Title: | Cluster Robust Wild Bootstrap Meta Regression |
---|---|
Description: | In meta regression sometimes the studies have multiple effects that are correlated. For this reason cluster robust standard errors must be computed. However, since the clusters are unbalanced the wild bootstrap is suggested. See Oczkowski E. and Doucouliagos H. (2015). "Wine prices and quality ratings: a meta-regression analysis". American Journal of Agricultural Economics, 97(1): 103--121. <doi:10.1093/ajae/aau057> and Cameron A. C., Gelbach J. B. and Miller D. L. (2008). "Bootstrap-based improvements for inference with clustered errors". The Review of Economics and Statistics, 90(3): 414--427. <doi:10.1162/rest.90.3.414>. |
Authors: | Michail Tsagris [aut, cre] |
Maintainer: | Michail Tsagris <[email protected]> |
License: | GPL (>= 2) |
Version: | 1.0 |
Built: | 2024-12-12 06:55:55 UTC |
Source: | CRAN |
In meta regression sometimes the studies have multiple effects that are correlated. For this reason cluster robust standard errors must be computed. However, since the clusters are unbalanced the wild bootstrap is suggested.
Package: | crwbmetareg |
Type: | Package |
Version: | 1.0 |
Date: | 2023-10-18 |
License: | GPL-2 |
Michail Tsagris [email protected].
Michail Tsagris [email protected].
Chatzimichael K., Daskalaki C., Emvalomatis G., Tsagris M. and Vangelis Tzouvelekas V. (2023). Factors Shaping Innovative Behavior: A Meta-Analysis of Technology Adoption Studies in Agriculture. https://economics.soc.uoc.gr/el/market/998/factors-shaping-farmers-innovative-behavior-a-meta-analysis-of-technology-adoption-studies-in-agriculture
Oczkowski, E. and Doucouliagos, H. (2015). Wine prices and quality ratings: a meta-regression analysis. American Journal of Agricultural Economics, 97(1): 103-121.
Cameron, A. C., Gelbach, J. B. and Miller, D. L. (2008). Bootstrap-based improvements for inference with clustered errors. The Review of Economics and Statistics, 90(3): 414-427.
Column-wise weighted least squares meta analysis.
colwlsmeta(yi, vi)
colwlsmeta(yi, vi)
yi |
A matrix with the observations. |
vi |
A matrix with the variances of the observations. |
The weighted least squares (WLS) meta analysis is performed in a column-wise fashion. This function is suitable for simulation studies, where one can perform multiple WLS meta analyses at once. See references for this.
A vector with many elements. The fixed effects mean estimate, the
estimate, the
, the
, the Q test statistic and it's p-value,
the
estimate and the random effects mean estimate.
Michail Tsagris.
R implementation and documentation: Michail Tsagris [email protected].
Stanley T. D. and Doucouliagos H. (2015). Neither fixed nor random: weighted least squares meta-analysis. Statistics in Medicine, 34(13), 2116–2127.
Stanley, T. D. and Doucouliagos, H. (2017). Neither fixed nor random: Weighted least squares meta-regression. Research synthesis methods, 8(1): 19–42.
y <- matrix( rnorm(50* 5), ncol = 5) vi <- matrix( rexp(50* 5), ncol = 5) colwlsmeta(y, vi) wlsmeta(y[, 1], vi[, 1])
y <- matrix( rnorm(50* 5), ncol = 5) vi <- matrix( rexp(50* 5), ncol = 5) colwlsmeta(y, vi) wlsmeta(y[, 1], vi[, 1])
FAT-PET test using cluster robust wild bootstrap.
fatpet(target, se, cluster, weights, boot.reps = 1000, prog.bar = FALSE, seed = NULL)
fatpet(target, se, cluster, weights, boot.reps = 1000, prog.bar = FALSE, seed = NULL)
target |
A vector with the effect sizes. |
se |
A vector with the standard errors, or the variances, of the effect sizes. |
cluster |
A vector indicating the clusters. |
weights |
A vector with the inverse of the the variances of the effect sizes. |
boot.reps |
The number of bootstrap re-samples to generate. |
prog.bar |
If you want the progress bar to appear set this equal to TRUE. |
seed |
IF you want the results to be rerpoducible set this equal to TRUE. |
It implements the FAT-PET test using cluster robust wild bootstrap to compute the p-values. See references for this.
