Humans lie. Some lies are big and some are small, but all humans lie and they do it regularly. Of course not all lies at designed to deliberately mislead, because often the lie is a form of denial. Further, lies are often told in order to please someone else. In other words people often say what they think others want to hear. The psychology of lying is interesting and incredibly complex, and this has certainly lead to problems in scientific research. Asking subjects a series of questions, or asking them to record data is often problematic because the validity of the data relies on the subject telling the truth and being accurate with their observations. And as has been shown many times in studies investigating subject reliability, this is often not the case. In fact many questionnaires designed to collect study data deliberately include questions to assess the validity of the data being reported by the subject for this reason. Self denial and the wish to please others are two reasons that food diaries from obese subjects should always be questioned.
Because of the inherent problem with subject reliability, one group of researchers assessed the accuracy and usefulness of data reported in three studies from obese patients1. In an initial study the authors evaluated the ability of obese subjects to accurately record food intakes in the absence of training by assessing their ability to estimate the energy content and serving size of various foods in a laboratory setting. Such self monitoring of food intakes, often in a continuous longitudinal record, has been questioned when compared to methods such as the 24-hour recall. The results of the investigation showed that quantities of foods were generally overestimated, and this included as 6 % overestimate in the case of a cola drink, and a 260 % overestimate for crisps. There were also poor estimates of calorie intakes which ranged from a 5 % underestimate for cottage cheese to a 119 % overestimate for green beans. The mean error in calculating food qualities was 64 % and the mean error in calculating energy intake was 53 %.
Obese subjects therefore make large errors in estimating the calorie content and serving sizes of foods in a laboratory setting. When the authors looked at dropout rates in studies they found that subjects who dropped out of weight loss studies underestimated calories in 58 % of cases but subjects who did not drop out underestimated calories in only 23 % of cases. In a second experiment, the authors assessed the accuracy and completeness of food diaries from obese subjects that were performed in a real world situation away from the laboratory. However, of the diaries assessed by the researchers, 53 % had errors in estimates of quantities and energy content, 23 % were without any quantity data, and 24 % of entries had no data on either the quantity of food or an energy estimate. The authors found that subjects who tended to make errors in the laboratory also made errors in the real world situation. Self monitoring in obese subjects may therefore not be accurate when using untrained self-reported estimates of food intakes.
RdB
