This study sought to examine whether specific food and activity choices are associated with weight-control activities and with weight change. The findings of the present study are generally consistent with those of similar studies in other populations regarding weight-control efforts, weight and behaviors. Leaner women report eating diets that are lower in foods of high energy density and higher in foods with low energy density, and also report more physical activity . Cross-sectional correlates of reported efforts at weight-control and behaviors are similar to those for the same behaviors and body weight and longitudinal findings for behaviors and weight-control efforts are also consistent with these patterns [12, 13]. Increases in consumption of low energy dense foods and of physical activity are reported in those trying to control their weight compared to those who are not, while decreases are reported in consumption of foods with high energy density and to a lesser degree in sedentary behaviors. In addition, similar to the report of French et al. , there seemed to be no noticeable qualitative difference between the behavioral correlates of effort to lose weight as efforts to avoid weight gain, although effect sizes are generally larger for efforts to lose weight than they are for efforts to avoid weight gain, perhaps indicating greater intensity of effort in the weight losers.
One difference between the present findings and previous work in this area, however, was that we found no relationships between change in weight over time and reported efforts to lose weight or avoid gaining weight. Our sample was smaller than that of some previous studies, so that lack of power is a plausible contributing explanation . Both the weight gain prevention group and the no weight-control effort group were also relatively lean with mean BMI at baseline within the normal range. However, our sample was also purposefully selected to over-represent lower SES women by drawing a greater sample from socioeconomically disadvantaged neighborhoods, and relationships seen previously in other populations have been more ambiguous in lower SES groups .
Considering the overall question of why weight-control effort in the general population are not more successful, two logical possibilities are that people know what to do but implement this knowledge with insufficient frequency or intensity and that their knowledge is not complete enough or specific enough to avoid errors in behavioral choices that undermine their efforts. Conceptually these factors might be particularly important in low SES populations . Although we have little direct evidence to bring to bear on intensity of efforts, in the present results we note five observations that could be interpreted as supporting an incomplete knowledge hypothesis. There were two food items in particular that women trying to lose weight reported eating less of that are not especially high in energy density, namely regular potatoes and bread. While avoiding these foods has been recommended in low carbohydrate diets, that at various points in time have been quite popular, making concerted efforts to reduce their intake may not be the best focal point from the perspective of energy balance. Similarly, we note two food items, nuts and cheese, whose increase in those trying to lose or maintain weight could be self-defeating, since nuts and cheese are high in energy density. The fifth choice that might be suspect was increased frequency of egg consumption in those trying to control their weight. Increased egg consumption is usually not recommended in public health messages about weight control, although it is noted that some recent empirical data has supported the idea that it might be helpful . We are not aware of similar item specific data being reported in other studies, but suggest that further research on potentially counter-productive food choices among dieters is merited and might be informative for formulating weight-related nutrition messages. It should be noted, however, that the inference that the patterns of food selections seen in this population were mediated by knowledge and beliefs is not directly verifiable here because we did not have a measure of nutrition knowledge or beliefs. Food selections could have alternatively been the result of a number of factors unrelated to knowledge (e.g., lack of access or budgetary constraints). More direct confirmation would be necessary before making recommendations for public health messaging.
A second finding of the present study that merits discussion is the asymmetry of findings with respect to physical activity and sedentary behavior . Our data show consistent positive relationships between total self-reported activity and its leisure-time and transport components and both body weight cross-sectionally and weight-control activity both cross-sectionally and longitudinally, but only very weak associates with sedentary behavior, e.g., sitting and screen time. At face value, the finding suggests that active behavior is more important than sedentary behavior for weight-control, however, it also points to possible methodological problems in measurement, e.g., frequency and duration of active behaviors may be better recalled than sedentary behaviors because they are more discrete. More attention to the methods for measuring activity components is recommended.
This study had a number of strengths. These include a unique population of women residing in neighborhoods of relative deprivation, a behavioral survey with broad assessment of obesity related behaviors in the domains of both energy intake and energy expenditure, and a lengthy follow-up period. Offsetting weaknesses include a modest response rate to the survey, which probably makes the sample less representative of the entire population of the areas from which they were drawn than is desirable, complete reliance on self-report measures, some of which were not validated, and measurement tools that were not fashioned specifically on the hypothesis being investigated (e.g., self-reports of food intake over the last three months may not be reflective of diet intake over the three years between observations). It is also noted that the large number of statistical test examined increases the likelihood of type two error. It is believed in sum, however, that the investigation was unique and that the findings advance understanding of what it means behaviorally when people say that they are trying to lose or maintain weight and how those activities might impact the time course of weight over a period of years.
The methodological limitations of this research limit strong recommendations for practitioners or public health messaging. However, should it prove to be the case that failure of self-initiated weight loss and weight gain prevention efforts is related to inadequate knowledge about the energy content of foods, modified public health messaging related to food choices may be in order.