Analysis of Job Characteristics Dataset: Utrecht Work Engagement Scale Subfactors

Introduction/Literature Review

The dataset contains survey data from a single organization. There are 2000 simulated observations on a number of variables. Most of the variables in the dataset are one item variables (e.g. mean scores) relating to organisaitonal varibales or demographic variables. The important element of the dataset is the 3 sub-factors from the Utrecht Work Engagement Scale (UWES; Schaufeli, Bakker & Salanova, 2006); engagement, absorption and dedication. You will need to calculate the mean score of these subfactors and at least one should be included in your analysis. The aim of the analysis is to examine either

• Which organisational variables predict some or all of the work engagement subfactors.
• Which work engagement subfactors predict organisational variable/s

This assignment involves writing a report to present the results from the analysis of the "jobcharact.csv" dataset. The analysis of the data at least one multiple regression model including the relevant assumption tests.

Thus you can use the work engagement subfactors as IVs or DVs. It is suggested that you first read the Schaufeli et al. (2006) paper which presents validation of the Utrecht Work Engagement Scale.

For the regression model you can conduct a replication study (justifying why replication is important), create new hypotheses justified by existing research, or use one of the suggested hypotheses (you will need to do an independent literature search to locate the studies to support these suggested hypotheses).  The analysis can be for multiple regression or hierarchical regression. 

The report is in effect a short academic paper of 2000 words. Thus, it should include an abstract, an introduction/literature review, a section on methodology, a results section and a discussion. The main focus of this report is the methods and results sections and it is expected that you would devote about 1000 words to these. The remaining 1000 words can be used for your introduction and discussion. 

The introduction should present arguments that lead towards your hypotheses.

For the methodology, you will need to describe the design and participants, e.g., the characteristics of the sample, etc. Next include a materials section and describe the scales used (see FAQ below).  Plus a procedures section; for any info not provided you can make assumptions (as long as these are reasonable). For instance, you can make your own assumptions about the response rate or whether the survey was distributed online or not. 

You will need to include a regression hypothesis based on your literature review to test your hypotheses. This can involve any number of variables from the dataset but must include at least one of the Utrecht Work Engagement Scale subfactors. For example, you may evaluate the effect of the engagement factors on burnout, or performance or job satisfaction (see suggested hypotheses section below).  Consider the needs for and justify any control variables. 

• a descriptives/correlation table of the continuous variables used in your analysis also including the results of your reliability analysis (cronbach's alpha)
• a table to present the results from the regression analysis

Methodology

Note that none of the items in this dataset require reverse scoring.

Below are some suggested hypotheses, for which relevant literature exists.  You may choose one set of these to use for your assignment but will have to do a literature search to find the relevant studies to support these hypotheses.  Alternatively, you can create your own hypotheses supported by literature.  Any number of the UWES factors can be the IV or DV in your hypotheses.

Skills variety positively predicts dedication

Job autonomy positively predicts dedication

Performance feedback positively predicts dedication

Vigour negatively predicts burnout

Dedication negatively predicts burnout

Absorption positively predicts burnout

Autonomy positively predicts dedication

Feedback positively predicts dedication

Task significance positively predicts dedication

Autonomy positively predicts absorption

Feedback positively predicts absorption

Task significance positively predicts absorption

Vigour positively predicts job satisfaction

Dedication positively predicts job satisfaction

Absorption positively predicts job satisfaction

Q: Can we use the citations from the lectures to support justifications?

A: Yes, you can use the citations from the lecture, but will need to find the full reference.  This should be easy to trace online, or in the Andy Field book.  You can also cite Andy Field if you read something in there that is appropriate.

Q: I am worried I can’t reach the word count.

A: If you explain each step of the analysis this should bring you up to the word count.  Explain what you did, justify each of the decisions you made, include supporting evidence where you can (e.g. citations). The aim here is to let the reader know that you understand the steps of the analysis.

Q: How do we report the results?

A: Use the exercise from Week 4 as a guide, But Do Not Plagiarise The Wording , this will result in marks lost.

Q: Do we reanalyse if we remove outliers?

A: Yes. Conduct the regression and ask SPSS to identify outliers that are 3 SDs away from the mean (see the slides). If outliers are identified decide on whether to remove that case (e.g. that one response). You will then need to rerun the regression without that response/s.  For this reason, it is important that you look at the outliers output first before spending time reporting the results – just in case it needs to be rerun.  When you remove the outliers it is best to create a new version of the SPSS datafile so that you can go back to your original if for some reason you decide to later.

Q: Can I get a first by just doing a multiple regression, e.g. without doing a hierarchical regression or a moderation/mediation

A: Yes, excellent reports using any type of regression analysis can score a 1^st.

Q: If I want to include demographic variables as control variables should I use hierarchical analysis?

A: The answer to this question is two-fold. 

• First, you must have a justification for including any variable (demographic or psychological) as a control variable, e.g.
• There is prior evidence (that you can cite) which shows it is related to your DV.
• You run a correlation analysis before your regression analysis to examine if any demographics are related to your DV which would justify their removal.
• Second, although some papers show that demographics are entered in the first block of a hierarchical regression, this is not always the convention and some papers just include it in the multiple regression. I am happy for you to use either method.

Q: When do I create the variables?

A: Before you conduct the regression you need to create a score each of the factors of the UWES (i.e. vigour, dedication, absorption).  Do this by using the mean of the items for each factor.

Q: Do we also need to talk about/know/include how variables such as burnout, task significance, task variety etc are measured?  Because we don’t have any information on how they came about.

A: That is correct, with this dataset you do not have the name of the original scale, how many items, an example item, the response scale or the Cronbach’s alpha.  In a research report you would expect to see all of this information.  My advice would be to use some creative licence here to demonstrate that you know what should go in a proper research report.  You could provide a citation for a well-known scale for burnout (e.g. Maslach Burnout Inventory), include the details I’ve pointed out above and invent a cronbach’s alpha.