Transcript Stata - The University of Arizona College of Medicine

Stata
Workshop #1
Chiu-Hsieh (Paul) Hsu
Associate Professor
College of Public Health
[email protected]
Outline
•
•
•
•
•
•
Do files
Data entry
Data management
Data description
Estimation: Confidence Interval
Hypothesis testing
Do files
• Stata programs
– Easy to add or skip comments
– One click/command can run the whole
program
• Reproducible
– Don’t need to retype all of the commands
• Interactive work vs. do files
Data Entry
Stata Commands
1. cd: Change directory
2. dir or ls: Show files in current directory
3. insheet: Read ASCII (text) data created by a
spreadsheet
4. infile: Read unformatted ASCII (text) data
5. infix: Read ASCII (text) data in fixed format
6. input: Enter data from keyboard
7. save: Store the dataset currently in memory on disk
in Stata data format
8. use: Load a Stata-format dataset
9. count: Show the number of observations
10. list: List values of variables
11. clear: Clear the entire dataset and everything else
12. memory: Display a report on memory usage
13. set memory:Set the size of memory
Ways to enter data
•
•
•
•
•
•
•
•
•
•
•
•
•
Change the directory to the folder you like
cd c:\Stata
Common separated values (.csv) format files
insheet using test.csv,clear (with variable names)
infile gender id race ses schtyp str10 prgtype read write
math science socst using hs0.raw, clear (without variable
names)
Stata (.dta) files
use test
Type in data one by one
input id female race ses str3 schtype prog read write math
science socst
End (when you are done)
What’s in the dataset?
describe
list
Data Management
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
Stata Commands
pwd: show: current directory (pwd=print working directory)
keep if: keep observations if condition is met
Keep: keep variables or observations
drop: drop variables or observations
append: append a data file to current file
sort: sort observations
merge: merge a data file with current file
codebook: show codebook information for file
label data: apply a label to a data set
order: order the variables in a data set
label variable: apply a label to a variable
label define: define a set of a labels for the levels of a categorical variable
label values: apply value labels to a variable
encode: create numeric version of a string variable
rename a variable
recode: recode the values of a variable
notes: apply notes to the data file
generate: creates a new variable
replace: replaces one value with another value
egen: extended generate - has special functions that can be used when
creating a new variable
Merging two datasets
•
•
test1 and test2 have the same variables but
different subjects
use test1
append using test2
save test12
test3 and test4 have the same subjects and only
share a link variable, e.g. ID
use test3, clear
sort id
save test3,replace
use test4, clear
sort id
save test4,replace
use test3
merge id using test4
save test34
Play with Variables
•
•
•
•
•
•
•
•
•
use test
label variable gender "Male"
rename gender male
gen female=1-male
order id male female
encode prgtype, gen(prog)
codebook prog
keep if female==1 (delete male)
drop female
Dummy Variables
•
•
•
•
•
•
•
–
–
–
–
A categorical variable with K possible levels
Need K-1 dummy variables (one as the reference)
Dummy variables are convenient for regression
analysis
How to create dummy variables?
Use generate command
gen female=1-gender
Use tabulate command
tabulate gender, gen(male)
Use factor variables
xi i.gender
list,clean
Data Description
1.
2.
3.
4.
5.
6.
7.
8.
9.
Stata Commands
describe: describe a dataset
log: create a log file
summarize: descriptive statistics
tabstat: table of descriptive statistics
table: create a table of statistics
stem: stem-and-leaf plot
graph: high resolution graphs
kdensity: kernel density plot
histogram: histogram for continuous and categorical
variables
10. tabulate: one- and two-way frequency tables
11. correlate: correlations
12. pwcorr: pairwise correlations
Example: raw data
•
•
•
•
•
•
•
•
•
•
•
•
•
log using test.txt, text replace
use lead
describe
sum maxfwt, detail
histogram maxfwt, by(Group) normal
graph box maxfwt, by(Group)
stem maxfwt
kdensity maxfwt
tab Group sex
cor ageyrs maxfwt,sig
cor ageyrs maxfwt if sex==1 (male only),sig
pwcorr ageyrs maxfwt fwt_r,sig
log close
Example: grouped data
•
•
•
•
use group (a grouped dataset)
sum age [fweight=freq],detail
hist age [fweight=freq]
Pretty much the same as raw data. Just need to
specify the weight.
Some Review
•
•
•
•
Use both location and spread measures to
summarize a dataset
Mean, standard deviation and range are
easily affected by extreme observations
Median and inter-quartile range are less
affected by extreme observations
Coefficient of variation (standard deviation
divided by mean) removes the scale effect.
Estimation
Estimation of Parameters
•
Binomial distribution
–
–
•
Parameters n (usually known) and p
How to estimate p?
Poisson distribution
–
Parameter λ
– How to estimate λ?
• Normal distribution
– Parameters µ and σ2
– How to estimate µ and σ2?
