Balance_MAGMA {MAGMA.R}

Balance_MAGMA

Description

This function computes all four balance criteria of 'MAGMA.R', namely Pillai's Trace, d-ratio, mean g, and adjusted d-ratio. The estimation involves both binary and metric variables. Balance estimation is performed across various sample sizes. See Details for more information.

Usage

Balance_MAGMA(Data, group, covariates, step = "step", verbose = TRUE)

Arguments

Data

A data frame containing at least the grouping variable, the step variable from the main MAGMA-function (or other matching algorithms), and all covariates of interest.

group

A character specifying the name of your grouping variable in data. Note that MAGMA can only match your data for a maximum of four groups. For matching over two grouping variables (e.g., 2x2 design) is possible by specifying group as a character vector with a length of two. In this case, each of the two grouping variables can only have two levels.

covariates

A character vector listing the names of all binary and metric covariates of interest.

step

A character specifying the step variable of the matching. Per default, it is set to step, which corresponds the resulting name of the main MAGMA function.

verbose

TRUE or FALSE indicating whether matching information should be printed to the console.

Details

This function computes all four balance criteria of 'MAGMA.R', namely Pillai's Trace, d-ratio, mean g, and adjusted d-ratio. This is an iterative process including more cases with each iteration according to the step variable. Thus, starting with cases having a small within-match distance, larger distances are included with increasing iterations. As a minimum the function specifies n >= 20 per group. This does not imply that balance criteria with such a small sample size can be estimated consistently. For Pillai's Trace a higher minimum sample size can be possible. It depends on the number of covariates to ensure a positive model identification. Missing data for Pillai's Trace are excluded listwise, while for the other balance criteria pairwise exclusion is applied.

Value

A list of length four containing all balance criteria and all pairwise effects with respect to group sample size.

Author(s)

Julian Urban

References

Pastore, M., Loro, P.A.D., Mingione, M., Calcagni, A. (2022). overlapping: Estimation of Overlapping in Empirical Distributions. R package version 2.1, (https://CRAN.R-project.org/package=overlapping).

Revelle, W. (2023). psych: Procedures for Psychological, Psychometric, and Personality Research. Northwestern University, Evanston, Illinois. R package version 2.3.6, (https://CRAN.R-project.org/package=psych).

Viechtbauer, W. (2010). Conducting meta-analyses in R with the metafor package. Journal of Statistical Software, 36(3), 1-48. (doi:10.18637/jss.v036.i03)

Fisher, Z., Tipton, E., Zhipeng, H. (2023). robumeta: Robust Variance Meta-Regression. R package version 2.1, (https://CRAN.R-project.org/package=robumeta).

Examples

# Defining the names of the metric and binary covariates
covariates_vector <- c("GPA_school", "IQ_score", "Motivation", "parents_academic", "gender")


# Estimating balance of a two-group matching using the data set
# 'MAGMA_sim_data'.
# Matching variable 'gifted_support' (received giftedness support yes or no)
# Using subsample only for faster execution.
Balance_gifted <- Balance_MAGMA(Data = MAGMA_sim_data[MAGMA_sim_data$step_gifted < 200, ],
                                group = "gifted_support",
                                covariates = covariates_vector,
                                step = "step_gifted")
str(Balance_gifted)


# 2x2 matching using the data set 'MAGMA_sim_data'
# Matching variables are 'gifted_support' (received giftedness support yes
# or no) and 'enrichment' (participated in enrichment or not)
# 'MAGMA_sim_data_gift_enrich' contains the result of the matching
# 2x2 matching is equivalent to four-group matching
MAGMA_sim_data_gift_enrich <- MAGMA(Data = MAGMA_sim_data,
                                   group = c("gifted_support", "enrichment"),
                                   dist = "ps_2x2",
                                   cores = 2)


# Estimating balance. Covariates same as above
Balance_2x2 <- Balance_MAGMA(Data = MAGMA_sim_data_gift_enrich,
                             group = c("gifted_support", "enrichment"),
                             covariates = covariates_vector,
                             step = "step") # step created during matching
str(Balance_2x2)