East Carolina University

Department of Psychology

Students in PSYC 7433 are required to deliver, in class, a research paper in which they present the results of a multivariate analysis. The data for such analysis may be real data that have been collected by the student (such as the data for the student's thesis or dissertation), real data that have been obtained from an archive, or simulated data (obtained from a text book or similar source, or simulated by the student).

The statistical analysis must include one or more of the following: log-linear analysis of multidimensional contingency tables, multiple regression analysis which includes testing of a moderation or mediation model, logistic regression, canonical correlation analysis, discriminant function analysis, multiple analysis of variance, multiple analysis of covariance, the multivariate approach to repeated measures ANOVA, principal components analysis, factor analysis, path analysis, univariate factorial ANOVA (at least three factors), univariate factorial ANCOV (at least two factors, at least one covariate), or other techniques for which the student obtains prior approval.

The presentations will like those one sees at a scholarly convention. Each presenter will be allocated 20 minutes, and asked to reserve the last 5 minutes for questions from the audience. The presentation should be accompanied by a PowerPoint show.

The presentation should include:

- A succinct introduction of the question being investigated, including a brief review of the relevant literature.
- A description of the methods that were employed to collect the data. If the data were simulated, describe the real-world process that was being simulated as well as the method actually employed to simulate the data. If the data were obtained from an archive, cite it and describe how they were originally obtained.
- The results of the statistical analysis, including descriptive and inferential statistics. This, and interpretation of the results, should be the focus of the presentation. Given that this is a statistics class, you might want to spend a bit more time explaining how you did the statistical analysis than you would with the typical presentation at a scholarly convention.
- Discussion of the results. Evaluate and interpret the results, relate them to past research and/or theory, mention any limitations, discuss practical implications, suggest additional research, and so on.

Each presenter will also prepare an APA-style abstract and post it in the Research Presentation Forum in BlackBoard. Do not prepare the abstract in Word, as text pasted from Word into BlackBoard is not properly displayed. Instead of Word, use a plain text editor (like Notepad) or an htm editor or just type the abstract directly into BlackBoard. Consult the APA Publication Manual for details on what should be included in the abstract. The abstract should not exceed approximately 120 words in length.

To post your abstract, enter the forum, click new thread, and then type your last name as the title of the thread. Enter your abstract in the message area. Then create a bulleted list containing the links to the supporting documents -- click the Attach File icon and then browse to location of the file on your computer. The Insert Content Link window will ask "Launch in new window?" Tell it "yes."

After posting the abstract in BlackBoard, you should attach supporting documents

- The data in a plain text file, Excel file, or system data file.
- The statistical program(s) (and, if you used SAS, a copy of the SASLOG) in a Word document. If you used SPSS, either send me the journal file with all of the commands you used or leave the syntax in the output file.
- The statistical output in a separate Word document. If the output is excessively long, cut and paste into a shorter document the most important parts of the output. Annotate the output in ways which will help me read it.
- A reference list, in a Word document, APA-style.
- If you simulated your data, a copy of the program you used to simulate them.

Prefix each file name with your last name followed by an underscore character and then indication of the type of file -- for example:

- Fechner_PsychophysicsOfTheDead.pptx -- the PowerPoint show
- Fechner_Output.docx -- the statistical output from SAS and/or SPSS -- Please paste or export the output to a Word document rather than submitting a system file.
- Fechner_Data.dat -- the data in a plain text file
- Fechner_Data.### -- where ### is something like sas7bdat -- the data in a SAS system data file.
- Fechner_Data.xlsx -- the data in an Excel file
- Fechner_Program.txt -- the SAS program followed by the SAS log
- Fechner_Data.sav -- the data in SPSS format
- Fechner_Journal.jnl -- the syntax from SPSS
- Fechner_Simulator.sas -- the program used to simulate the data

I suggest that you zip your files to make it less of a hassle to upload them into BlackBoard.

I suggest that have a good look at the examples provided by previous students in this class: Jonathan Highsmith, Taylor Rush, and Hotaka Maeda.

There are lots of data sets in archives out on the Internet. You could retrieve one of these to use for your project. Links to some sources can be found on my **Data Files Page**. It is not appropriate to use any data file on Dr. Wuensch's StatData Page -- those are the files that we have been using in class.
If you are going to be using a data file you found on the Internet, email all
members of the class to claim that file as yours. It is not appropriate
for two or more students to use the same data file, unless Professor Karl
approves a partitioning of the variables, one set for one student, a different
set for the other student.

Here are some suggestions which might help you avoid some of the mistakes that have been common in the past:

1. Be sure to check your data regarding whether or not they meet the assumptions of the procedures you are using (for example, for parametric ANOVA, normality and homogeneity of variance). If there are serious violations of assumptions, take appropriate corrective action (transformation, adjustment of degrees of freedom, etc.). If you cannot correct the problem, at least acknowledge that the problem exists.

2. Do not neglect to report the important descriptive statistics.

3. In your program, name the variables in descriptive fashion ("income," "sex," "race," "education" rather than "Y," "A," "B," "C") and use PROC FORMAT or VALUE LABELS or such to provide descriptive labels to numeric codes (for example, "female" and "male" rather than "0" and "1". Annotate your output to make it easier for Karl to read and understand it.

Contact Information for the Webmaster,

Dr. Karl L. Wuensch

This page most recently revised on the 31^{st }of August, 2013.