Page:Sm all cc.pdf/40

 * undetected, involuntary change of part of the measurement procedure during the measurement series;
 * undetected change in standards;
 * temporal change in an unidentified relevant variable, i.e., a source of ‘noise’.

Examples
We can gain insight into the statistical techniques described above by considering their application to a few datasets of different types. Our examples come from a variety of original sources, but I got almost all of them from the World Almanac [Hoffman, 1990]. The exceptions are the examples of random normal numbers and of the hare and tortoise. I have deliberately chosen familiar examples rather than artificial data or real scientific data, because the explanation for observed statistical behavior is easier to comprehend with familiar examples. The examples are:
 * reexamination of the random normal numbers of Figures 1 and 2;
 * race between the hare and the tortoise;
 * percentage of high school students that graduate, by state;
 * state population (1990 census);
 * state taxes, per capita, by state;

Table 6 summarizes the statistical results for these examples, as well as some examples introduced in the next chapter.