The function uses a modification of the function "cluster.wild.glm()" of the package "clusterSEs".
A vector with two p-values. One for the constant and one for the cofficient of the "vse".
Michail Tsagris.
R implementation and documentation: Michail Tsagris [email protected].
Oczkowski, E. and Doucouliagos, H. (2015). Wine prices and quality ratings: a meta-regression analysis. American Journal of Agricultural Economics, 97(1): 103–121.
Cameron, A. C., Gelbach, J. B. and Miller, D. L. (2008). Bootstrap-based improvements for inference with clustered errors. The Review of Economics and Statistics, 90(3): 414–427.
y <- rnorm(50) se <- rexp(50, 3) cluster <- sample(1:20, 50, replace = TRUE) fatpet(y, se, cluster, weights = se^2, boot.reps = 500)
y <- rnorm(50) se <- rexp(50, 3) cluster <- sample(1:20, 50, replace = TRUE) fatpet(y, se, cluster, weights = se^2, boot.reps = 500)
Meta regression using cluster robust wild bootstrap.
crwbmetareg(target, se, dataset, cluster, weights, boot.reps = 1000, prog.bar = FALSE, seed = NULL)
crwbmetareg(target, se, dataset, cluster, weights, boot.reps = 1000, prog.bar = FALSE, seed = NULL)
target |
A vector with the effect sizes. |
se |
A vector with the standard errors, or the variances, of the effect sizes. |
dataset |
A matrix or data.frame with the independent variables. |
cluster |
A vector indicating the clusters. |
weights |
A vector with the inverse of the the variances of the effect sizes. |
boot.reps |
The number of bootstrap re-samples to generate. |
prog.bar |
If you want the progress bar to appear set this equal to TRUE. |
seed |
IF you want the results to be rerpoducible set this equal to TRUE. |
It implements metaregression using cluster robust wild bootstrap to compute the p-values. See references for this.
The function uses a modification of the function "cluster.wild.glm()" of the package "clusterSEs".
A vector with two p-values. One for the constant and one for the cofficient of the "se".
Michail Tsagris.
R implementation and documentation: Michail Tsagris [email protected].
Oczkowski, E. and Doucouliagos, H. (2015). Wine prices and quality ratings: a meta-regression analysis. American Journal of Agricultural Economics, 97(1): 103–121.
Cameron, A. C., Gelbach, J. B. and Miller, D. L. (2008). Bootstrap-based improvements for inference with clustered errors. The Review of Economics and Statistics, 90(3): 414–427.
y <- rnorm(50) se <- rexp(50, 3) cluster <- sample(1:20, 50, replace = TRUE) dataset <- matrix( rnorm(50 * 2), ncol = 2 ) fatpet(y, se, dataset, cluster, weights = se^2, boot.reps = 100)
y <- rnorm(50) se <- rexp(50, 3) cluster <- sample(1:20, 50, replace = TRUE) dataset <- matrix( rnorm(50 * 2), ncol = 2 ) fatpet(y, se, dataset, cluster, weights = se^2, boot.reps = 100)
Weighted least squares meta analysis.
wlsmeta(yi, vi)
wlsmeta(yi, vi)
yi |
The observations. |
vi |
The variances of the observations. |
It implements weighted least squares (WLS) meta analysis. See references for this.
A vector with many elements. The fixed effects mean estimate, the
estimate, the
, the
, the Q test statistic and it's p-value,
the
estimate and the random effects mean estimate.
Michail Tsagris.
R implementation and documentation: Michail Tsagris [email protected].
Stanley T. D. and Doucouliagos H. (2015). Neither fixed nor random: weighted least squares meta-analysis. Statistics in Medicine, 34(13): 2116–2127.
Stanley, T. D. and Doucouliagos, H. (2017). Neither fixed nor random: Weighted least squares meta-regression. Research synthesis methods, 8(1): 19–42.
y <- rnorm(30) vi <- rexp(30, 3) wlsmeta(y, vi)
y <- rnorm(30) vi <- rexp(30, 3) wlsmeta(y, vi)