– σ2 unknown  t distribution
Stata Commands
•
Raw data
–
ci [varlist] [if] [in] [weight] [, options]
•
•
confidence intervals for mean, proportion (b) and count
(p)
Summarry statistics
–
cii #obs #mean #sd [, ciin_option]
•
–
Normal
cii #obs #succ [, ciib_options]
•
Binomial
Examples
•
•
•
–
–
–
•
–
–
–
gen female=sex-1
tab female Group
What’s the average maxfwt for females in the
exposed group?
ci maxfwt if female==1 & Group==2 (raw data)
sum maxfwt if female==1 & Group==2
cii 16 59 20.887,level(95) (summary statistics)
What’s the proportion of females in the exposed
group?
gen expose=Group-1
ci expose if female==1,b
cii 48 16,level(95)
Hypothesis Testing
Stata Commands (mean)
•
ttest
–
Raw data
•
•
•
•
–
ttest varname == # [if] [in] [, level(#)]
ttest varname1 == varname2 [if] [in], unpaired [unequal
welch level(#)]
ttest varname1 == varname2 [if] [in] [, level(#)]
ttest varname [if] [in] , by(groupvar) [options1]
Summarry statistics
•
•
ttesti #obs #mean #sd #val [, level(#)]
ttesti #obs1 #mean1 #sd1 #obs2 #mean2 #sd2 [,
options2]
Examples
•
–
One sample
Is the average maxfwt for females in the exposed group
significantly lower than 45?
•
•
•
–
ttest maxfwt==45 if female==1 & Group==2
ttesti 16 59 20.887 45 (summary statistics)
Two samples
Do females have a higher average maxfwt than males in the
exposed group?
•
•
•
ttest maxfwt if Group==2, by(female)
sum maxfwt if female==0 & Group==2
ttesti 16 59 20.887 30 60.167 27.28
Stata Commands (variance)
•
sdtest
–
Raw data
•
•
•
–
sdtest varname == # [if] [in] [, level(#)]
sdtest varname1 == varname2 [if] [in] [, level(#)]
sdtest varname [if] [in] , by(groupvar) [level(#)]
Summarry statistics
•
•
sdtesti #obs {#mean | . } #sd #val [, level(#)]
sdtesti #obs1 {#mean1 | . } #sd1 #obs2 {#mean2 | .
} #sd2 [, level(#)]
Examples
•
–
One sample
Is the variance of maxfwt for females in the exposed group
significantly greater than 100?
•
•
•
–
sdtest maxfwt==10 if female==1 & Group==2
sdtesti 16 59 20.887 10 (summary statistics)
Two samples
Do females have a greater variation in maxfwt than males
in the exposed group?
•
•
•
sdtest maxfwt if Group==2, by(female)
sum maxfwt if female==0 & Group==2
sdtesti 16 59 20.887 30 60.167 27.28
Stata Commands (proportion)
•
prtest
–
Raw data
•
•
•
–
prtest varname == #p [if] [in] [, level(#)]
prtest varname1 == varname2 [if] [in] [, level(#)]
prtest varname [if] [in] , by(groupvar) [level(#)]
Summarry statistics
•
•
prtesti #obs1 #p1 #p2 [, level(#) count]
prtesti #obs1 #p1 #obs2 #p2 [, level(#) count]
Examples
•
•
–
–
One sample
Is it more than 50% of females in the exposed group?
•
•
prtest expose==0.5 if female==1
prtesti 48 0.3333333 0.5
Two samples
Are there more females in the exposed group than the
control group?
•
•
•
prtest female, by(expose)
tab expose female, r
prtesti 78 0.4103 46 0.3478
Power and Sample Size
Stata Command (sample size)
• One sample
– continuous
• sampsi μ0 μ1, sd(.) p(.) a(.) onesam
• sampsi 3500 3800, sd(420) p(.9) onesam
– Binary proportions
• sampsi p0 p1, p(.) onesam
• sampsi 0.4 0.25, p(0.9) onesam
• Two samples
– continuous
• sampsi μ1 μ2, p(.) sd1(.) sd2(.) a(.)
• sampsi 132.86 127.44, p(0.8) sd1(15.34) sd2(18.23)
– Binary proportions
• sampsi p1 p2, p(.)
• sampsi 0.4 0.25, p(0.9)
Stata Command (power)
• One sample
– continuous
• sampsi μ0 μ1, sd(.) n(.) a(.) onesam
• sampsi 84.4 90.1, sd(10.3) n(5) onesam onesided
– Binomial proportion
• sampsi p0 p1, n1(.) onesam
• sampsi 0.25 0.4, n1(100) onesam
• Two samples
– continuous
• sampsi μ1 μ2, n1(.) n2(.) sd1(.) sd2(.) a(.)
• sampsi 9 14, n1(100) n2(100) sd1(15.34) sd2(18.23)
– Binomial proportions
• sampsi p1 p2, n1(.) n2(.)
• sampsi 0.4 0.25, n1(100) n2(150)
Useful links
• http://www.ats.ucla.edu/stat/stata/
• Once the D2L site is created, all of
the handouts and related materials
will be posted on the D2L site.