#MARGINS STATA HOW TO#

Thus it considers the effect of changing the Honda Civic's weight from 1,760 pounds as well as changing the Lincoln Continental's from 4,840 (the weight squared term is more important with the latter than the former). In doing so, margins looks at the actual data. What margins does here is take the numerical derivative of the expected price with respect to weight for each car, and then calculates the mean. The margins command can very easily tell you the mean effect: Thus the net effect of changing weight for any given car will very much depend on its starting weight. This case is particularly confusing (but not unusual) because the coefficient on weight is negative but the coefficient on weight squared is positive. But consider changing weight: since the model includes both weight and weight squared you have to take into account the fact that both change. In this simple case, the derivative is just the coefficient on mpg, which will always be the case for a linear model. If you want to look at the marginal effect of a covariate, or the derivative of the mean predicted value with respect to that covariate, use the dydx option: (The actual weights range from 1760 to 4840.) Again, this is a good candidate for a graphic: This calculates the mean predicted value of price with weight set to 1500, 2000, 2500, etc. You can also define a numlist with the by specifying start ( interval ) end: The simplest method is just to list the numbers you want, as above. A numlist is a list of numbers just like a varlist is a list of variables and, like a varlist, there are many different ways to define a numlist. The values are specified using a numlist. The margins output first assigns a number to each scenario, then gives their results by number. Think of each value as a 'scenario'-the above scenarios are very simple, but you can make much more complicated scenarios by listing multiple variables and values in the at option. This calculates the mean predicted value of price with weight set to 2000 pounds, and then again with weight set to 4000 pounds. The marginsplot command takes the results of the previous margins command and turns them into a graph:ДЇor continuous variables margins obviously can't look at all possible values, but you can specify which values you want to examine with the at option: Thus:ДЇirst asks, 'What would the mean price be if all the cars were domestic?' (but still had their existing weights, displacements, etc.) and then asks 'What would the mean price be if all the cars were foreign?'Дoes the same for all five values of rep78, but since there are so many of them it's a good candidate for a graphical presentation. Then, for each value it calculates what the mean predicted value of the dependent variable would be if all observations had that value for the categorical variable. If margins is followed by a categorical variable, Stata first identifies all the levels of the categorical variable. Reg price c.weight#c.weight i.foreign i.rep78 mpg displacement Levels of the Outcome VariableДЄll by itself, Stata will calculate the predicted value of the dependent variable for each observation, then report the mean value of those predictions (along with the standard error, t-statistic, etc.). For our first example, load the auto data set that comes with Stata and run the following regression: The margins command can only be used after you've run a regression, and acts on the results of the most recent regression command. If you're familiar with that material you can to skip to section 3. Sections 1 and 2 are taken directly from the Statistics section of Stata for Researchers (they are reproduced here for the benefit of those looking specifically for information about using margins).

#MARGINS STATA HOW TO#

The margins command is a powerful tool for understanding a model, and this article will show you how to use it. Once you've run a regression, the next challenge is to figure out what the results mean.

Margins stata