mi (version 0.06-5)

mi.pooled: Modeling Functions for Multiply Imputed Dataset

Description

Modeling Function that pulls together the estimates from multiply imputed dataset.

Usage

mi.pooled(coef, se, m)
lm.mi(formula, mi.object, ...)
glm.mi(formula, mi.object, family = gaussian, ...)
bayesglm.mi(formula, mi.object, family = gaussian, ...)
polr.mi(formula, mi.object, ...)
bayespolr.mi(formula, mi.object, ...)
lmer.mi(formula, mi.object, rescale=FALSE, ...)
glmer.mi(formula, mi.object, family = gaussian, rescale=FALSE, ...)
## S3 method for class 'mi.pooled':
print(x, \dots)
## S3 method for class 'mi.pooled':
coef(object)
## S3 method for class 'mi.pooled':
se.coef(object)
## S3 method for class 'mi.pooled':
display(object, digits=2)

Arguments

coef

list of coefficients

se

list of standard errors

m

number of chains for the imputation

formula

See lm, glm, polr, lmerfor detail.

mi.object

mi object

family

See glm, polr, lmerfor detail.

rescale

default is FALSE, see rescale for detail.

x

mi.pooled object.

object

mi.pooled object.

digits

number of significant digits to display, default=2.

...

Any option to pass on to lm, glm, bayesglm, bayespolr<

Value

• call
  the matched call.
• mi.pooled
  pulled estimates from the multiple dataset.
• mi.fit
  estimates from each dataset.

References

Andrew Gelman and Jennifer Hill, Data Analysis Using Regression and Multilevel/Hierarchical Models, Cambridge University Press, 2007.

See Also

lm, glm, bayesglm, bayespolr, polr, and lmer

Examples

# true data
  n <- 100
  x <- rbinom(n,1,.45) 
  z <- ordered(rep(seq(1, 5),n)[sample(1:n, n)])
  y <- rnorm(n)
  group <- rep(1:10, 10)

  # create artificial missingness
  dat.xy <- data.frame(x, z, y)
  dat.xy <- mi:::.create.missing(dat.xy, pct.mis=10)  

  # imputation 
  dat.mi <- mi(dat.xy, n.iter=6, preprocess=FALSE, add.priors=FALSE)

  # fit models
  M1 <- lm.mi(y ~ x + z, dat.mi)
  display(M1)
  coef(M1)
  se.coef(M1)

  M2 <- glm.mi(x ~ y , dat.mi, family = binomial(link="logit"))
  display(M2)
  coef(M2)
  se.coef(M2)

  M3 <- bayesglm.mi(x ~ y , dat.mi, family = binomial(link="logit"))
  display(M3)
  coef(M3)
  se.coef(M3)

  M4 <- polr.mi(ordered(z) ~  y, dat.mi)
  display(M4)
  coef(M4)
  se.coef(M4)

  M5 <- bayespolr.mi(ordered(z) ~  y, dat.mi)
  display(M5)
  coef(M5)
  se.coef(M5)

  M6 <- lmer.mi(y ~ x  + (1|group), dat.mi)
  display(M6)
  coef(M6)
  se.coef(M6)

  M7 <- glmer.mi(x ~ y  + (1|group), dat.mi, family = binomial(link="logit"))
  display(M7)
  coef(M7)
  se.coef(M7)