**Stata FAQ**

**Path Analysis**

It is not very difficult to perform path analysis using Stata's **regress** command, but it
does require the use of a **regress** command for each stage in the path analysis model. **pathreg**
is a convenience command that can compute the path analysis with a single command.
You can download **pathreg **over the internet by typing **findit
pathreg** (see
How can I used the findit command to search for programs and get additional
help? for more information about using **findit**).

Let's say that we want to estimate the following path model using the **hsb2** dataset.

We will begin by downloading the data over the internet and getting the correlation between
the two exogenous variables,

**read** and

**write**.

**use http://www.ats.ucla.edu/stat/data/hsb2, clear**
**corr read write**
(obs=200)
| read write
-------------+------------------
read | 1.0000
write | 0.5968 1.0000

This path analysis is really just two regression models. The first model is

**math = _cons + read + write** while
the second model is

**science = _cons + math + read + write**. Using

**pathreg** we just
place each model inside parentheses (leaving off the equal signs, plus signs and constants).

**pathreg (math read write)(science math read write)**
------------------------------------------------------------------------------
math | Coef. Std. Err. t P>|t| Beta
-------------+----------------------------------------------------------------
read | .4169486 .0564838 7.38 0.000 .4563134
write | .3411219 .0610982 5.58 0.000 .3451322
_cons | 12.86507 2.82162 4.56 0.000 .
------------------------------------------------------------------------------
n = 200 R2 = 0.5153 sqrt(1 - R2) = 0.6962
------------------------------------------------------------------------------
science | Coef. Std. Err. t P>|t| Beta
-------------+----------------------------------------------------------------
math | .3190094 .0766753 4.16 0.000 .301854
read | .3015317 .0686815 4.39 0.000 .3122533
write | .2065257 .0707644 2.92 0.004 .1977167
_cons | 8.407353 3.192799 2.63 0.009 .
------------------------------------------------------------------------------
n = 200 R2 = 0.4999 sqrt(1 - R2) = 0.7071

We will use the standardized regression coefficients (Beta) as our path coefficients. Now we can
add the path coefficients and errors,
sqrt(1 - R^{2}) to the path diagram as shown below.

The content of this web site should not be construed as an endorsement
of any particular web site, book, or software product by the
University of California.