R help - Aug 2010 - chisq.test on samples of different lengths

Hello,

I am trying to see whether there has been a significant difference in whether
people experienced damages from wildlife in two different years. I therefore
have two columns:

year 1:
yes
no
no
no
yes
yes
no

year 2:
no
yes
no
yes

I wanted to do a chisq.test, but if I enter it this way:

chisq.test(year1, year2)

I get the error saying the columns are two different lengths. So then I tried
doing:

damages<-matrix(c(3,4, 2,2), ncol=2, dimnames=list(answer=c("yes",
"no"), year=c("year1", year2)))
chisq.test(damages)

Does that make sense? Should I maybe be doing a different test instead?

Any help would be appreciated, thank you.

Agnese

On Aug 24, 2010, at 10:12 AM, Marino Taussig De Bodonia, Agnese wrote:
> Hello,
>
> I am trying to see whether there has been a significant difference  
> in whether people experienced damages from wildlife in two different  
> years. I therefore have two columns:
>
> year 1:
> yes
> no
> no
> no
> yes
> yes
> no
>
> year 2:
> no
> yes
> no
> yes
>
> I wanted to do a chisq.test, but if I enter it this way:
>
> chisq.test(year1, year2)
>
> I get the error saying the columns are two different lengths. So  
> then I tried doing:
>
> damages<-matrix(c(3,4, 2,2), ncol=2,
dimnames=list(answer=c("yes",
> "no"), year=c("year1", year2)))
> chisq.test(damages)
Which should throw an error because year2 is not quoted. Consider  
using prop.test:

?proptest

So your matrix is the transpose of what is needed for prop.test, at  
least as I read the docs:

 > damages<-matrix(c(3,4, 2,2), ncol=2, byrow=TRUE,  
dimnames=list(year=c("year1",
"year2"),success=c("yes", "no")))
 > damages
        success
year    yes no
   year1   3  4
   year2   2  2
 > prop.test(damages)

	2-sample test for equality of proportions with continuity correction

data:  damages
X-squared = 0, df = 1, p-value = 1
alternative hypothesis: two.sided
95 percent confidence interval:
  -0.7548099  0.6119528
sample estimates:
    prop 1    prop 2
0.4285714 0.5000000

Warning message:
In prop.test(damages) : Chi-squared approximation may be incorrect

>
> Does that make sense? Should I maybe be doing a different test  
> instead?
>
> Any help would be appreciated, thank you.
>
> Agnese
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
David Winsemius, MD
West Hartford, CT

On Aug 24, 2010, at 4:12 PM, Marino Taussig De Bodonia, Agnese wrote:
> Hello,
> 
> I am trying to see whether there has been a significant difference in
whether people experienced damages from wildlife in two different years. I
therefore have two columns:
> 
> year 1:
> yes
> no
> no
> no
> yes
> yes
> no
> 
> year 2:
> no
> yes
> no
> yes
> 
> I wanted to do a chisq.test, but if I enter it this way:
> 
> chisq.test(year1, year2)
> 
> I get the error saying the columns are two different lengths. So then I
tried doing:
> 
> damages<-matrix(c(3,4, 2,2), ncol=2,
dimnames=list(answer=c("yes", "no"),
year=c("year1", year2)))
> chisq.test(damages)
> 
> Does that make sense? Should I maybe be doing a different test instead?
The procedure is fine as such. A more automated way would be to 

mat <- cbind(table(year1),table(year2))
chisq.test(mat)

(some may prefer rbind(...), but the chi-square won't care)

The issue with the two-variable format is that it  expects cross-classifying
factors of the same individuals, not two independent groups. So you might do

answer <- c(year1,year2)
year <- rep(1:2, length(year1),length(year2))
table(answer, year) # just for enlightenment
chisq.test(answer, year)


Another matter is that you are below the usual rule of thumb for chi-square:
expected >5 obs in all 4 cells, which is obviously not going to happen with
10 observations in total. fisher.test is an option, but you need pretty extreme
configurations to obtain significance.

(BTW, all of the above assumes that there are no empty cells. Caveat emptor.)
> 
> Any help would be appreciated, thank you.
> 
> Agnese
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
-- 
Peter Dalgaard
Center for Statistics, Copenhagen Business School
Solbjerg Plads 3, 2000 Frederiksberg, Denmark
Phone: (+45)38153501
Email: pd.mes at cbs.dk  Priv: PDalgd at gmail.com