Common Errors in Statistics (and How to Avoid Them), 4th Edition

DESCRIPTIVE STATISTICS

In this section, we consider how to most effectively summarize your data whether they comprise a sample or the entire population.

Binomial Trials

The most effective way of summarizing the results of a series of binomial trials is by recording the number of trials and the number of successes. For example, the number of coin flips and the number of heads, the number of patients treated and the number who got better, and so forth.

When trials can have three to five possible outcomes, the results are best presented in tabular form (as in Table 8.1) or in the form of a bar chart, whether the outcomes are ordered (no effect, small effect, large effect) or unordered (win, lose, tie). Both forms also provide for side-by-side comparisons of several sets of trials.

TABLE 8.1. RBI’s per game

For the reasons discussed in the next chapter, we do not recommend the use of pie charts.

Categorical Data

When data fall into categories such as male versus female, black versus Hispanic versus oriental versus white, or in favor versus against versus undecided, we may display the results for a single categorical variable in the form of a bar chart. If there are multiple variables to be considered, the best way to display the results is in the form of a contingency table, as shown in Tables 8.2 and 8.3. Note that Table 8.2 is a highly effective way of summarizing the data from nine ...