Thesis - Open Access
Doctor of Philosophy (PhD)
Mathematics and Statistics
Profile by Sanford is a membership based weight loss program that helps its members make lifestyle changes with diet, exercise, and one-on-one interactions with a weight loss coach. Discovery of characteristics and behaviors influencing weight loss will benefit current and future members of Profile. This research utilizes massive data from Profile by Sanford to analyze member behavior. Fourteen data sets are evaluated, some containing millions of observations. All data is combined into one comprehensive table of 33,487 members. Members of Profile by Sanford are 77% female and two-thirds of all members start the program classified as obese. Attending meetings with a weight loss coach decreases rapidly over time for Profile members but a higher frequency of meetings is found to have a positive association with weight loss. Increasing a member’s coach meeting attendance to one more meeting a month results in 2.5 percentage points more weight loss for Profile members who weigh themselves consistently each month for the first 12 months in the program. The same group of Profile members experience 2.3 percentage points less weight loss if taking antidepressants after controlling for sex and starting BMI. A mixed model generates weight loss predictions. An additional attendance of a coach meeting is associated with 0.13 percentage points more monthly weight loss. With one more weight recording members lose 0.02 percentage points more per month. A unit increase in starting BMI is associated with an increase of 0.03 percentage points more weight loss. By month 6 more than half of members have dropped out of Profile and 80% have dropped out by month 12. The probability of dropping out of the program is produced by a joint model. Higher age, married members, and females are associated with a lower risk of dropping out of Profile. The joint model suggests that the risk of dropping out of the weight loss program increases by 140% with each percentage point increase in monthly weight gain. Application of the statistical models can allow coaches to interact proactively with members based on their likelihood of dropping out of the program.
Includes bibliographical references (pages 98-101)
Number of Pages
South Dakota State University
In Copyright - Educational Use Permitted
Bares, Valerie, "Identifying Predictors of Weight Loss and Drop-Out Using Joint Modeling" (2017). Theses and Dissertations. 1716.