Location: Children's Nutrition Research CenterTitle: Children's accuracy of portion size estimation using digital food images: effects of interface design and size of image on computer screen) Author
|Thompson, Deborah - Debbe|
Submitted to: Public Health Nutrition
Publication Type: Peer reviewed journal
Publication Acceptance Date: 6/2/2010
Publication Date: 3/1/2011
Citation: Baranowski, T., Baranowski, J.C., Watson, K.B., Martin, S., Beltran, A., Islam, N., Dadabhoy, H., Adame, S., Cullen, K., Thompson, D., Buday, R., Subar, A. 2011. Children's accuracy of portion size estimation using digital food images: effects of interface design and size of image on computer screen. Public Health Nutrition. 14(3):418-425. Interpretive Summary: Estimation of portion size consumed has been a major source of error in 24 hour dietary recalls, especially among children. This study tested the use of food images on computer screens for portion size estimation. We especially tested the number and size of the images on the screen (our images on whole screen versus eight/nine images per screen) and the presence of food size cues (checkered table cloth and cutlery, versus not). This was a perception task in that the food item was placed right next to the child, thereby not requiring any memory of the food. 60.3% of the foods were correctly classified. There were no differences in accuracy, the number and size of the images on the screen, nor by the presence of food size cues. Children were able to more quickly estimate portion size when all sizes were on the screen at the same time. Future computer based dietary assessment programs do not need to use food size cues, but should place multiple images on the screen at the same time to abbreviate task completion.
Technical Abstract: The objective was to test the effect of image size and presence of size cues on the accuracy of portion size estimation by children. Children were randomly assigned to seeing images with or without food size cues (utensils and checked tablecloth) and were presented with sixteen food models (foods commonly eaten by children) in varying portion sizes, one at a time. They estimated each food model's portion size by selecting a digital food image. The same food images were presented in two ways: (i) as small, graduated portion size images all on one screen or (ii) by scrolling across large, graduated portion size images, one per sequential screen. The setting was laboratory-based with computer and food models. Volunteers used were multi-ethnic, sample of 120 children, equally distributed by gender and ages (8 to 13 years) in 2008-2009. Results showed an average percentage of correctly classified foods was 60.3%. There were no differences in accuracy by any design factor or demographic characteristic. Multiple small pictures on the screen at once took half the time to estimate portion size compared with scrolling through large pictures. Larger pictures had more overestimation of size. In conclusion, multiple images of successively larger portion sizes of a food on one computer screen facilitated quicker portion size responses with no decrease in accuracy. This is the method of choice for portion size estimation on a computer.