Statistical Analysis with Excel For Dummies, 4th Edition

Chapter 12

Testing More Than Two Samples

IN THIS CHAPTER

Understanding why multiple t-tests won’t work

Working with ANOVA

Taking the next step after ANOVA

Working with repeated measures

Performing a trend analysis

Statistics would be limited if you could only make inferences about one or two samples. In this chapter, I discuss the procedures for testing hypotheses about three or more samples. I show what to do when samples are independent of one another, and what to do when they’re not. In both cases, I discuss what to do after you test the hypotheses.

I also introduce Excel data analysis tools that do the work for you. Although these tools aren’t at the level you’d find in a dedicated statistical package, you can combine them with Excel’s standard features to produce some sophisticated analyses.

Testing More Than Two

Imagine this situation. Your company asks you to evaluate three different methods for training its employees to do a particular job. You randomly assign 30 employees to one of the three methods. Your plan is to train them, test them, tabulate the results, and make some conclusions. Before you can finish the study, three people leave the company — one from the Method 1 group and two from the Method 3 group.

Table 12-1 shows the data.

TABLE 12-1 Data from Three Training Methods

Method 1	Method 2	Method 3
	95	83	68
	91	89	75
	89	85	79
	90	89	74
	99	81	75
	88	89	81
	96	90	73
	98	82	77
	95	84
		80
Mean