Your reflection response is to be written in PROSE, but NO more than 500 words single spaced.

You may NOT include any direct quotes, or you will get NO credit.

Include a citation referencing the Carpenter and Dobbin paper using the proper format (economists often use Chicago stylein the text include the authors name and publication year such as Abrams, Clarke, and Mehta (2021) and Chicago for the reference listed at the end).

As with all of your work, please write CLEARLY and CONCISELY.

In your prose, be sure to address the following questions:

1. What research question do the authors examine in this study?
2. What are the three limitations with prior work the authors list?
3. What research approach do the authors use to examine this question?
4. What data do the authors use to address these questions?
5. What age adults do the authors focus on in their analysis?
6. Which outcomes from the NHIS do the authors examine?
7. How do the authors measure heavy drinking?
8. What problems do the authors identify with the NHIS data?
9. What do the authors mean by studying deaths due to external causes?
10. Consult Figure 1: does drinking participation behavior change much at age 21?
    

11. Which regression specification is preferred? What is different about the preferred regression specification than the most parsimonious specification in column (1)? (Authors talk about this when discussing Table 1)
12. Consult Figure 2. Does drinking intensity change much at age 21?
    
13. What explanation do the authors give for the findings in Table 2?
14. Explain in words what assumption the authors are testing with Table 3, and what they find.
    

15. Explain in words what the authors observe in Figure 3.

16. Explain in words why the authors include deaths due to internal causes in Figure 3 and Table 4.
17. Explain in words what the authors show in Figure 4

18. In Table 5, the preferred specification is in column (2). There are no asterisks to denote statistical significance, so you will have to use the rule of thumb of divide the coefficient estimate by the standard error and if the ratio is >=2, can conclude relationship is statistically significant. For which outcomes are there statistically significant relationships?
    

19. Which commonly held beliefs about the relationship between drinking and mortality do the authors debunk with these results?
    

20. How many more deaths do the authors estimate would occur each year if the drinking age were lowered to 20?