Download Excercises For Stata and more Exercises Windows Programming in PDF only on Docsity!
Exercises for Stata
Svend Juul, June 2011
The purpose of these exercises is to learn Stata by doing. Use Svend Juul and Morten Frydenberg's Stata book (ISHR3) to look up commands, etc., during the exercises.^1
From http://www.biostat.au.dk/teaching/senior you can download datasets for exercises to your hard-disk; at the computer-lab download them to D:\StataCourse\your name.^2 For this exercise we will use the smoke.dta dataset. Further instruction will be given at the course.
In the exercise questions I often give a hint about the procedure to be used (e.g. describe). Look up the command syntax in ISHR3.
The smoke.dta dataset includes the following variables:
Variable Meaning Possible values id ID number 1- sex Sex 1 male 2 female age Age in years 18- weight Weight in kg 40- height Height in cm 120- smoker Smoker? 0 no 1 current smoker 2 former smoker cigaret Cigarettes/day 0- cheroot Cigars or cheroots per day 0- pipe Packs of pipe tobacco per week 0-
- In the Command window enter: cd "D:\StataCourse" to make D:\StataCourse the current file path mkdir "your name" to create the folder D:\Statacourse\your name cd "your name" to make D:\StataCourse\your name the current file path use "smoke.dta" to load the smoke.dta dataset into Stata
Look at the Variables window; it displays the names of variables in the smoke.dta dataset. Look in the lower left corner, just below the Variables window; it displays the current file path
(^1) Juul S, Frydenberg M. An introduction to Stata for health researchers, 3rd ed. College Station, TX: Stata Press,
- I refer to the book as ISHR3. (^2) Not "your name", of course, but a folder with your name.
- Open the Data window with the command: browse and take a look. Now close the Data window. Open it again to see the values of age and weight for the first 5 observations by: browse age weight in 1/ There is a button for browse, too, but it does not let you select observations and variables.
- In the Command window now enter: summarize and look at the Results window. summarize displays the number of valid observations, mean, standard deviation, and the minimum and maximum values of all variables. Use the mouse to highlight the table, and print it. (Click the printer button. In the dialog that appears, click the Selection radio button; the rest can be done by intuition).
- Look at the Review window. It displays the two commands issued so far (and possibly also some errors you made). Now, click the last command (summarize) and watch it being pasted to the Command window. In the Variables window click weight and watch it being pasted to the Command window, which now displays: summarize weight Hit the Enter key and see what happens.
- Try the dialog system to obtain the same result. If you know the command name it is easiest to enter the db (dialog box) command: db summarize If you are searching a command, but don't know its name, use the menu system; for summarize it is:^3 Statistics ► Summaries, tables, and tests ► Summary and descriptive statistics ► Summary statistics In the dialog that appears, use the drop-down list in the Variables field and select age. The dialogs generate commands. In the case of summarize age it obviously was easier to write the command than to create it using the dialog, but for complex commands and graphs the dialogs can be an advantage.
(^3) Stata's menu system is not Scandinavian design, and it can be difficult to guess which branch to follow to
locate a command. If you know the command name, use the db command.
- Look at the tables for sex and smoker; they display value labels. To see the codes instead: tab1 smoker , nolabel Use the numlabel command to include the code in the value label: numlabel , add tab1 smoker
Would it be wise to give this command? tab1 id
- Examine graphically with a scatterplot the relation between height and weight (use the menu system: Graphs ► Twoway graph ► scatter). After having studied the graph, look at the Review window; here you see the graph command; it should be: twoway (scatter weight height) Click the command; now it is copied to the Command window. Add: twoway (scatter weight height)(lfit weight height) To restrict the graph to women only, use the if qualifier (ISHR3, section 8.1): twoway (scatter weight height)(lfit weight height) if sex==
- Create a new variable, agegrp, which is a reasonable grouping of age (recode).
Create a new variable, tobacco: tobacco use in grams per day (1 cigarette = 1 g, 1 cigar or cheroot = 2 g, 1 pack of pipe tobacco = 40 g) (generate). Define labels for agegrp and tobacco (label variable, label define, label values). Create a new dataset, smoke1.dta, including the two new variables (save). Since this leads to a new version of the dataset, it is strongly recommended to do it in a do-file, starting with: cd "D:\StataCourse\your name" use smoke.dta , clear and ending with: save smoke1.dta , replace with the calculation commands between the use and the save command. Did you save the do-file creating smoke1.dta? What name did you give it? (My recommendation: gen_smoke1.do; the do-file that generates smoke1.dta):
- From smoke1.dta: see a frequency table for tobacco (tab1). Compare with the frequency tables for cigarettes etc. (from question 7) and decide if the result makes sense. Also, list the first 30 cases to see if calculations gave the result intended (list or browse with the in qualifier). If wrong, redo exercise 9.
- agegrp could have been made with egen, using the cut function (described in the booklet under recode). Try to do that.
- Describe with a crosstable the joint age and sex distribution of the study population (tab2). Would you prefer the original age or the grouped age for this kind of output?
- Create a new variable, bmi (Body Mass Index) = weight/height^2 (weight in kg, height in m). See a summarize table for bmi and check that it looks reasonable. See the average bmi by sex and age groups (tabstat). Test if the bmi distribution is different for men and women (ttest).
- Group bmi in 3 groups (recode). See a crosstable between the grouped BMI and age variables (tab2). Add column percentages, a chi-squared and a Fisher's exact test to the crosstable (ISHR3, section 11.3)
- The new variables bmi and bmigrp are useful and should be included in the dataset. One option is to include the commands in gen_smoke1.do and run it again, another option is to include them in a gen_smoke2.do, which generates smoke2.dta.
- Make a list of all whose BMI is larger than 30, showing the variables id weight height bmi (list with the if qualifier). Look at the list. Any surprises? (Read ISHR3, section 5.4 on missing values).
Using log files
In the computer lab, an output log is created automatically; it contains a copy of everything you saw in the Results window. It is a simple text file, stata.log, and it is located in D:\StataCourse\tmp.^5
The following unofficial commands are helpful for inspecting the log: nlog Opens the log file in NoteTab Light, a rather simple text editor. Here you can read it, print it, or print selected part of it. It works much like a word processor, but without the formatting options.^6 newlog deletes the log file and opens a new log.
Besides the general output log described above, you can create a specific output log, typically to preserve the results from an analysis. Imagine that we perform a regression analysis and want to preserve the output from this analysis. This is important, so we do it with a do-file. Note the capture log close, the log using ..., and the log close commands.
- regres1_bwt.do cd C:\docs\proj capture log close // Closes log in case one is open log using regres1_bwt.log , replace use lbw1.dta , clear regress bwt i.race i.smoke lwt testparm i.race predict rbwt if e(sample) , residual histogram rbwt rvfplot , yline(0) log close
(^5) This is not default behavior by Stata, but something we made for the course. If you want your own computer
to generate an output log automatically, study section 1.6 in ISHR3. (^6) Find a short description of NoteTab Light at http://www.folkesundhed.au.dk/uddannelse/software. Note
especially that if you want to print a selected part of the text, press the Shift key while clicking the Print icon, and select Selection.
