display {arm} | R Documentation |
This generic function gives a clean printout of lm, glm, mer and polr objects.
display (object, ...) # methods for display() display.lm (object, digits=2) display.glm (object, digits=2) display.mer (object, digits=2) display.polr (object, digits=2)
object |
The output of a call to lm, glm, mer, polr, or related regressions function with n data points and k predictors. |
... |
further arguments passed to or from other methods. |
digits |
number of significant digits to display. |
This generic function gives a clean printout of lm, glm, mer and polr objects, focusing on the most pertinent pieces of information: the coefficients and their standard errors, the sample size, number of predictors, residual standard deviation, and R-squared. Note: R-squared is automatically displayed to 2 digits, and deviances are automatically displayed to 1 digit, no matter what.
Coefficients and their standard errors, the sample size, number of predictors, residual standard deviation, and R-squared
Output are the model, the regression coefficients and standard errors, and the residual sd and R-squared (for a linear model), or the null deviance and residual deviance (for a generalized linear model).
Andrew Gelman gelman@stat.columbia.edu; Yu-Sung Su ys463@columbia.edu; Maria Grazia Pittau grazia@stat.columbia.edu
Andrew Gelman and Jennifer Hill, Data Analysis Using Regression and Multilevel/Hierarchical Models, Cambridge University Press, 2006.
# Here's a simple example of a model of the form, y = a + bx + error, # with 10 observations in each of 10 groups, and with both the # intercept and the slope varying by group. First we set up the model and data. group <- rep(1:10, rep(10,10)) mu.a <- 0 sigma.a <- 2 mu.b <- 3 sigma.b <- 4 rho <- 0 Sigma.ab <- array (c(sigma.a^2, rho*sigma.a*sigma.b, rho*sigma.a*sigma.b, sigma.b^2), c(2,2)) sigma.y <- 1 ab <- mvrnorm (10, c(mu.a,mu.b), Sigma.ab) a <- ab[,1] b <- ab[,2] # x <- rnorm (100) y1 <- rnorm (100, a[group] + b[group]*x, sigma.y) y2 <- rbinom(100, 1, prob=invlogit(a[group] + b*x)) # # display a simple linear model # M1 <- lm (y1 ~ x) display (M1) # # display a simple logit model # M2 <- glm (y2 ~ x, family=binomial(link="logit")) display (M2) # # Then fit and display a simple varying-intercept model: # M3 <- lmer (y1 ~ x + (1|group)) display (M3) M3.sim <- mcsamp (M3) print (M3.sim) plot (M3.sim) # # Then the full varying-intercept, varying-slope model: # M4 <- lmer (y1 ~ x + (1 + x |group)) display (M4) M4.sim <- mcsamp (M4) print (M4.sim) plot (M4.sim) # # Then the full varying-intercept, logit model: # M5 <- lmer (y2 ~ x + (1|group), family=binomial(link="logit")) display (M5) M5.sim <- mcsamp (M5) print (M5.sim) plot (M5.sim) # # Then the full varying-intercept, varying-slope logit model: # M6 <- lmer (y2 ~ x + (1 + x |group), family=binomial(link="logit")) display (M6) M6.sim <- mcsamp (M6) print (M6.sim) plot (M6.sim) # # Then the ordered logit model from polr # M7 <- polr(Sat ~ Infl + Type + Cont, weights = Freq, data = housing) display(M7) M8 <- bayespolr(Sat ~ Infl + Type + Cont, weights = Freq, data = housing) display(M8)