Three methodological and statistical points about benchmark data

The coefficient of variation is the ratio of the standard deviation of a set of numbers to the average value of the numbers. The lower the coefficient the tighter the data set. For example, using more than 350 law departments that have participated in the General Counsel Metrics survey of Global Benchmarks, the standard deviation of the number of lawyers was 70.4, the average was 32.8 and therefore the coefficient of variation was 2.15. In comparison, for paralegals the standard deviation was 14.1, the average 7.96, and the coefficient of variation 1.78. Finally, for all other legal staff those figures were 38.6, 16.3, and a coefficient of 2.2.4.

Cross-tabulation analysis can tell whether there are significant differences between respondents and non-respondents on industry, size, location and other characteristics.

Third, analysts can scrutinize respondents to a survey for non-response bias. A Kolmogorov-Smirnove test indicates whether there are significant differences between respondents and non-respondents on such factors as geographic location, industry distribution, age, size or performance.” But you must know your survey group’s demographics. If a trade group invites its members to take part in a study, this statistical tool for benchmarking comes in handy.