I survived the class, woohoo! The topic was fun, but taking a summer grad class was brutal. I wish I could have taken it during the fall semester like I’d planned so I could have more time to enjoy it, but alas, it’s only offered at the undergrad level in the fall.
I wasn’t able to use this survey for my project because it wasn’t complex enough. Instead my project analyzed the firm size effect on predicted returns of the stock market in the last decade. That wasn’t near as fun as this topic (to me), but it had the complexity needed for the project.
Since I didn’t use the data for class, it’s now just “a random chick on the internet wants to know…” and given the small sample size of 70 observations, the results aren’t conclusive, but it’s still fun anyway.
I used a linear probability model to try to discern how someone would answer the “do you follow a recipe” question based on the person’s age, gender, and if the person is a rule-follower in life and if the person’s parents followed a recipe.
The models below predict the correct outcome 50% of the time, and they’re really good at predicting “failures” (i.e. not this answer), but they aren’t good at predicting “successes” (i.e. yes, this answer). None of the coefficients are statistically significant, and including and excluding variables and slope dummies changed the coefficients but did not produce any statistically significant results, so these models, given the sample set, aren’t good, but it was still fun to play around with.
^Exactly = 0.09 + 0.07Parents - 0.10Life + 0.04Gender + 0.002Age
^Dietary = 0.61 - 0.09Parents - 0.07Life - 0.11Gender - 0.003Age
^Substitution = 0.37 - 0.02Parents + 0.06Life - 0.05Gender - 0.005Age
^Suggestion = -0.06 + 0.04Parents + 0.11Life + 0.12Gender + 0.006Age
= the predicted probability that a person will follow a recipe exactly
= the predicted probability that a person will change a recipe for dietary reasons or taste preferences
= the predicted probability that a person will make substitutions with ingredients on hand
= the predicted probability that a person will consider a recipe just a suggestion
= 1 if the person’s parents generally followed a recipe, 0 otherwise
= 1 if the person considers themselves a rule follower in life, 0 otherwise
= 1 if the person is male, 0 if the person is female
= the age of the person (in the survey, these were ranges, so I used the midpoint
So for myself, here are the predicted probabilities:
^Exactly = 0.09 + (0.07*0) – (0.10*1) + (0.04*0) + (0.002*36) = 0.102 => (less than 0.5) => interpreted as 0 (failure) => No, won’t follow exactly
^Dietary = 0.43 => (< 0.5) => 0 => No, won’t change for dietary or taste preferences
^Substitution = 0.25 => (< 0.5) => 0 => No, won’t substitute with what’s on hand
^Suggestion = 0.27 => (< 0.5) => 0 => No, doesn’t see recipes as a suggestion
I picked “Dietary,” which had the highest probability given my inputs, so it was on the right track, but it still predicted it as a “failure” when rounded to a boolean Yes/No value.
Anyway, thank you to everyone who played along for a random chick on the internet’s Question of the Day. I enjoyed playing with the data