chapter  6
28 Pages

Two Simple Designs

Many of the simplest experiments compare two conditions, perhaps using the two-independent-groups design or the paired design. You may be familiar with these designs in the context of the independent-groups t test and paired t test. This chapter discusses how to use CIs to present and interpret results from these two designs. Here’s the agenda:

• The dreadful ambiguity of conventional figures

The two-independent-groups design:

• Displaying the data • A rule of eye for estimating the p value • Presenting your own data • Illustrating the great variability with replication

The paired design:

• Displaying the data • The correlation between measures • Presenting your own data • Illustrating why this design can be sensitive

Your wine club gives you the onerous task of evaluating a fine red wine from the Napa Valley, California, and another from McLaren Vale, South Australia. To avoid bias, you mask the bottles and label them A and B. It’s easy to recruit volunteers, but you need to choose whether to use independent groups of tasters for the two wines, or a single group who taste both, some A then B, and others B then A. After collecting the data, your assistant prepares Figure 6.1. What extra information does a reader need to understand it?