Effects of training on portion size estimation.

Abstract

The effectiveness of training individuals in portion size estimation skills in order to increase their accuracy of food quantity estimation was investigated. The retention of gain in estimation accuracy over six months and the effects of the completion of diet records on the retention of estimation accuracy were also investigated. A multi-technique training program combining the measurement and estimation of real foods and food models by standard household measurements and dimensions was used. Three foodforms were represented: solids, liquids, and amorphous foods. Two subject samples, composed of Caucasian females, 28-42 years old, were recruited. One sample (n = 25) was pretested on portion size estimation accuracy, trained, and posttested. They subsequently completed four randomly assigned, nonconsecutive days of diet records and were posttested again six months after their first posttest. The other sample (n = 66) was randomly assigned to one of four groups using the Solomon four-group design: (1) control, pre- and posttested, (2) control, posttested only, (3) trained, pre- and posttested, and (4) trained, posttested only. One control group (Group 2, n = 14) and one trained group (Group 4, n = 14) were administered a second posttest six months after their first posttest for comparison with the other subject sample. Groups 2 and 4 did not complete diet records during the six month period between posttests. Using 14 separate 2 x 2 ANOVAs, results showed that training significantly increased portion size estimation accuracy calculated by difference ("D" method, p =.007) and by absolute value ("Ab" method, p =.003). This training effect for the total test was attributed specifically to the increase in estimation accuracy for solids estimated in cups ("D" method, p =.000; "Ab" method, p =.008) and for amorphous foods ("D" method, p =.000; "Ab" method, p =.000). Comparisons between first and second posttest scores for Groups 2, 4, and the group that completed diet records were analyzed using paired t-tests and mixed model ANOVAs. Post hoc tests included Tukey's HSD test and orthogonal contrasts (p≤.05). Both trained groups retained their gain in estimation accuracy over six months for all but one variable, and the completion of diet records improved estimation of solid foods by dimensions.