Missing data with nlme

The big difference, which may not be relevant in your particular case, is that using the na.omit() function on the entire dataset removes any row that has a missing value in it anywhere, even if you are not using that variable in themodel. Using the na.action = na.omit only removes rows for the variables you are using in your model.

Here's a toy example dataset.

dat1 = data.frame(x = c(1:4),
                 y = c(3, 3, 4, 5),
                 x2 = c(1, 4, 5, NA))

I want to regress y vs x but I have a second variable, x2, that has a missing value in it.

Using the na.omit() function, the analysis is only done on three rows since the last row is removed due to the missing value in x2.

summary(lm(y ~ x, data = na.omit(dat1)))
Call:
lm(formula = y ~ x, data = na.omit(dat1))

Residuals:
      1       2       3 
 0.1667 -0.3333  0.1667 

Coefficients:
            Estimate Std. Error t value Pr(>|t|)
(Intercept)   2.3333     0.6236   3.742    0.166
x             0.5000     0.2887   1.732    0.333

Residual standard error: 0.4082 on 1 degrees of freedom
Multiple R-squared:   0.75,	Adjusted R-squared:    0.5 
F-statistic:     3 on 1 and 1 DF,  p-value: 0.3333

Using na.action = na.omit all four rows are used.

summary(lm(y ~ x, data = dat1, na.action = na.omit))

Call:
lm(formula = y ~ x, data = dat1, na.action = na.omit)

Residuals:
   1    2    3    4 
 0.3 -0.4 -0.1  0.2 

Coefficients:
            Estimate Std. Error t value Pr(>|t|)  
(Intercept)   2.0000     0.4743   4.216   0.0519 .
x             0.7000     0.1732   4.041   0.0561 .
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.3873 on 2 degrees of freedom
Multiple R-squared:  0.8909,	Adjusted R-squared:  0.8364 
F-statistic: 16.33 on 1 and 2 DF,  p-value: 0.05612

For your last question about how to calculate specific statistical results from mixed models fit with lme(), I'd recommend asking a new question. Make sure to include a reproducible example so folks can help you. See this FAQ on how to include a reproducible example.

2 Likes