Using paradata to investigate food reporting patterns in AMPM

Authors: Steinfeldt, Lois; Anand, Jaswinder; Murayi, Theophile

Submitted to: International Blaise Users Conference
Publication Type: Proceedings
Publication Acceptance Date: 2/20/2012
Publication Date: 4/23/2012
Citation: Steinfeldt, L.C., Anand, J., Murayi, T. 2012. Using paradata to investigate food reporting patterns in AMPM. 14th International Blaise Users Conference Proceedings. p. 61-68. Available: http://www.blaiseusers.org/2012/papers/02d.pdf

Interpretive Summary: The USDA Automated Multiple Pass Method (AMPM) collects approximately 10,000 24-hour dietary recalls for the What We Eat In America, National Health and Nutrition Examination Survey each year. The 5-step multiple pass approach used in AMPM is the result of more than 10 years of research by USDA to improve the 24-hour dietary recall methodology. AMPM’s automated collection now provides opportunities for research into the interview process and how individuals report the foods they eat. AMPM keeps track of the order in which foods are reported in the interview and the step in which each food was reported. This and other interview process information collected by AMPM on a large number of individuals in a nationally representative dietary survey is a unique resource. It can be used to identify overall patterns in reporting foods for different groups of people. It can also be used to describe and evaluate the use of each of the 5-steps within AMPM and the questions within each step. Identifying how individual components of the interview are used and the patterns people use to report foods can then be used to further improve the collection of 24-hour dietary survey data in AMPM.

Technical Abstract: The USDA Automated Multiple Pass Method (AMPM) Blaise instrument collects 24-hour dietary recalls for the What We Eat In America, National Health and Nutrition Examination Survey. Each year it is used in approximately 10,000 interviews which ask individuals to recall the foods and beverages that were consumed the day before the interview. AMPM uses a research-based, multiple-pass approach which has 5 steps designed to encourage complete and accurate food recall and reduce respondent burden. Details collected for each food or beverage include the time of day the food was eaten and what the respondent called the eating occasion (breakfast, lunch, dinner, snack, etc). The interview begins with the Quick List where respondents are asked to report all the foods and beverages consumed. The Quick List is an unstructured, uninterrupted listing which allows foods to be reported in any order. At any point during the interview the respondent can add foods that were not reported at the time of the Quick List. AMPM keeps track of the order in which foods are reported in the interview and the step in which the food was reported. Comparing the order that foods were reported to the order in which they were consumed provides insight into the patterns of how individuals report foods. For example, in the day 1 interview about 66% of individuals began the Quick List with their first eating occasion for that day. However, only about 17% reported all their eating occasions on the Quick List in the order they were consumed. In the day 2 interview, with respondents now familiar with the AMPM, these percentages increased to 83% and 35% respectively. Paradata collected by the AMPM can be used for detailed investigations of food reporting patterns in a nationwide food consumption survey.