ConMET (Construct Measurement Evaluation Tool) is an application that makes it easier to analyze typical CFA models that you have to report in most journals. With this app you can work much faster because it automatically creates, tests, and compares nested measurement models. It also provides you with a table that summarizes the results, which you can insert directly into your manuscript. Another advantage is that the application has a point-and-click user interface, which means you don't have to program to run CFAs.
Step 1: Loading Data
First, you need to load your data. Currently SPSS and Excel files are accepted.
Step 2: Selecting Items
Once the data is uploaded, you can select items that constitute your measurement model. There are two options: (1) let the app automatically detect the factors or (2) manually select which items belong to which construct.
Clear labeling is crucial for the application to work well. Choose a clear name for a scale and then add the item number at the end of the name. For example, the first item of a 5-item performance scale should, for instance, be labelled as “Perf1” or “Performance1”. To indicate that an item is reverse coded, you could add an “R” or “Rev” at the end. For example, “Perf3R” or “Perf3_Rev”.
The app clusters items using a simple process:
- First it removes characters that indicate that an item was reversed. It will remove the following substrings if they appear at the end: “R”, “Rev”, “REV”, “_R”, “_Rev” and “_REV”. For example, “OCB1R” becomes “OCB1”. Or “CSR1_REV” becomes “CSR1”.
- Then it deletes all digits in the variable names. For example, “OCB1” becomes “OCB”, “Trust1_2” becomes “Trust_”, or “Proc13” becomes “Proc”. Now all items belonging to the same factor should have the same unique substring.
- After this, it checks if some variable names still end with a special character (e.g., “_”) and this will be removed.
- Finally, the stripped down variable name will be used as the label of the latent factor (e.g., OCB) and all items that have these same string will be treated as an indicator (e.g., OCB1 and OCB2_R but not OCBO1).
Step 3: Run CFA
Once you have selected “Extract Factors”, a proposed measurement model appears. This should reflect the model you want to test. To run the model you can go the “CFA Specifications Tab”.