You run the following regression of the number of drowning d

You run the following regression of the number of drowning deaths in the US on a given day on the ice cream sales on that day.

Drownt=beta zero +beta one*icecreamt+Et

a.) Do you expect the estimate beta one to be positive, negative or zero? Explain why.

b.) Do you think the estimate beta one reflects the causal effect of ice cream sales on drowning? Explain why.

c.) What do you think would happen to the coefficient beta one if you included an indicator for the summer months (June-August) in the regression? Explain why.

d.) If you could include just one additional variable in the regression, what would it be? (Note that the data is a time-series of the U.S., so the unit of observation is the day and the thing being observed is the U.S. as a whole.) What would you expect to happen to beta one when you include the variable? Explain your reasoning.

Solution