Stata version 12people.umass.edu/biep640w/pdf/stata v 12 lab session 1.pdfStata version 12 Lab...

$: Stata version 12people.umass.edu/biep640w/pdf/stata v 12 lab session 1.pdfStata version 12 Lab Session 1 February 2013 …\stata v 12 lab session 1.doc 2/18/2013 Page 1 of 22 ... Follow$
Stata version 12 Lab Session 1 February 2013

…\stata v 12 lab session 1.doc 2/18/2013 Page 1 of 22

Stata version 12

Lab Session 1 February 2013

1. Preliminary: How to Screen Capture …………………………….. 2. Preliminary: How to Keep a Log of Your Stata Session ……….. 3. Preliminary: How to Save a Stata Graph …………………..……. 4. Enter Data: How to Create a New Data Set in Stata…………….… 5. Enter Data: How to Import an Excel Data Set …………..……….. 6. Enter Data: How to Import a Stata Data Set from the Internet … 7. Describe Your Data – Numerical Descriptions …………………… 8. Describe Your Data – Graphical Descriptions ……………………. 9. One and Two Sample Inference ……………………………………. 10. Simple and Multiple Linear Regression ……………………………

2

4

6

8

10

17

18

19

21

22



1. Preliminary: How to Screen Capture

When to Screen Capture “Screen capture” is useful when you want to capture a “picture” to be pasted elsewhere. The “picture” might be what is on your screen following a “google image” search. Or it might be what is on your screen after obtaining an error message. There are two steps: (1) capturing the picture, followed by (2) pasting the picture elsewhere. Step 1 – Capture the Picture (PC Users)



Step 1 – Capture the Picture (for MAC Users)

Step 2 – Paste the Picture __1. Launch WORD __2. Scroll to the location where the picture is to be placed. Tip – Make sure there is a blank line above this location. __3. Use the TABLE commands to create a table that has 1 row and 1 column __4. Position the cursor inside your table. Tip – Center the cursor so that your picture will be centered. __5. Use INSERT > PICTURE > FROM FILE to insert your picture.

YOUR TURN: __a. Create a word document called lab1.doc __b. Using your browser, go to the welcome page for PubHlth 640 __c. “Capture” the picture of the lighthouse __d. “Paste” into lab1.doc



2. Preliminary: How to Keep a Log of Your Stata Session

When to Keep a Log of Your Stata Session The short answer is: ALWAYS! This is both an archive of your work that can be reproduced later and a record of what you have learned how to do (thus, saving you having to re-learn it later!). It is also useful in report writing. You can import sections of your log, results mostly, into the report that you write. A Stata log can be saved in either of two formats: “smcl” or “log”

.smcl – smcl stands for “stata markup and control language”. This format preserves all the Stata formatting and controls.

.log – This is a plain text format. This format is easily imported into MS Word or Notepad. Tip – Save your log in format “.log”

Step 1: From the main menu at upper left: FILE > LOG > BEGIN



Step 2: Click on drop down menu at right of FILE FORMAT:

Step 3: Choose Stata Log

Step 4: Enter name at “SAVE AS:” Then click SAVE

YOUR TURN: __a. Launch Stata __b. Start a log of your lab session that you name feb20_lab



3. Preliminary: How to Save a Stata Graph

When to Save a Stata Graph Again, the short answer is: ALWAYS! Surely you will want to display your graph somewhere, yes?

Step 1: Create your graph (for now – just follow along by typing the following in the command window) use “http://people.umass.edu/biep640w/datasets/week02.dta”, clear graph twoway (scatter boiling temp, symbol(d)) (lfit boiling temp), title(“Simple Linear Regression”) subtitle(“Y=boiling on X=temp”) caption(“boiling.png”, size(vsmall))



Step 2: With the graph window still active, click on the SAVE icon.

Step 3: From drop down menu for File format: Choose PORTABLE NETWORK GRAPHICS (*.png)

Step 4: Enter your choices at “SAVE AS” and at “WHERE”. Then click on the SAVE icon.

YOUR TURN: __a. Launch Stata. __b. In the command window, type: use “http://www.pauldickman.com/survival/ivf”, clear __c. In the command window, type: histogram hyp, discrete __d. Save the graph to your desktop under the name hypertension_bar.png



4. Enter Data – How to Create a New Data Set in Stata Using the Data Editor Icon

When to Create a New Data Set in Stata Most of the time, you will not create a data set in Stata. Instead, you will import it from Excel (described next) or access an existing Stata data set (described later). Still, it’s good to know how to do this.

