michaela: m i c h a e l a

In phoebehlam/michaela: convert statistical metrics for meta-analytic purpose

m i c h a e l a

description

michaela is beta. please report any bugs
a package for converting various stats metrics to correlation coefficient
simple functions aimed to be user-friendly for folks without r knowledge
all functions are in the format of "from.to"
e.g., from regression b and se to r: bse.r()
e.g., from odd ratio to r: or.r()

all functions are documented, use ?function to see usage, arguments, and examples (e.g., ?bse.r())

a note on converting unadjusted vs. adjusted metrics:
given bivariate metrics, all michaela functions can estimate bivariate pearson's correlation
given adjusted metrics, some michaela functions can estimate the partial correlation coefficient
please use with caution: if you extracted adjusted metrics, make sure the function can actually estimate partial correlation
for your convenience, equations that can take adjusted metrics are marked with *.

see references section below for a list of aloe and becker's work on synthesis of partial effect size and inaccuracies in replacing bivariate corr with regression results

correlational designs

*r.z(): r to fisher's z
*z.r(): fisher's z to r
*zvar(): variance of z
*rsq.r(): r squared to r (estimates semi-partial correlation instead of partial correlation)
*bse.t(): regression coefficient and standard error to t-statistics
*t.r(): t-statistics to r
*bse.r(): regression coefficient and standard error to r
*bci.r(): regression coefficient and confidence interval to r
*expb.r(): exponentiated logistic regression coefficient and confidence interval to r
*percent.r(): percent change derived from transformed exponentiated regression coefficients and confidence interval to r
f.r(): f-statistics (df1=1) to r
*regp.r(): regression p-values (or other t distribution-based p-values) to r

group difference designs

or.r(): odd ratio to r
rr.r(): risk ratio to r
meansd.d(): means and standard deviations of two groups to cohen's d
meanse.d(): means and standard errors of two groups to cohen's d
meanci.d(): means and confidence intervals of two groups to cohen's d
meansd.r(): means and standard deviations of two groups to r
meanse.r(): means and standard errors of two groups to r
meanci.r(): means and confidence intervals of two groups to r
geo.d(): geometric means and cis to cohen's d
geo.r(): geometric means and cis to r
d.t(): cohen's d to t-statistics
d.r(): cohen's d to r
median.r(): medians, minimums, and maximums to r

dichotomized/extreme group designs

use with caution, please refer and follow recommendations by pustejovsky (2014).
dicho.d.r(): cohen's d from dichotomizing/extreme group design to r
dicho.d.z(): cohen's d from dichotomizing/extreme group design to fisher's z
dichod.zvar(): variance of z from a cohen's d derived from dichotomizing/extreme group design
dicho.meansd.r(): means and sds from dichotomizing/extreme group design to r
dicho.meansd.z(): means and sd from dichotomizing/extreme group design to fisher's z
dicho.meansd.zvar(): variance of z converted from dichotomizing/extreme group design mean and sd
dicho.t.r(): t-statistics from dichotomizing/extreme group design to r
dicho.t.z(): t-statistics from dichotomizing/extreme group design to fisher's z
dicho.t.zvar(): variance of z derived from dichotomizing/extreme group design t-statistics
dicho.se.r(): mean and se from dichotomizing/extreme group design to r
dicho.se.z(): mean and se from dichotomizing/extreme group design to fisher's z
dicho.se.zvar(): variance of z derived from dichotomizing/extreme group design mean and se
dicho.meanci.r(): mean and ci from dichotomizing/extreme group design to r
dicho.meanci.z(): mean and ci from dichotomizing/extreme group design to fisher's z
dicho.meanci.zvar(): variance of z derived from dichotomizing/extreme group design mean and ci

auxillary conversions

*ci.se(): confidence interval to standard error
*sediff(): difference between standard errors computed using upper ci and lower ci
*geoci.se(): confidence interval around geometric mean to standard error of arithmetic mean of logged raw units
se.sd(): standard deviation to standard error
median.mean(): median, minimum, and maximum to estimated mean
median.sd(): median, minimum, and maximum to estimated standard deviation

references

re conversion of adjusted effect sizes and combining unadjusted and adjusted effect sizes, please refer and follow recommendations below

1. aloe, a. m., & thompson, c. g. (2013). the synthesis of partial effect sizes. journal of the society for social work and research, 4(4), 390-405.

2. aloe, a. m. (2015). inaccuracy of regression results in replacing bivariate correlations. research synthesis methods, 6(1), 21-27.

3. aloe, a. m., tanner‐smith, e. e., becker, b. j., & wilson, d. b. (2016). synthesizing bivariate and partial effect sizes. campbell systematic reviews, 12(1), 1-9.

4. becker, b. j., & wu, m. j. (2007). the synthesis of regression slopes in meta-analysis. statistical science, 22(3), 414-429.

phoebehlam/michaela documentation built on Oct. 23, 2024, 4:10 p.m.