E2 Design

In many cases, this will be just as described in Section 3.1 of the book. However, in a few cases you may have a large number of DVs to report. If so, then you may only have room to describe and illustrate each type of DV in the DESIGN.

For example, you might have measured attitudes using a set of semantic differential scales, such as good-bad, strong-weak, pleasant-unpleasant, etc. In the DESIGN you would mention that the attitude items were measured using semantic differential scales and leave description of the precise wording of the attitude items and the actual scales used on the semantic differential until the MATERIALS.

On the other hand, if you had a set of “belief” items that you measured with the same 7-point scale, you could state in the DESIGN that the belief items were measured using a 7-point scale. You would describe in the MATERIALS the precise wording of the belief items themselves and the end points of the scale (and any other verbal labels used on the scale).

Either way, you should make sure that you describe the DVs as they are reported and analysed in the RESULTS. For example, you may run the analyses on a mean or total score obtained from combining a set of separate ratings (for more on this see Section 13.9.2 of the book and E7.1 of this Web site). If so, this composite score is the DV and not the individual ratings that comprise it. For example, you might write something like:

Just a word of warning: If you do have a large number of DVs to analyse and these are not combined into composite scores, then be careful about your type I error rate. The more statistical tests that you run on a set of data, the more you increase the chances of making a type I error. You may need to make adjustments to the significance level for each test to keep the overall type I error rate to an acceptable level. You will find more about type I error in Section 11.3 of the book. You can find discussions of this problem and ways of controlling for it in most textbooks of statistics. In the statistics books paired with the latest edition of Designing and Reporting Experiments in Psychology, you will find this in Chapter 16 of Greene and D’Oliveira’s Learning to Use Statistical Tests in Psychology and in Part Five of Pallant’s SPSS Survival Manual.

It may be that you intended some of your items to measure the same underlying variable. If so, you can avoid some of the above problems if you are able to combine them into a composite measure or scale. The statistic Cronbach’s alpha can help you decide whether this is possible and which of the items to include. You can find out more about this in Section 13.9.2 of the book and Section E9 of this Web site.