TABLE 2

Exposure and Outcomes Variables Used in the Analysis

Exposure Variables
 Parental stressParental stress was measured via signal contingent EMA surveys by using items adapted from the daily health diary.32 Parents were asked about their current level of stress (ie, how stressed are you feeling right now?), the main source of stress (ie, what is the main source of your stress?; response options, eg, a lot of work to get done at job or school, conflicts or arguments with my spouse or romantic partner), and their perceived ability to cope with stress32 (ie, right now, how certain do you feel that you can handle all the things that you have to do today?) since they woke up or since the last survey. EMA-reported stress before noon (ie, morning stress) was analyzed as a continuous random variable with Likert scale values ranging from 0 to 4 (0, “not at all”; 1, “a little”; 2, “moderately”; 3, “quite a bit”; 4, “extremely”). Multiple morning stress observations on signal contingent EMA surveys were averaged by participant and by observation day if the stress measure occurred before noon to ensure temporal ordering of the predictor variable.
 Parent depressed moodParent depressed mood was measured during signal contingent EMA surveys by using an item adapted from Kessler-6 measure of depressive symptoms.33 Parents were asked about their current level of depressed mood (ie, how sad or depressed are you feeling right now?). EMA-reported depressed mood before noon (ie, morning stress) was also analyzed as a continuous random variable with Likert scale values ranging from 0 to 4 (0, “not at all”; 1, “a little”; 2, “moderately”; 3, “quite a bit”; 4, “extremely”). Temporal ordering was handled for morning depressive symptoms in the same way as the morning stress measure.
Outcome Variables
 Parent feeding practicesParent restriction and pressure-to-eat parent feeding practices were measured during event contingent (ie, meal occasions) EMA surveys by using 2 items modeled after the child feeding questionnaire.34 Parent restriction (ie, did you have to make sure [child’s name] didn’t eat too much food at this meal?) and pressure-to-eat (ie, did you have to encourage [child’s name] to eat more food at this meal?) feeding practices at meal occasions were measured as a dichotomous variable (0, “no”; 1, “yes”). Dinner and snacking event contingent meal occasions occurring at 4 pm or later were included for analysis (ie, breakfast and lunch meal occasions were excluded to avoid reverse causation). Evening feeding practices were averaged at the participant and observation day level and treated as a binomial outcome variable ranging between 0 and 1 in all quantitative analyses. If a participant reported pressuring feeding practices at 2 of 3 meal occasions, the binomial outcome variable at the day level would be equal to 0.667 (ie, 2 meal occasions in which the feeding practice was observed divided by the 3 total meals occurring that day after 4 pm at dinner or snacking occasions).
 Type of foods served at mealsThe type of foods served at meals was assessed during event contingent EMA surveys (ie, which best describes the type of food served at the meal?), adapted from previous survey research questions. Response options included: homemade, preprepared or fast food. Parents could report multiple types of foods served at any one meal occasion.4,35 The proportion of evening meal occasions (at the participant, observation day level) in which the food type was present was operationalized as a binomial outcome variable in the same way as the parent feeding practices variable (ie, range 0–1). Like the example provided above, if the daily proportion of fast food at evening dinner and snacking meal occasions was equal to 0.667, then two-thirds of 1 day’s evening meal occasions had fast food item present. Parents could report multiple types of foods served at any 1 meal occasion. Lunch meals were excluded to ensure temporal ordering of early parental stress and later-day parent feeding practices and type of foods served at meals.