73G-1 |
A comparison of computer and paper ballots |
M. SWANEY-STUEVE and H. Heymann. Food Science and Engineering Unit, University of Missouri, 122 Eckles Hall, Columbia, MO 65211
Many researchers have compared results from experiments using paper ballots and computer ballots, yet few have studied the interchangeability of the two data collection methods. If computers fail during or between sessions, one would like to use paper ballots for that session with some confidence that scoring will not be affected. The objective of this study was to compare results of panelists alternately using both computer and paper ballots to determine if ballot type has a significant influence on the results. Twelve panelists evaluated twelve marketed peanut butters using a Williams Latin Square Design. The samples were evaluated in triplicate using paper and computer ballots, identical in sequencing of the thirty-two attributes and in line scale length. The design was balanced such that six panelists evaluated the peanut butter samples using a paper ballot, while the remaining six panelists used the computer ballot. The panelists evaluated the products using the same ballot for twelve consecutive sessions. The ballot type was reversed for sessions thirteen through twenty-four. Mixed model multivariate analyses of variance indicated no significant differences (p<0.05) between paper and computer ballots. Univariate analysis of variance also indicated no significant interaction between main effects of sample and ballot. However, from this analysis, the fourteen peanut butter samples did significantly differ across thirty attributes. One can conclude from this experiment that if a situation occurs forcing panelists to use an alternative ballot, the results should not be significantly affected.
Session 73G, Sensory Evaluation
|