Data Analysis with Microsoft Excel: Updated for Office 2007

Chapter 10 Analysis of Variance 413

The assumptions for two-way ANOVA are essentially the same as those for

one-way ANOVA. For one-way ANOVA, all observations on a treatment were

assumed to have the same mean, but here all observations in a cell are assumed

to have the same mean. The two-way ANOVA assumes independence, con-

stant variance, and normality, just as the one-way ANOVA (and regression).

Two-Way Analysis Example: Comparing

Soft Drinks

The Cola workbook contains data describing the effects of cola (Coke, Pepsi,

Shasta, or generic) and type (diet or regular) on the foam volume of cola soft

drinks. Cola and type are the factors; cola has four levels, and type has two

levels. There are, therefore, eight combinations, or cells, of cola brand and

soft drink type. For each of the eight combinations, the experimenter pur-

chased and cooled a six-pack, so there are 48 different cans of soda. Then

the experimenter chose a can at random, poured it in a standard way into a

standard glass, and measured the volume of foam.

Why would it be wrong to test all of the regular Coke fi rst, then the diet

Coke, and so on? Although the experimenter might make every effort to keep

everything standardized, trends that infl uence the outcome could appear.

For example, the temperature in the room or the conditions in the refrig-

erator might change during the experiment. There could be subtle trends in

the way the experimenter poured and measured the cola. If there were such

trends, it would make a difference which brand was poured fi rst, so it is best

to pour the 48 cans in random order.

The Cola workbook contains the variables shown in Table 10-2.

Table 10-2 Data for Cola Workbook

Range Name Range Description

Can_No A2:A49 The number of the can (1–6) in the six-pack

Cola B2:B49 The cola brand

Type C2:C49 Type of cola: regular or diet

Foam D2:D49 The foam content of the cola

Cola_Type E2:E49 The brand and type of the cola

To open the Cola workbook:

1 Open the Cola workbook from the Chapter10 data folder.

2 Save the workbook as Cola ANOVA. The workbook appears as

shown in Figure 10-15.