discrete data can usually be preceded by the phrase “number of …”—for

Comparing the two outcomes of binary variables for two independent groups happens to be one of the most frequently computed procedures in applied statistics. Surprisingly, there is little agreement on the best statistical test for this situation. For large sample sizes, the chi-square test is typically recommended. For small sample sizes, the

Errors provide excellent diagnostic information on why users are failing tasks and, where possible, are mapped to UI problems. Errors can also be analyzed as binary measures: the user either encountered an error (1 = yes) or did not (0 = no).

If there is more variation in a population, each sample taken will fluctuate more and therefore create a wider confidence interval. The variability of the population is estimated using the standard deviation from the sample.

Because the total area must add up to 100% under the curve, we can express a z-score of 1.28 as being higher than 90% of values or less than 10% of values

seek to minimize systematic bias in your sample but remember that representativeness is more important than randomness.

For practical tips on collecting metrics in usability tests, see A Practical Guide to Measuring Usability (Sauro, 2010) and Measuring the User Experience (Tullis and Albert, 2008).

interactive lessons with many visualizations and examples on the www.measuringusability.com

Just because the goal is to find and fix as many problems as you can does not mean there is no opportunity for quantification. You can quantify the problems in terms of frequency and severity, track which users encountered which problems, measure how long it took them to complete tasks, and determine whether they completed the tasks successfully.