ReplicationSuccess (version 1.3.3)

protzko2020: Data from Protzko et al. (2020)

Description

Data from "High Replicability of Newly-Discovered Social-behavioral Findings is Achievable" by Protzko et al. (2020). The variables are as follows:

experiment

Experiment name

type

Type of study, either "original", "self-replication", or "external-replication"

lab

The lab which conducted the study, either 1, 2, 3, or 4.

smd

Standardized mean difference effect estimate

se

Standard error of standardized mean difference effect estimate

n

Total sample size of the study

Usage

data("protzko2020")

Format

A data frame with 80 rows and 6 variables

Details

This data set originates from a prospective replication project involving four laboratories. Each of them conducted four original studies and for each original study a replication study was carried out within the same lab (self-replication) and by the other three labs (external-replication). Most studies used simple between-subject designs with two groups and a continuous outcome so that for each study, an estimate of the standardized mean difference (SMD) could be computed from the group means, group standard deviations, and group sample sizes. For studies with covariate adjustment and/or binary outcomes, effect size transformations as described in the supplementary material of Protzko (2020) were used to obtain effect estimates and standard errors on SMD scale. The data set is licensed under a CC-By Attribution 4.0 International license, see https://creativecommons.org/licenses/by/4.0/ for the terms of reuse.

References

Protzko, J., Krosnick, J., Nelson, L. D., Nosek, B. A., Axt, J., Berent, M., ... Schooler, J. (2020, September 10). High Replicability of Newly-Discovered Social-behavioral Findings is Achievable. tools:::Rd_expr_doi("10.31234/osf.io/n2a9x")

Protzko, J., Berent, M., Buttrick, N., DeBell, M., Roeder, S. S., Walleczek, J., ... Nosek, B. A. (2021, January 5). Results & Data. Retrieved from https://osf.io/42ef9/

Examples

data("protzko2020", package = "ReplicationSuccess")

## forestplots of effect estimates
graphics.off()
parOld <- par(mar = c(5, 8, 4, 2), mfrow = c(4, 4))
experiments <- unique(protzko2020$experiment)
for (ex in experiments) {
  ## compute CIs
  dat <- subset(protzko2020, experiment == ex)
  za <- qnorm(p = 0.975)
  plotDF <- data.frame(lower = dat$smd - za*dat$se,
                       est = dat$smd,
                       upper = dat$smd + za*dat$se)
colpalette <- c("#000000", "#1B9E77", "#D95F02")
cols <- colpalette[dat$type]
yseq <- seq(1, nrow(dat))

## forestplot
plot(x = plotDF$est, y = yseq, xlim = c(-0.15, 0.8),
     ylim = c(0.8*min(yseq), 1.05*max(yseq)), type = "n",
     yaxt = "n", xlab = "Effect estimate (SMD)", ylab = "")
abline(v = 0, col = "#0000004D")
arrows(x0 = plotDF$lower, x1 = plotDF$upper, y0 = yseq, angle = 90,
       code = 3, length = 0.05, col = cols)
points(y = yseq, x = plotDF$est, pch = 20, lwd = 2, col = cols)
axis(side = 2, at = yseq, las = 1, labels = dat$type, cex.axis = 0.85)
title(main = ex)
}
par(parOld)