160 likes | 536 Views
Stata Review Session. Economics 1018 Abby Williamson and Hongyi Li November 17, 2006. Agenda. Administrative Issues Empirical exercise - clarifications Log and .do files in Stata Introducing Variables Reviewing Stata Commands An Example Questions. Administrative Issues.
E N D
Stata Review Session Economics 1018 Abby Williamson and Hongyi Li November 17, 2006
Agenda • Administrative Issues • Empirical exercise - clarifications • Log and .do files in Stata • Introducing Variables • Reviewing Stata Commands • An Example • Questions
Administrative Issues • Office Hours • Abby – Monday, November 27, 10-12 Tuesday, November 28, 4-6 • Hongyi – Wednesday, November 29, 6-8 • Problem Set Questions • E-mail BOTH TFs with questions. We will reply to the entire class. Keep track of these e-mails as common questions will arise. • Addresses: Abby.Williamson@gmail.com, hongyi@gmail.com
Empirical Exercise – some clarifications • Question 2b: you may organize your graphs in whichever way you see fit. • For example: with happiness, generate 1 bar graph for each wave, with 1 bar per country representing average level of happiness • Question 3: Don’t have to include too many controls (which would reduce significance). • Also, don’t use educational levels (which are unevenly measured across countries) • Question 4: Don’t just use US data. You have to show that trust has declined, relative to other countries, in younger cohorts in the US.
Keeping Track of Stata Work • Opening data • Open Stata, use “open” button on menu (folder) to open relevant dataset • Dataset too big? Set memory higher: • set mem 500m (or bigger …) • Open a log • Ex: log using c:/WilliamsonPS • Remember to use: capture log close • Start a .do file • Use the envelope button to open a .do file, save it under a name you’ll remember. Write and run your program from there so that you can re-run analyses without rewriting everything.
Example: A Few Variables • E037.- Government responsibility • People should take more responsibility to provide for themselves vs. The government should take more responsibility to ensure that everyone is provided for • 1 'People should take more responsibility' • 2 '2' • 3 '3' • 4 '4' • 5 '5' • 6 '6' • 7 '7' • 8 '8' • 9 '9' • 10 'The government should take more responsibility'
Example: A Few Variables • A025.- Respect and love for parents With which of these two statements do you tend to agree? (CODE ONE ANSWER ONLY) • A. Regardless of the qualities and faults of one's parents, one must always love and respect them. • B. One does not have the duty to respect and love parents who have not earned it by their behavior and attitudes. • 1 'Always' • 2 'Earned' • 3 'Neither'
Reviewing Stata Commands • Rename • Ex. rename e037 big_govt (Renames variable e037 “big_govt”) • Ex. Rename a025 obedience(Renames variable a025 “obedience”) • Recode • Ex. recode obedience (1=1) (0=2) (Recodes the variable so that always=1, earned=0)
Reviewing Stata Commands • Tabulate • Ex. tab obedience(Tells what proportion of respondents falls into each category.) • Ex. tab obedience if sex==1(Tabulates obedience for female respondents.) • Summarize • Ex. sum big_govt (Gives basic summary statistics (mean, range, etc.) for big_govt.)
Reviewing Stata Commands • Sort and By • Ex. sort country by country: tab obedience (Sorts by country and then tabulates obedience for each country.) • Collapse • Ex. collapse (mean) age height, by (country) (Makes a new dataset with one observation per country, and two variables: the mean age and mean height for the people of that country)
Reviewing Stata Commands • Graph • Ex. graph twoway scatter age height • (Generates a scatter plot with age and height as the axes) • Ex. graph bar (mean) age, over (country) • (Generates a bar graph where each bar corresponds to a country and the height of each bar is the average age for that country)
Reviewing Stata Commands • Generate • Ex. gen highhealth=0 replace highhealth=1 if health==1 (Generates a dummy in which all observations equal 0 unless respondent reports excellent health, in which case highhealth is replaced with 1.) • Ex. gen agesq=age*age(Creates a variable that is the square of the respondent’s age.) • Ex. gen femeduc=sex*educ(Creates an interaction variable that measures the differential impact of education on women.)
Reviewing Stata Commands • Drop • Ex. drop a001 a002 a003 a004(Drops those four variables, keeping all others.) • Ex. drop if age<18(Drops all children.) • Keep • Ex. keep a001 a002 a003 a004(Keeps those four variables, dropping all others.) • Ex. keep if age<18(Keeps only children.)
Reviewing Stata Commands • Correlate • Ex. corr big_govt obedience (Gives the correlation between health and happiness.) • Regress • Ex. reg big_govt obedience educ age income, r (Regresses happiness on health controlling for education, age, and income. “,r” indicates use of robust standard errors – almost always a good idea.)
Reviewing Stata Commands • Regress • Ex. reg big_govt obedience educ age income, r • (Regresses happiness on health controlling for education, age, and income. “,r” indicates use of robust standard errors – almost always a good idea.) • Ex. xi: reg big_govt obedience educ i.country, r • (Regresses happiness on health controlling for education, age, and income, with country fixed effects.)
Reviewing Stata Commands • IV Regress • Ex: ivreg2 big_govt (obedience=religion) educ age income, r • (Regresses happiness on health controlling for education, age, and income, while instrumenting for happiness using religion.) • ivreg2 is preferred to ivreg, although both run 2SLS. You have to install it – search for it from the help window.