Psych 350
Data Collection
1. Data Collection
Now that you have your 10-item questionnaire available via metaForm, please print the address/URL for your study clearly on a piece of paper and pass that paper to the person to your right (or in whatever direction makes sense, given where you're sitting). Someone to your left should be passing you their URL.
Take 1 to 2 minutes to fill out the survey that was passed to you. When you are done and/or on the TA's call, please pass that URL to the person to your right and get a new one from the person on your left. Repeat this process until everyone has had the opportunity to take everyone else's personality questionnaire.
Please enter a consistent alias or nickname in the ID field of each survey, assuming that the demographic questions were used.
2. Import Your Data
After everyone has had an opportunity to take your survey, it is time to import your data. Please do so by using the download data section on the metaForm homepage. Download your txt file to your desktop and import it into SPSS. Remember that the SPSS Import Wizard has 6 steps and you can roll with all the defaults, minus 2: (a) Indicate that the variable names are on the first row of the text file and (b) make sure "comma" is the only box selected in the delimiter section (uncheck "space").
Once you're imported your data successfully, be sure to save it as an SPSS data file. Email it to yourself at the end of class.
3. Reverse Key
If you have any items that need to be reverse keyed, please do so. Save those reverse keyed items with a name you'll recognize later.
4. Test the Uni-Dimensionality of your 10 Items
Conduct a Principal Components Analysis (PCA) and study the Scree Plot to determine whether your items behave in a uni-dimensional fashion. Remember that the rule of thumb we'll use in this class is as follows: If there is only one component with an eigenvalue of 1 or higher, then the items are behaving in a uni-dimensional fashion. If there are two or more components with eigenvalues of 1 or higher, then your items are not functioning in a uni-dimensional manner. Figure out which items don't really "belong" and re-run the analysis without them.
Look at the component loadings matrix. These represent the correlations between each item and a "composite" of all the items generated by the PCA. Which variables correlate highly (postiviely or negatively) with the composite? Which ones correlate less strongly? Pick the 5 items that correlate the most strongly and, for now, consider those 5 items to represent the "short" version of your personality measure.
5. Alpha Reliability
What is the Cronbach's alpha for your 10-item test? Is it high? Is it low?
Use the advanced options in SPSS to determine what the alpha would be when each item is systematically excluded from the test. Which 5 items contribute the LEAST to the reliability of the scale? If you were to select the 5 "best performing" items according to these criteria, what would the alpha be for your shortened 5-item test?
Are these the same 5 items you would keep if you used the PCA analyses?