After statistical analysis, how can i show the significant differences (S.D) in bar plots? In SPSS 16 and Minitab 15 there is no such function ( to show any symbol above the bar). I've analysed the data (ANOVA), but the problem was to show the S.D. in chart, because I couldn't find any automated S/W to denote symbols above the bars.
In R you can get such a plot by
values = c(8,6,4,7,4,7.5)
errors = c(1,1,0.5,0.5,1,1)
years = c("1996","1997")
habitats = c("Temporary","Permanent","Lake")
symbols = c("A","B","B","C","AC","AC")
colors = c("black","white")
x = matrix(values,nrow=2,dimnames=list(years,habitats))
se = matrix(errors,nrow=2)
v = barplot(x,beside=TRUE,las=1,xlab="Habitat",ylab="Mean Percent Hatch",ylim=c(0,max(x+2*se)), col=colors)
arrows(v,x,v,x+se,code=2,angle=90)
xpos = grconvertX(0.5,"ndc","user")
ypos = grconvertY(1,"ndc","user")
legend(xpos,ypos,legend=rownames(x),fill=colors,bty="n",cex=1.2,xpd=NA,xjust=0.5,yjust=1)
text(v,x+se,labels=symbols,cex=1.2,adj=c(0.5,-0.5))
In R you can get such a plot by
values = c(8,6,4,7,4,7.5)
errors = c(1,1,0.5,0.5,1,1)
years = c("1996","1997")
habitats = c("Temporary","Permanent","Lake")
symbols = c("A","B","B","C","AC","AC")
colors = c("black","white")
x = matrix(values,nrow=2,dimnames=list(years,habitats))
se = matrix(errors,nrow=2)
v = barplot(x,beside=TRUE,las=1,xlab="Habitat",ylab="Mean Percent Hatch",ylim=c(0,max(x+2*se)), col=colors)
arrows(v,x,v,x+se,code=2,angle=90)
xpos = grconvertX(0.5,"ndc","user")
ypos = grconvertY(1,"ndc","user")
legend(xpos,ypos,legend=rownames(x),fill=colors,bty="n",cex=1.2,xpd=NA,xjust=0.5,yjust=1)
text(v,x+se,labels=symbols,cex=1.2,adj=c(0.5,-0.5))
You have to manually prepare the pairwise comparision then go for Excel presentation you can show the value above bars toooooooooooo
Sunil, could you let on about how this can be achieved in Excel? I found no easy way to let Excel draw the symbols (e.g. as labels of another data series). Of course, one can manually add them - but that was not the question (a manual finishing of a figure, be it in Excel, a statistic software or a image editor (Photoshop or even Powerpoint), is anyway always possible). I would be happy to know a tolerably simple solution for Excel.
Dear Wilhelm, I was confused by pic given in the question where values were above the bar, which is possible in excel but after seeing your question I saw the pic very carefully and it is the alphabetic letters so I am sorry for my answer................I have not done that by using EXCEl it can be done by manual editing as you mentioned......
I you have a large number of group, the task the most time consuming will be to assess the letter coding, rather than to plot the letters.
If you are interested in an automatic way, have a look to the insert-absorb algorithm from Piepho 2004
in SAS:
http://agrobiol.sggw.waw.pl/~cbcs/articles/CBCS_7_1_2.pdf
in R, long time ago I hacked the code below from a package I don't remember the name. Sorry for the authors. This code implements the same insert-absorb algorithm
# Function implements the insert-absorb (sweep) heuristic of Piepho 2004:
# "An Algorithm for a Letter-Based Representation of All-Pairwise Comparisons"
#
# x ... vector of logicals indicating significant comparisons with hyphenated
# names e.g. A-B, treatmentA-treatmentB, ...
# Letters ... a set of user defined letters { default is Letters=c(letters, LETTERS) }
# separator ... a separating character used to produce a sufficiently large set of
# characters for a compact letter display (default is separator=".") in case
# the number of letters required exceeds the number of letters available
# Decreasing ... Inverse the order of the letters
insert_absorb
Hey Pierre,
Can you maybe provide me with an example of how to use the R code above?
No need to give the functions again, but maybe provide an example of a usable data set? and an example of how to call the functions?
Thanks in advance, it would be great if I would get this to work!!
I think I figured out what the data vector should look like, e.g.:
signdata
- Badges
- Science topic