pred_int: Compute prediction interval for replication study given...

In Replicate: Statistical Metrics for Multisite Replication Studies

Description

Given point estimates and their variances for one or multiple original studies and one or more replication studies, returns a vector stating whether each replication estimate is in its corresponding prediction interval. Assumes no heterogeneity.

Usage

pred_int(yio, vio, yir = NULL, vir, level = 0.95)

Arguments

yio	Effect estimate in the original study. Can be a vector for multiple original studies.
vio	Estimated variance of effect estimate in the original study (i.e., its squared standard error). Can be a vector for multiple original studies.
yir	Effect estimate in the replication study. Can be a vector for multiple replication studies. Can be omitted, in which case function returns only the prediction interval.
vir	Estimated variance of effect estimate in the replication study (i.e., its squared standard error). Can be a vector for multiple replication studies.
level	Coverage level of prediction interval. Typically 0.95.

Examples

# calculate prediction interval for a single replication study
pred_int( yio = 1, vio = .5, yir = 0.6,
vir = .2 )

# calculate prediction intervals for a one-to-one design
pred_int( yio = c(1, 1.3), vio = c(.01, .6),
yir = c(.6, .7), vir = c(.01,.3) )

# no need to pass yir if you only want the intervals
pred_int( yio = c(1, 1.3), vio = c(.01, .6),
vir = c(.01,.3) )

# calculate prediction intervals for a many-to-one design
pred_int( yio = c(1), vio = c(.01), yir = c(.6, .7), vir = c(.01,.3) )

Example output

$int.lo
[1] -0.6398235

$int.hi
[1] 2.639824

$rep.inside
[1] TRUE

$int.lo
[1]  0.7228192 -0.5593851

$int.hi
[1] 1.277181 3.159385

$rep.inside
[1] FALSE  TRUE

$int.lo
[1]  0.7228192 -0.5593851

$int.hi
[1] 1.277181 3.159385

$rep.inside
[1] NA

$int.lo
[1]  0.72281924 -0.09126176

$int.hi
[1] 1.277181 2.091262

$rep.inside
[1] FALSE  TRUE

Replicate documentation built on Dec. 1, 2019, 1:14 a.m.