YOUR TURN – Create a Stata Data Set Containing the Following Data Follow the commands below to create a data set containing n=4 observations on the variables id, dob, gender, and weight. Tip – This illustration utilizes the different types of variables! Before you begin, type the following two commands into the command window: clear set more off

id

type: numeric

dob

type: date

gender

type: string/character

weight

type: numeric 1 3/26/1926 male 161.3 2 6/9/1956 female 120.1 3 4/1/1954 male 223.2 4 11/4/1951 female 124.0

Tip – The following pertains to “step 2” on the next page - how to find the data editor icon



. * STEP 1: Define your variables (lower case recommended), set type, and intialize . generate id=. . generate str8 dob_string="" . generate str8 gender="" . generate weight=. . * STEP 2: Click on DATA EDITOR icon to access an initially empty spreadsheet . * ---- See picture on page 8. Enter the data. Then close the data editor window ---* . * STEP 3: Create dob (date of birth) that is a Stata date variable . generate dob=date(dob_string, "MDY") . format dob %tdNN/DD/CCYY . drop dob_string . list . * STEP 4: Create 0/1 indicator of female gender . generate female=(gender=="female") . list . * STEP 5: Attach labels to variable names . label variable id "Subject id" . label variable weight "weight (lbs)" . label variable dob "Date of birth" . label variable female "0/1 female" . * STEP 6: Define dictionary of discrete variable values . label define femalef 0 "male" 1 "female" . * STEP 7: Attach coding labels to discrete variable values . label values female femalef . list . * To see data with numeric labels provided . numlabel, add . list . * To drop display of numeric labels . numlabel, remove . list . * STEP 8 - Save data set using FILE > SAVE AS



5. Enter Data – How to Import an Excel Data Set

When to Import an Excel Data Set There may be many occasions where you have an excel file and you want to look at it or analyze it in Stata. There are two ways to import an excel data set: (1) copy and paste; and (2) FILE > IMPORT in Stata. Tip – Take care that you have saved each variable type correctly. METHOD 1 – Copy and Paste

Step 1 – While still in Excel, use FORMAT > CELLS to format each column of data (text, date, numeric, as appropriate)

Step 2 – While still in Excel, select the data to be copied, including row headings with variable names. Use EDIT > COPY to complete selection.



Step 3 – Launch Stata. From tool bar, click on the icon DATA EDITOR

Step 4 – Position cursor in cell A1. From the STATA tool bar, use EDIT > PASTE to paste data here.



Step 5 – TIP! Be sure to click on VARIABLE NAMES so that Stata knows first row is variable names

You should now see the following



Step 6 – Close the DATA EDITOR window Tip – Don’t worry; data is not lost. You will save it next

Step 7 – SAVE your imported data using FILE > SAVE AS …



METHOD 2 – FILE > IMPORT in Stata

Step 1 – Launch Stata. Click on FILE > IMPORT > EXCEL SPREADSHEET (*.xls, *.xlsx)

Step 2 – Take care to complete the dialog boxes: (a) BROWSE to locate file and (b) IMPORT 1st row as variable names . Now you can click on the OK icon



You should see the following (but with your path and name, not mine) in your results window

Step 3 – In the command window, type the following: describe

You should now see the following in your results window

Step 4 – From the menu bar, save your data using FILE > SAVE AS … Complete dialog boxes.



YOUR TURN – Create an Excel Data set. Bring it into Stata by method 1 or 2. Save. __1. Launch Excel. __2. Create an Excel data set called stata_lab1.xls __3. Create a Stata data set stata_lab1.dta using your excel data. __4. Save your Stata data set.

Here is the excel data for you:

id

type: numeric

dob

type: date

gender

type: string/character

weight

type: numeric 1 3/26/1926 male 161.3 2 6/9/1956 female 120.1 3 4/1/1954 male 223.2 4 11/4/1951 female 124.0



6. Enter Data – How to Import a Stata Data Set from the Internet

When to Import a Stata Data Set You will do this often for class and perhaps not so often in your work. Tips – (1) Be sure to enclose the url in quotes, (2) Be sure to use the option clear after the comma; and (3) Be sure to save the data onto your computer so that you have it for your use later.

