Fitting generalized linear mixed effect models - serverside function

glmerSLMADS2 fits a generalized linear mixed effects model (glme) - e.g. a logistic or Poisson regression model including both fixed and random effects - on data from one or multiple sources with pooling via SLMA (study level meta-analysis)

glmerSLMADS2(
  formula,
  offset,
  weights,
  dataName,
  family,
  control_type = NULL,
  control_value.transmit = NULL,
  nAGQ = 1L,
  verbose = 0,
  theta = NULL,
  fixef = NULL
)

Arguments

formula

see help for ds.glmerSLMA

offset

see help for ds.glmerSLMA

weights

see help for ds.glmerSLMA

dataName

see help for ds.glmerSLMA

family

see help for ds.glmerSLMA

control_type

see help for ds.glmerSLMA

control_value.transmit

see help for argument <control_value> for function ds.glmerSLMA

nAGQ

integer scalar, defaulting to 1L. IN PRACTICE, IT MAY BE NECESSARY TO SET nAGQ TO 0L when the model appears to converge perfectly well (e.g. verbose=2 demonstrates good initial convergence of both the log-likelihood and regression coefficients) but formal convergence does not get declared - so no output is produced - despite running the model for many iterations. The nAGQ argument is set by the nAGQ argument for ds.glmerSLMA and further details can be found in help(ds.glmerSLMA) and in the native R help for glmer()

verbose

see help for ds.glmerSLMA

theta

see help for argument <start_theta> for function ds.glmerSLMA

fixef

see help for argument <start_fixef> for function ds.glmerSLMA

Value

all key model components see help for ds.glmerSLMA

Details

glmerSLMADS2 is a serverside function called by ds.glmerSLMA on the clientside. The analytic work engine is the glmer function in R which sits in the lme4 package. ds.glmerSLMA fits a generalized linear mixed effects model (glme) - e.g. a logistic or Poisson regression model including both fixed and random effects - on data from a single or multiple sources. When there are multiple data sources, the glme is fitted to convergence in each data source independently and the estimates and standard errors returned to the client thereby enabling cross-study pooling using study level meta-analysis (SLMA). By default the SLMA is undertaken using the metafor package, but as the SLMA occurs on the clientside which, as far as the user is concerned is just a standard R environment, the user can choose to use any approach to meta-analysis they choose. Additional information about fitting glmes using the glmer engine can be obtained using R help for glmer and the lme4 package

Author

Tom Bishop, with some additions by Paul Burton