metafor (version 0.5-0)

plot.rma.uni.infl: Plot Method for rma.uni.infl Objects

Description

Plot method for objects of class "rma.uni.infl".

Usage

## S3 method for class 'rma.uni.infl':
plot(x, plotdfb=FALSE, pch=21, bg="black", 
     bg.infl="red", col.na="lightgray", ...)

Arguments

x

an object of class "rma.uni.infl".

plotdfb

logical indicating whether the DFBETAS values should be plotted (default is FALSE).

pch

plotting symbol to use. By default, a filled circle is used. See points for other options.

bg

color to use for filling the plotting symbol (default is "black").

bg.infl

color to use for filling the plotting symbol when the point is considered influential (default is "red").

col.na

color to use for lines connecting two points with NAs in between (default is "lightgray").

...

other arguments.

Details

The function plots the externally standardized residuals, DFFITS values, Cook's distances, covariance ratios, leave-one-out $\tau^2$ estimates, leave-one-out (residual) heterogeneity test statistics, hat values, and weights. If plotdfb=TRUE, the DFBETAS values are plotted after confirming the page change. Points are considered influential if at least one of the following is true:

• The absolute DFFITS value is larger than$3 \sqrt{p/(k-p)}$, where$p$is the number of model coefficients and$k$the number of studies.
• The lower tail area of a chi-square distribution with$p$degrees of freedom cut off by the Cook's distance is larger than$1/2$.
• The hat value is larger than$3(p/k)$.
• Any DFBETAS value is larger than$1$.

These cut-offs are also indicated in the plot with horizontal reference lines. In addition, on the plot of the externally standardized residuals, horizontal reference lines are drawn at -1.96, 0, and 1.96. On the plot of the hat values, a horizontal reference line is drawn at $p/k$. Since the sum of the hat values is equal to $p$, the value $p/k$ indicates equal hat values for all $k$ studies. Finally, on the plot of weights, a horizontal reference line is drawn at $100/k$, corresponding to the value for equal weights (in %) for all $k$ studies. Note that all weights will automatically be equal to each other when using unweighted model fitting. Also, except for their scaling, the hat values and weights will be equal to each other in models without moderators. The chosen cut-offs are (somewhat) arbitrary. Substantively informed judgment should always be used when examining the influence of each case on the results.

See Also

influence.rma.uni

Examples

### load BCG vaccine data
data(dat.bcg)

### meta-analysis of the log risk rates using a mixed-effects model
### with two moderators (absolute latitude and publication year)
res <- rma(ai=tpos, bi=tneg, ci=cpos, di=cneg, mods=cbind(ablat, year),
           data=dat.bcg, measure="RR", method="REML")
plot(influence(res))