Lab 3- Q1d

[malmufre]

Hello,

I am having trouble interpreting Q1d since I am getting the following:

accidents<- data$accidents
taxes<-data$taxes

reg.acc <- lm ( accidents ~ taxes )
stargazer( reg.acc, 
           type = "html", 
           dep.var.labels = (" Car Accidents"),
           column.labels = c("Pooled OLS model"),
           covariate.labels = c("Intercept", "Taxes"),
           omit.stat = "all", 
           digits = 2, intercept.bottom = FALSE )

Based on question 1d: Interpret the coefficient of “taxes”. What is its effect on car accidents? I am interpreting it as the following, but I am unsure because 760 cannot be expressed in percentage terms as it is in the lecture. The number of accidents increases by 760.12, for each dollar increase in beer taxes. Am I thinking of this correctly?

Thank you!

[lecy]

I would revisit the data dictionary and the measure of taxes.

Beer taxes as percentage of cost, from 0 to 1

What does a one-unit change represent in this context?

Check the descriptives (min, max, mean, sd) for taxes. It might help with interpretation.

A standard deviation of the inter-quantile range are both common ways of representing a significant change with regards to an intervention (raise in taxes). Compare that to the default one-unit used for the coefficient estimate.

Is 760 a reasonable amount (going from a 0% tax to 100% tax), or would it be larger / smaller than what we would expect from an actual moderate to large intervention (significant tax increase)?

Note that the slope coefficient is an abstract representation of your input-output machine (if you change inputs you get a different level of outputs). But the default change of one unit is rarely a meaningful change that would correspond with an actual policy intervention. Make sure you understand what the coefficient represents, but also try to think about what a reasonable unit of change might look like in a given domain.