The basic command is of the following form and is issued in the command window use “http://fullurlpath”, clear To save the data onto your computer, from the top menu bar issue: FILE > SAVE AS ..

Examples - use “http://www.pauldickman.com/survival/ivf.dta”, clear use “http://people.umass.edu/biep640w/datasets/week02.dta”, clear use “http://people.umass.edu/biep640w/datasets/larvae.dta”, clear Tip – The extension “.dta” is optional. I recommend using it.

YOUR TURN – Import ivf.dta from the internet __1. Launch Stata (if you have not already done so) __2. In the command window, type: set more off __3. In the command window, type: use “http://www.pauldickman.com/survival/ivf”, clear __4. Use FILE > SAVE AS… to save it to your computer



7. Describe Your Data - Numerical Descriptions

YOUR TURN The following is a session that you can duplicate on your own. Tip - Commands that begin with an asterisk (*) are comments. The highlights in blue are my doing not Stata.

. * Use the command USE to access data set from the internet, if you don’t already have it. . use “http://www.pauldickman.com/survival/ivf”, clear . * Use the command aorder to reorder variables in alphabetic order . aorder . * Descriptives for dicrete variable - Use the command tab1 . tab1 sex, missing . * Descriptives again, this time with the labels and missing values . numlabel, add . tab1 sex, missing . * Descriptives for continuous variable - Use either tabstat or summarize . summarize bweight . summarize bweight, detail . tabstat bweight, stat(n, mean sd sem min q max cv) missing . * Descriptives for TWO dicrete variables - Use the command tab2 . * tab2 rowvariable columnvariable, options . tab2 sex hyp . tab2 sex hyp, row . tab2 sex hyp, row column cell exact chi2 . * Descriptives for ONE continuous variable, by groups defined by a dicrete variable . * Must sort by the discrete variable first . sort sex . tabstat bweight, by(sex) col(stat) stat(n mean sd sem min q max) . * use option LONGSTUB if you want to be reminded of the variable you are summarizing . tabstat bweight, by(sex) col(stat) stat(n mean sd sem min q max) longstub



8. Describe Your Data – Graphical Descriptions

YOUR TURN Again, follow along …

. * Tell Stata what graph scheme you want to use. I like the scheme s1color . set scheme s1color . * BAR GRAPH - Use the command histogram with the option discrete . histogram hyp, discrete percent addlabels ylabel(0(20)100) xlabel(0 "Normotensive" 1 "Hypertensive") gap(50) title("Bar Graph -Hyp") subtitle("n=639") caption("hyp_barchart.png") . * DOT PLOT – Use the command dotplot . dotplot matage, center msize(vsmall) xlabel(1 "All") title("Distribution of Maternal Age") subtitle("n=641") caption("dot_matage.png", size(vsmall)) . DOT PLOT for more than 1 group – Use the command dot plot with option over( ). Sort first. . sort sex . dotplot matage, over(sex) title("Distribution of Maternal Age") subtitle("by infant sex") caption("dot2_matage.png", size(vsmall)) . * BOX PLOT for more than 1 group – Use the command graph box with option over( ). . sort sex . graph box matage, over(sex) title("Distribution of Maternal Age") subtitle("by infant sex") caption("box2_matage.png", size(vsmall)) .* HORIZONTAL BOX PLOT for more than 1 group – Use the command graph hbox . graph hbox matage, over(sex) title("Distribution of Maternal Age") subtitle("by infant sex") caption("hbox2_matage.png", size(vsmall))



