psychmeta (version 2.3.4)

metareg: Compute meta-regressions

Description

This function is a wrapper for metafor's rma function that computes meta-regressions for all bare-bones and individual-correction meta-analyses within an object. It makes use of both categorical and continuous moderator information stored in the meta-analysis object and allows for interaction effects to be included in the regression model. Output from this function will be added to the meta-analysis object in a list called follow_up_analyses. If using this function with a multi-construct meta-analysis object from ma_r or ma_d, note that the follow_up_analyses list is appended to the meta-analysis object belonging to a specific construct pair within the construct_pairs list.

Usage

metareg(ma_obj, formula_list = NULL, ...)

Arguments

ma_obj

Meta-analysis object.

formula_list

Optional list of regression formulas to evaluate. NOTE: If there are spaces in your moderator names, replace them with underscores (i.e., "_") so that the formula(s) will perform properly. The function will remove spaces in the data, you only have to account for this in formula_list when you supply your own formula(s).

...

Additional arguments.

Value

ma_obj with meta-regression results added (see ma_obj$follow_up_analyses$metareg).

Examples

# NOT RUN {
## Meta-analyze the data from Gonzalez-Mule et al. (2014)
## Note: These are corrected data and we have confirmed with the author that
## these results are accurate:
ma_obj <- ma_r_ic(rxyi = rxyi, n = n, hs_override = TRUE, data = data_r_gonzalezmule_2014,
                  rxx = rxxi, ryy = ryyi, ux = ux, indirect_rr_x = TRUE,
                  correct_rr_x = TRUE, moderators = Complexity)

## Pass the meta-analysis object to the meta-regression function:
ma_obj <- metareg(ma_obj)

## Examine the meta-regression results for the bare-bones and corrected data:
ma_obj$metareg[[1]]$barebones$`Main Effects`
ma_obj$metareg[[1]]$individual_correction$true_score$`Main Effects`


## Meta-analyze simulated d-value data
dat <- data_d_meas_multi
## Simulate a random moderator
set.seed(100)
dat$moderator <- sample(1:2, nrow(dat), replace = TRUE)
ma_obj <- ma_d(ma_method = "ic", d = d, n1 = n1, n2 = n2, ryy = ryyi,
               construct_y = construct, sample_id = sample_id,
               moderators = moderator, data = dat)

## Pass the meta-analysis object to the meta-regression function:
ma_obj <- metareg(ma_obj)

## Examine the meta-regression results for the bare-bones and corrected data:
ma_obj$metareg[[1]]$barebones$`Main Effects`
ma_obj$metareg[[1]]$individual_correction$latentGroup_latentY$`Main Effects`
# }