This Lab-assignment is about factor analysis in two parts; one for Exploratory Factor Analysis (using SPSS) and one for Confirmatory Factor Analysis (using AMOS).
The data used for this lab is provided in a data file DataLab2.xls. Export the data to SPSS and save the data. AMOS uses a SPSS data file (.sav).
Lab 2 requires reading of chapters 3, 9 and 10 in Hair et al (2018), in particular the applied parts of the chapters. Some of the instructions are implicitly pointed out by the reading instead of explicit lab instructions.
It is essential that you describe what you are doing and why. Just providing output tables from SPSS/AMOS without showing understanding of what you are doing is not what this exercise is about. It is also essential that you show that you have taken part of the literature via your explanatory texts and the use of references.
Part 1: Exploratory Factor Analysis Read chapter 3 of Hair et al (2018).
Use the DataLab2.xls file and export the data available in the sheet “data” to SPSS. In the data sheet “description”, you will have variable names and data levels to each variable. Export the data and characteristics into SPSS. You can rename the variables if you want a personal meaning to the variables.
Select the data necessary to do an EFA (Exploratory Factor Analysis). All data used must fulfill the criterions of an EFA. Present summary statistics of the data you are using. Explain the selection criterion and why you (if you do) deselected variables.
Reduce the data by the EFA in SPSS. The analysis should include creating (and naming) latent variables, interpretation of the loadings, communalities etc. Explain what you are doing and why using references to Hair et al. (2018) Note that in SPSS you cannot select any EFA described in Hair et al (2018), but you can describe which one you are using.
Evaluate the EFA and undertake adjustments to cross-loading problems. Describe your adjustments and explain why you are doing these adjustments with references to Hair et al (2018). Re-analyze the outcome.
Submission part 1
- Summary statistics of the used data and a description of the criterion used for deselecting variables.
- Analysis of the EFA.
- Descriptions of adjustments made and explanations why.
- Re-analysis of the EFA.
Part 2: CFA (Confirmatory Factor Analysis)
In this chapter you should do a simple CFA using the same data as in Part 1 (DataLab2.xls). The data must be in SPSS format (.sav) before you start.
- Read chapters 9 and 10 in Hair et al (2018)
- In AMOS, draw a graph like the one presented below and open the data. Use variable (A) as your endogenous variable and use two sets E,F,G and z1-z4 as your exogenous variables as inputs in your constructs. These are the observed values. Use drag-anddrop once you have opened the data in ‘list variables in data’.
- Prepare your analysis
• Go to plug ins => ‘name unobserved variables’. This action will name all your unobserved variables (errors and latent variables). You can rename and/or label your variables if you double click on the item.
4) Go to ‘analysis properties’ and chose the kind of analysis you want to undertake.
For instance;
- In estimation: do a Maximum Likelihood in discrepancy.
- In output: select indirect, direct and total effects, factor score weights and tests for normality and outliers.
- Go to analyze => ‘calculate estimates’ You will get a note about F1 and F2. For your first estimation, just proceed. (But you should adjust in (7))
- Go to view => ‘text output’. Analyze the outcomes. Describe the latent variables, the relationships and the GOF.
- Elaborate with the model. See if you can increase the GOF by changing the relations in any way. You can for instance account for correlations between the measured variables (the error terms), change the relationships from one-way to mutual relationships by changing the arrows or add/drop variables or constructs or changing the analysis properties. Note that this part of the examination is just supposed to be a laboratory exercise and engage in a critical mindset to the outcome, not one to search for optimum in infinity.
Note: If you add other measured variables, you need to add an error (unobserved) variable or you will receive an error note when you run the test). This is a circle with a single headed arrow.
Submission part 2:
- A summary of the most essential outcomes from the initial test (variables, relationships and GOF). Describe the output you are analyzing and the results.
- Describe briefly how you elaborated with the outcomes and your results.
Review Lab 2
| Name | ||||||
| Not included | Included | Complete | Passlevel | Well done | Very well done | |
| EFA data selection | ||||||
| EFA analysis | ||||||
| Adjusted EFA | ||||||
| CFA description of analysis items | ||||||
| CFA analysis of results | ||||||
| CFA elaboration | ||||||
| Use of the literature | ||||||
| Passing grade (Yes/no) | ||||||
| Comment |
