pool.analyses: Pool analyses results given M latent variable data sets, and...

In rosiezou/440proj: Analyze Continuous Latent Variables From Ordinal Survey Responses

Description

Pool analyses results given M latent variable data sets, and estimate parameters

Usage

pool.analyses(latent.datasets, formula, method = lm)

Arguments

latent.datasets	a (non-empty) list of lists returned by the gen.latent.vars function.
formula	a valid formula in the form of response.var ~ independent.vars
method	a regression function (e.g. lm). Currently only supports lm

Value

A list of 10 elements consisting of: point estimate for parameter Q, within-imputaiton variance, estimates of parameter Q obtained from M multiple imputations, estimates of variance, difference between estimates of parameter Q and the final point estimate, between-imputation variance, total variance, relative increase in variance due to nonresponse, fraction of missing information, and results from hypothesis testing with H_0: beta_i = 0 for i = 1 ... number_of_parameters. The mean and variance are calculated according to Rubin's Rules for Multiple Imputation. For more mathematical details, please refer to page 5 of this UCLA paper https://stats.idre.ucla.edu/wp-content/uploads/2016/02/multipleimputation.pdf \(section title "Combining Inferences from Imputed Data Sets"\)

Examples

# Create indicators (a label indicating which latent variable the question corresponds to)
grp.indicator <- sapply(names(multiis), FUN =
                         function(x){strsplit(x, split = "_")[[1]][2]})

latent.datasets <- gen.latent.datasets(5, multiis, grp.indicator = grp.indicator, num.iter = 1)

lm.pool <- pool.analyses(latent.datasets, cat~comp+int, lm)

rosiezou/440proj documentation built on May 12, 2019, 6:25 p.m.