This study uses a subset of the survey questions which measured job satisfaction. Income was represented by respondent reported employment, settlement, maintenance and investment income. Lifestyle is measured using questions in the survey questionnaire based on the Extravagant Lifestyle Scale (ELAS) (Balogun, Selemogwe & Akinfala, 2013). Motivation was considered to exist where the lifestyle variable data was over the estimated threshold based on income; a negative attitude was considered to exist where there was little job satisfaction, as determined by a threshold score based on previous surveys which were conducted using FEVS. Where both motivation and attitude met the threshold, that respondent was considered to be a risk for intention to commit fraud. Based on the responsibilities of public servants, and the need for a high level of trust, it was assumed that those employees who so wished did have access to opportunities to commit fraud.

Reliability for this study should be high, given that the tools used and the study itself are easy to replicate for other populations, similar populations or the same population. Further reliability comes from the fact that the basis for the instrument was previously validated in other studies and found to provide an accurate representation. Validity is someone limited as discretion was used to determine thresholds, however the method was open to modelling to determine the best fit of thresholds and justifications for those thresholds for each variable. A further factor is the lack of baseline data that included people with confirmed intention to commit fraud behaviors. Those individuals who had been caught committing fraud would no longer be captured as having intention to commit fraud by this SEM tool as they would no longer have the job from which to measure satisfaction, and their lifestyle and income was likely transformed by these revelations.

Statistical Tools
The focus of statistical tools will be the determination of the extent to which the three variables (employee satisfaction, income and lifestyle) contribute to individual attitude and motivation, and how these two factors are correlated to intent to commit fraud. The Statistical Package for the Social Sciences (SPSS) supports structural equation modelling using Amos 19 (Arbuckle, 2010). First, confirmatory factor analysis is used to determine that there is in fact a relationship between the variables (Arbuckle, 2010). Following this verification, the structural equation model is used to statistically analyze the extent of those relationships. Amos 19 allows the user to load the data for each represented variable, and it then automatically creates the output for analysis of the relationship between the variables.

The regression model used to determine fraud intentions will be based on the correlation of attitude and motivation, and these variables will be determined through the use of the three identified variables. The data for these variables will be collected using the survey instrument.

This chapter provided an overview of the decisions and approaches to the methodology of this study of behavioral relationships in fraud intention of government employees. This included the quantitative research design, the purposive sampling approach, the research instrument and its bases as well as the statistical analysis tools. The next chapter provides the findings that resulted from using these procedures and analysis of the data.

  • Arbuckle, J. L. (2011). IBM SPSS Amos 19 user’s guide. Crawfordville, FL: Amos Development Corporation, 635.
  • Bakshi, M., & Mishra, P. (2016). Structural equation modelling of determinants of consumer-based brand equity of newspapers. Journal of Media Business Studies, 13(2), 73-94.
  • Balogun, S. K., Selemogwe, M., & Akinfala, F. (2013). Fraud and Extravagant Life Styles Among Bank Employees: Case of Convicted Bank Workers in Nigeria. Psychological Thought, 6(2), 252-263.
  • Etikan, I., Musa, S. A., & Alkassim, R. S. (2016). Comparison of convenience sampling and purposive sampling. American Journal of Theoretical and Applied Statistics, 5(1), 1-4.
  • Fernandez, S., Resh, W. G., Moldogaziev, T., & Oberfield, Z. W. (2015). Assessing the past and promise of the Federal Employee Viewpoint Survey for public management research: A research synthesis. Public Administration Review, 75(3), 382-394.
  • Kline, R. B. (2015). Principles and practice of structural equation modeling. Guilford publications.
  • Wolf, E. J., Harrington, K. M., Clark, S. L., & Miller, M. W. (2013). Sample size requirements for structural equation models: An evaluation of power, bias, and solution propriety. Educational and psychological measurement, 73(6), 913-934.