Snack product consumer surveys: large versus small samples

This study showed the comparison results of randomly selected small sample sizes and a larger consumer panel (522 respondents) from questionnaires on snacking behaviour (nominal data), sensory preference tests (interval data) and also the ranking of the packaging (ordinal data) of 7 representative Thai commercial snack products. The results from each of the smaller random samples were plotted against the results from the large consumer survey, and their R-squares were calculated. The histograms of the differences between the small sample sizes and the larger panel were created and the standard deviations of each histogram were considered. If the R-squares of the small group versus the large consumer group or their standard deviations of differences were plotted against the sample size of the small groups, the R-squares increased whilst the standard deviations of differences decreased when the small sample sizes increased. The R-squares and standard deviations did change noticeably between the sample sizes of 8-24 for nominal data and 8-32 for interval data. The degree of changes gradually decreased until a sample size of 64, after which it flattened out. For ordinal data, the changes of R-squares and standard deviations were rapid at the sample sizes of 8-40 but they reached a plateau after 40.