ABSTRACT
An anthropometrist wishes to establish whether a certain method of measuring adipose tissue is sufficiently reliable to be used in future research work. The anthropometrist recruits 20 subjects and obtains two repeated measurements under the same conditions. The researcher is not an expert in statistics and, therefore, refers to the literature for what are perceived to be the most appropriate statistical analyses and to see how such analyses are interpreted. Consequently, the researcher employs (i) a paired t-test, to check that there has been no change in the sample mean measurement over the two weeks, (ii) the intraclass correlation coefficient (ICC) to check that the degree of relationship between test and retest is above 0.8, and (iii) the technical error of measurement (TEM) to check that the “absolute variability” between test and retest is below 5%.
