Prior to examining real, as opposed to ideal failure data, it is instructive to examine the important role that visual inspection of failure probability plots can play in choosing the statistical life model that offers the best characterization. For practical reasons having to do with cost and schedule, the only failure data available for a manufacturer may have resulted from life testing small sample size populations, for example, SS = 10–20. Despite the recognition that such data may prove inadequate (Section 8.11) for making trustworthy reliability estimates, an analysis may be required nonetheless, because a quantitative analysis is better than a guess. The three prior chapters showed that when the data sets are large (SS = 50) it is easy to distinguish among ideal Weibull, lognormal, and normal times to failure. It is also possible, for example, to distinguish between ideal Weibull and lognormal times to failure when the sample sizes are small.