The year is 1979. A consortium of foreign car manufacturers starts running an ad

The year is 1979. A consortium of foreign car manufacturers starts running an advertising campaign claiming that a recent study showed that foreign cars have significantly better fuel efficiency than domestic vehicles. The advertisement goes on to explain that this is due to the superior technology
of foreign manufacturers. Detroit becomes concerned about this claim, particularly in relation to the high gas prices at the time. The big US auto makers ask your consulting company to look into this advertisement, to assess its accuracy, and suggest a possible counter-campaign.
Note: please copy/paste the STATA text output of any relevant regressions into your paper as anappendix. Normally, you wouldn’t do this in a real report, but it helps me understand what went wrong if your answers are off.
Using the auto.dta data that comes with STATA, do the following and write up your work into a cohesive report for the customer (this will probably come out to 4-6 pages with all the plots that you will want to include):
1. Do a quick check of the original claim – plot MPG versus domestic/foreign for the provided data.
Do the foreign cars look like they are more efficient on average? Do a quick regression of MPG
versus domestic/foreign. Remember the STATA command for making indicator variables:
tabulate [variable], generate([output variable name]). What is the average difference in MPG
between foreign and domestic cars? Is this figure statistically significant?
2. The US auto makers point out that US cars are generally bigger, so you would expect them to have lower MPG. [note: you can quickly check the mean of a variable by using the command “mean([variable])”. You can check the mean of just foreign/domestic cars by using the
command “mean([variable]) if [indicator variable name] == [0 or 1]”. An example: mean(weight)
if fd1 == 1] Look into this effect by plotting mpg versus weight (showing the foreign and
domestic points as different colors), then doing a regression of MPG versus weight and
domestic/foreign. Is there a noticeable difference between domestic/foreign on the plot? What does the regression say – is your foreign/domestic indicator variable significant? What does it mean that this variable is significant/not significant?
3. You also consider that engine size might have something to do with MPG and that domestic cars probably have larger engines. Check out this hypothesis by doing another regression of MPG
versus engine displacement and foreign/domestic. Which variables are significant in this regression? Interpret/explain what the regression results mean.
4. Do a regression of MPG against weight, displacement, and foreign/domestic (ie, using weight, displacement and the foreign/domestic indicator variable to predict MPG). Compare the R-squared, adjusted R-squared, and coefficient confidence intervals from this regression to the
two prior regressions (where MPG was a function of weight and displacement individually).
Which of these three models is the best?
5. Using any model (including the three above) that you chose as the best option, interpret the
results for the client (the US car makers). What are your findings from this work? What does this mean for them? How might they use your findings to create their own advertising campaignto counter the campaign by foreign manufacturers?
Note: if you think other regressions, plots, or comparisons would help your case here, feel free to pursuethem.

Leave a Reply

Your email address will not be published. Required fields are marked *