10.08.2013 Views

Introduction to Stata 8

Introduction to Stata 8

Introduction to Stata 8

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

10.7. Reshaping data<br />

collapse [R] collapse<br />

You want <strong>to</strong> create an aggregated data set, not with the characteristics of each individual, but<br />

of groups of individuals. One situation might be <strong>to</strong> characterize physicians by number of<br />

patient contacts, another <strong>to</strong> make a reduced data set for Poisson regression (see section 13):<br />

. // gen.stcollaps.cancer2.do<br />

. use c:\dokumenter\proj1\stsplit.cancer2.dta , clear<br />

. collapse (sum) risktime died , by(agegr drug)<br />

. save c:\dokumenter\proj1\stcollaps.cancer2.dta<br />

. summarize<br />

Variable | Obs Mean Std. Dev. Min Max<br />

-------------+-----------------------------------------------------<br />

drug | 15 2 .8451543 1 3<br />

agegr | 15 55 7.319251 45 65<br />

risktime | 15 4.133334 3.01067 .3581161 10.88906<br />

died | 15 2.066667 2.374467 0 8<br />

reshape [R] reshape<br />

E.g. with repeated measurements some analyses require a 'wide', some a 'long' data structure:<br />

Long structure Wide structure<br />

id time sbp<br />

1 1 140<br />

1 2 120<br />

1 3 130<br />

1 4 135<br />

2 1 155<br />

etc.<br />

id sbp1 sbp2 sbp3 sbp4<br />

1 140 120 130 135<br />

2 155 etc.<br />

Switch between structures by:<br />

reshape wide sbp , i(id) j(time) // Restructures long <strong>to</strong> wide<br />

reshape long sbp , i(id) j(time) // Restructures wide <strong>to</strong> long<br />

xpose [R] xpose<br />

You may transpose observations and variables, i.e. let observations become variables and<br />

variables become observations. This may be useful e.g. for restructuring data for a graph:<br />

xpose , clear // clear is not optional<br />

expand [R] expand, contract<br />

You may duplicate observations according <strong>to</strong> a weighting variable. Look at the example in<br />

section 7 about weighting observations. You can obtain the same result by expanding the four<br />

observations <strong>to</strong> 130:<br />

expand pop<br />

tabulate expos case , chi<br />

contract does the opposite of expand.<br />

26

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!