In order to test whether there is a statistically significant effect of a particular macro level variable on a particular micro level outcome, you will probably wish to use some from of regression analysis. For example, suppose you are interested in whether PR increases trust in politicians and you want to control for the effect of generalized trust in people, you could then run the following regression.

Example 8: Regression analysis with both micro and macro variables

```
regress trstplt ppltrst pr

Source |       SS       df       MS              Number of obs =    6148
-------------+------------------------------           F(  2,  6145) =  402.38
Model |  3801.87099     2   1900.9355           Prob > F      =  0.0000
Residual |  29030.2259  6145  4.72420276           R-squared     =  0.1158
Total |  32832.0969  6147  5.34115779           Root MSE      =  2.1735

------------------------------------------------------------------------------
trstplt |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
ppltrst |   .3159584   .0112979    27.97   0.000     .2938106    .3381062
pr |    .196908   .0793005     2.48   0.013     .0414513    .3523648
_cons |   2.082625   .0891842    23.35   0.000     1.907793    2.257457
------------------------------------------------------------------------------

```

The results suggest that there is a significant positive effect of PR on trust in politicians even after controlling for trust in people generally. Note that macro and micro variables are introduced into the regression in entirely the same way.

This problem can be dealt with by using a multilevel model as described in Unit 6 in this series.

Countries and Citizens: Unit 4 Combining macro and micro data by Steve Fisher, University of Oxford is licensed under a Creative Commons Attribution-Non-Commercial-Share Alike 2.0 UK: England & Wales Licence.