Skip to contents

Publication-Ready Covariate Balance Table

balance_table.Rd

Produces a publication-ready covariate balance table comparing means, standard deviations, standardized mean differences (SMD), and variance ratios between treated and control groups. Optionally returns a gt or knitr formatted table.

Usage

balance_table(
  data,
  treatment,
  covariates = NULL,
  weights = NULL,
  smd_threshold = 0.1,
  digits = 3,
  output_format = c("df", "kable", "gt"),
  caption = "Covariate Balance Table"
)

Arguments

data

A data frame containing the covariates and treatment indicator.

treatment

Character. Name of the binary treatment indicator column (0/1 or TRUE/FALSE).

covariates

Character vector. Names of covariate columns to include. If NULL (default), all numeric and logical columns except treatment are used.

weights

Numeric vector of length nrow(data). Optional analytical weights (e.g., inverse probability weights).

smd_threshold

Numeric. Threshold for flagging imbalanced covariates (|SMD| > threshold). Default 0.1.

digits

Integer. Number of decimal places. Default 3.

output_format

Character. Table format: "df" (data frame, default), "kable" (knitr table), or "gt" (gt table).

caption

Character. Table caption for formatted outputs.

Value

A data frame (when output_format = "df") or formatted table with columns:

covariate

Covariate name.

mean_control

Weighted mean in control group.

mean_treated

Weighted mean in treated group.

sd_control

Standard deviation in control group.

sd_treated

Standard deviation in treated group.

smd

Standardized mean difference.

var_ratio

Variance ratio (treated/control).

flag

Logical. Whether |SMD| exceeds smd_threshold.

Details

The standardized mean difference (SMD) is computed as: $$SMD = \frac{\bar{X}_t - \bar{X}_c}{\sqrt{(s_t^2 + s_c^2)/2}}$$ following @rubin2001using. Values below 0.1 are conventionally considered well-balanced.

References

Rubin, D. B. (2001). Using propensity scores to help design observational studies: application to the tobacco litigation. Health Services and Outcomes Research Methodology, 2(3–4), 169–188.

Examples

data(lalonde, package = "MatchIt")
balance_table(
  data = lalonde,
  treatment = "treat",
  covariates = c("age", "educ", "re74", "re75")
)
#>   covariate mean_control mean_treated sd_control sd_treated    smd var_ratio
#> 1       age       28.030       25.816     10.774      7.136 -0.242     0.439
#> 2      educ       10.235       10.346      2.852      2.005  0.045     0.494
#> 3      re74     5619.237     2095.574   6780.834   4873.395 -0.597     0.517
#> 4      re75     2466.484     1532.055   3288.157   3210.538 -0.288     0.953
#>    flag
#> 1  TRUE
#> 2 FALSE
#> 3  TRUE
#> 4  TRUE