. * HISTOGRAM – Preliminary: It’s a good idea to obtain min and max . tabstat matage, stat(min max) . * HISTOGRAM – Use the command histogram . histogram matage, width(5) start(20) percent ylabel(0(10)50) addlabels title("Distribution of Maternal Age") subtitle("n=641") caption("histogram_matage.png", size(vsmall)) . * X-Y SCATTERPLOT - Preliminary: Obtain min and max of the X and Y variables . tabstat matage gestwks, stat(min max) .* X-Y SCATTERPLOT . graph twoway (scatter gestwks matage, msymbol(d) msize(vsmall)), xlabel(20(5)45) ylabel(20(5)45) title("Scatterplot") ytitle("Weeks Gestation",size(small)) caption("scatter.png", size(vsmall)) .* X-Y SCATTERPLOT WITH OVERLAY LINEAR FIT . graph twoway (scatter gestwks matage, msymbol(d) msize(vsmall)) (lfit gestwks matage), xlabel(20(5)45) ylabel(20(5)45) legend(off) title("Scatterplot") subtitle("with overlay linear fit") ytitle("Weeks Gestation",size(small)) caption("lfit.png", size(vsmall)) . * X_Y SCATTERPLOT WITH OVERLAY FIT AND 95% CI . graph twoway (scatter gestwks matage, msymbol(d) msize(vsmall)) (lfitci gestwks matage), xlabel(20(5)45) ylabel(20(5)45) legend(off) title("Scatterplot") subtitle("with overlay linear fit and 95% CI") ytitle("Weeks Gestation",size(small)) caption("lfitci.png", size(vsmall))



9. One and Two Sample Inference

YOUR TURN Again, follow along!

. ** ONE CONTINUOUS VARIABLE - 99% CI for mean using command ci . ci gestwks, level(99) . ** ONE CONTINUOUS VARIABLE - Test of null: mean = 40 using command ttest . ttest gestwks=40 . ** ONE CONTINUOUS VARIABLE - Test of null: standard deviation = 1 using command sdtest . sdtest gestwks=1 . ** TWO CONTINUOUS VARIABLES - Test of null: Equality of 2 INDEPENDENT means using ttest . sort sex . ttest gestwks, by(sex) . ttest gestwks, by(sex) unequal . ** TWO CONTINUOUS VARIABLES - Test of equality of 2 independent variances using sdtest . sdtest gestwks, by(sex) . ** 1 DISCRETE (0/1)VARIABLE - Test of binomial proportion using command bitest and prtest . ** Test of null: proportion of female births = .50 . generate female=(sex==2) . bitest female=.50 . prtest female=.50 . ** 1 DISCRETE (0/1)VARIABLE - 95% CI for event probability using ci with option binomial . ci female, binomial level(95) . ** 2 DISCRETE (0/1)VARIABLES - Test of equality of probabilities prtest with option by( ) . sort sex . prtest hyp, by(sex) . ** 2 DISCRETE VARIABLES (any # rows and # columns) - chi square test: tab2 with option chi2 . tab2 sex hyp, row column chi2 . ** 2 DISCRETE VARIABLES (any # rows and # columns) -fisher exact test: tab2 with option exact . tab2 sex hyp, row column exact



10. Simple and Multiple Linear Regression

. clear . * Data are from Vittinghoff et al. Regression Methods in Biostatistics . * Y=glucose X=physact (5 levels), BMI . use "http://people.umass.edu/biep640w/datasets/hersdata.dta" . ** descriptives of glucose, by levels of physical activity . sort physact . tabstat glucose, by(physact) col(stat) stat(n mean sd min p50 max) format(%8.2f) longstub .** descriptive graphs . dotplot glucose, over(physact) center msymbol(d) msize(tiny) xlabel(1 "1" 2 "2" 3 "3" 4 "4" 5 "5") subtitle("Y=Glucose by Physical Activity") note("dot_glucose.png") . graph matrix glucose age BMI, half msize(tiny) subtitle("Y=glucose v Predictors") note("scattermatrix.png", size(vsmall)) . ** Assessment of normality of Y=glucose . swilk glucose . sfrancia glucose . tabstat glucose, stat(min max) . histogram glucose, start(0) width(25) percent addlabels normal ylabel(0(20)100) title("Y=Glucose") note("histogram_glucose.png", size(vsmall)) . ** One predictor - continuous . regress glucose BMI . ** One predictor - nominal physical activity with design variables . xi: regress glucose i.physact . ** multiple predictor model with both BMI and physact . xi: regress glucose BMI i.physact . ** Partial F test of BMI controlling for physact: 1 df Partial F . testparm BMI . ** Partial F test of physical activity controlling for BMI: 4 df Partial F . testparm _Iphysact*

Stata version 12people.umass.edu/biep640w/pdf/stata v 12 lab session 1.pdfStata version 12 Lab...

Documents

Transcript of Stata version 12people.umass.edu/biep640w/pdf/stata v 12 lab session 1.pdfStata version 12 Lab...