Magazine Ads. Advertising researchers F. Shuptrine and D. McVicker wanted to determine whether there were significant differences in the readability of magazine advertisements. Thirty magazines were classified based on their educational level-high, mid, or low - and then three magazines were randomly selected from each level. From each magazine, six advertisements were randomly chosen and examined for readability. In this particular case, readability was characterized by the numbers of words, sentences, and words of three syllables or more in each ad. The researchers published their findings in the paper "Readability Levels of Magazine Ads" (Journal of Advertising Research, Vol. 21. No. 5, pp. 45-51). The number of words of three syllables or more in each ad are provided on the WeissStats site.

MINITAB can generate individual normal probability plots and sample standard deviations.

MINITAB (low) method: Normal probability plot

Step 1: Select Probability Plot from the Graph menu.

Step 2: Click OK after selecting Single.

Step 3: In Graph variables, select the Low column.

Step 4: Press the OK key.

Medium normal probability plot

Procedure for MINITAB:

Step 1: Select Probability Plot from the Graph menu.

Step 2: Click OK after selecting Single.

Step 3: In the Graph variables section, type Medium in the column.

Step 4: Click the OK button.

Normal probability plot using the HIGH MINITAB technique

Step 1: From the Graph menu, choose Probability Plot.

Step 2: Click OK after selecting Single.

Step 3: From the Graph variables menu, choose High.

Step 4: Press the OK key.

The standard deviations of the samples are provided below, based on the normal plots:

• High: 12.10;
• Medium: 9.76;
• Low: 9.93

• A residual analysis is carried out using MINITAB.

MINITAB procedures:

Step 1: From the drop-down menu, choose Stat > ANOVA > One-Way.

Step 2: In Response, fill in the Words field.

Step 3: In Factor, in the column, write EDLEVEL.

Step 4: In the graph, select normal probability of residual and residual versus fits.

Step 5: Press the OK key.

Assumptions: -

• The sample chosen should consist of simple random samples.
• The samples have nothing in common.
• The population is about average.
• Standard deviations in the population are the same.
• The ratio of the highest to the lowest standard deviation is,

$\text{Ratio}=\frac{12.10}{9.93}$

$=1.22$

• The ratio of the biggest and smallest standard deviations cannot exceed 2.
• This shows that the equal standard deviation assumption is not broken.

• The data distribution is not approximately normally distributed because the points in the normal probability plot are not closer to the straight line.
• In other words, the points are not linear.
• As a result, the one-way ANOVA test is invalid for the supplied data set.