I'm using multiple regressions to determine relationships between my DV and each of my IV. Therefore, I compute a ß slope for every single relationship. I would like to ground my interpretation of these effects based on "The New Statistics" (Cumming, 2012), and not only calculate 95% Confidence Intervals on these slopes (which so far isn't a big deal), but also to plot my Confidence Intervals on a graph in order to have a meaningfull visual representation of these. Could you advise any particular script, function, on package in R likely to help me ?
I would use the package ggplot2. The function geom_errorbar(aes(ymin =LowerCI, ymax = UpperCI) within a ggplot is practiible, if you have already calculated the confidence interval. Additionally points, graphs, legend ect. can be plotted.
plot(y~x,type="n")
m = lm(y~x)
wx = par("usr")[1:2]
new.x = seq(wx[1],wx[2],len=100)
pred = predict(m, new=data.frame(x=new.x), interval="conf")
lines(new.x,pred[,"fit"],lwd=2)
lines(new.x,pred[,"lwr"],lty=3)
lines(new.x,pred[,"upr"],lty=3)
points(x,y,pch=16,col="steelblue")
### alternative: plot confidence band:
plot(y~x,type="n")
polygon(c(new.x,rev(new.x)),c(pred[,"lwr"],rev(pred[,"upr"])),border=NA,col=blues9[3])
lines(new.x,pred[,"fit"],lwd=2,col=blues9[8])
points(x,y,pch=16)
box()
Use this code, it works for me. I got from R help link.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R Statistical Package