R Regression

Let's assume we want to run a regression with lwage (the logarithm of the woman's wage) as dependent variable and a constant, exper (the years of experience) and the logarithm of the husbands wage (huswage as explanatory variables. First we should note that the logarithm of the woman's wage already exists as variable lwage, but the logarithm of the husband's wage doesn't exist as its own variable. Hence we are yet to calculate it.

The lm() function

The R function that does the heavy lifting for regression analysis is the lm() function (presumably an abbreviation for "linear model") and we will have a close up look at how it works. But let's get our first regression under the belt.

A first example

The following few lines of code (which you should save in a script) import the data, convert missing data to NAs (see R_Data#Data_Types) and eventually runs a regression:

    # This is my first R regression!
    setwd("T:/ECLR/R/FirstSteps")              # This sets the working directory
    mydata <- read.csv("mroz.csv")  # Opens mroz.csv from working directory
     
    # Now convert variables with "." to num with NA
    mydata$wage <- as.numeric(as.character(mydata$wage))
    mydata$lwage <- as.numeric(as.character(mydata$lwage))
     
    # Run a regression
    reg_ex1 <- lm(lwage~exper+log(huswage),data=mydata)

So let's look at the last line in which we ask R to run a regression. Whatever comes in the parenthesis after lm are parameters to the lm() function. Different parameters are separated by commas. So here we have two inputs. Let's start with the second data=mydata. This basically indicates to R that we are drawing the data for the regression from our dataframe called mydata. That means that for the first input, in which we actually specify the model we estimate, we can refer to the variable names of the variables that are contained in mydata. In that first input you should imagine writing down a regression model. The model we want to estimate is the following:

[math]\label{OLSModel} lwage = \beta_0 + \beta_1 * exper + \beta_2 * log(huswage) + \epsilon[/math]

The way how you tell lm() to estimate this model is to leave the coefficients and error term away, and replace the equal sign with a ~. This is where the bold part of

    reg_ex1 <- lm(lwage~exper+log(huswage),data=mydata)

comes from. One additional note here. The log(huswage) part of this model took the huswage variable from our dataframe and applied the log() function to it.

You should be familiar with the left hand side of the command, which assigns the results of the regression to a new object called reg_ex1. You should think of this as some sort of folder in which R has now saved all the regression results. If you look at the object reg_ex1 in your environment, you will see that it is a list object with 13 elements. So far so good, but if you click on the little triangle next to its name and see the detail you will most likely scratch your head and think "What a mess!" and I think you are quite right.

Regression Output

You would be familiar with a standard regression output containing estimated coefficients, standard errors and a set of regression statistics. Fortunately it is very easy to obtain this. The command is

    > summary(reg_ex1)
     
    Call:
    lm(formula = lwage ~ exper + log(huswage), data = mydata)
     
    Residuals:
         Min       1Q   Median       3Q      Max 
    -3.10089 -0.31219  0.02919  0.37466  2.11402 
     
    Coefficients:
                 Estimate Std. Error t value Pr(>|t|)    
    (Intercept)  0.534866   0.139082   3.846 0.000139 ***
    exper        0.016684   0.004243   3.933 9.81e-05 ***
    log(huswage) 0.236466   0.063684   3.713 0.000232 ***
    ---
    Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
     
    Residual standard error: 0.7031 on 425 degrees of freedom
      (325 observations deleted due to missingness)
    Multiple R-squared:  0.05919,	Adjusted R-squared:  0.05477 
    F-statistic: 13.37 on 2 and 425 DF,  p-value: 2.338e-06

Most things should sound familiar here. One thing you should note is that, on this occasion, R automatically ignored all the missing observations, all 325 of them, which left R to estimate the model with 428 observations

Accessing Regression Output

The next thing you should want to know as a buddying applied econometrician is how you actually access these regression statistics in order to use them in some further analysis. Let's see how you would access the regression residuals.