Abstract
Given the rise in health-related issues associated with changing food consumption patterns, we examine the role of not only demographics on food-away-from-home expenditures by meal occasion, but also transactional specific variables (e.g., promotion, ordering, and both facility and meal characteristics) and past away-from-home expenditures. Our results suggest that transactional variables that reflect both facility and meal characteristics as well as ordering, promotion, and party composition variables significantly influence food-away-from-home expenditures by meal occasion. Furthermore, both the timing and amount of previous away-from-home expenditures play a major role in expenditures as well.
Notes
2. 1 Exact variable descriptions and averages for the transaction specific variables can be found in the online appendix located http://public.homepages.uconn.edu/˜bec12003/.
3. 2 Using , 56.9% of the observations were FAFH transactions with the other 43.1% representing consumers that had no FAFH transaction the previous day. Accordingly, using the percentages in the amount of censoring ranged from 79.6% for lunch to 99.1% for brunch.
4. 3 As noted throughout the literature having a large degree of censored observations may cause maximum likelihood estimators to fail to converge. Given the high degree of censoring within our data, each maximum likelihood estimator failed to converge.
5. 4 Since the data only indicates whether a promotion was used and not whether a promotion was available but not used, interpretation of the promotion variables should be used with caution. Results detail the effect of a promotion if used, not the effect of promotion in general. For instance, a promotion may increase the number of consumers frequenting a business, but our estimates will not capture this effect, we are only capturing the effect of using a promotion during a transaction.