John Fox
2021-Jan-19 23:17 UTC
[R] Different results on running Wilcoxon Rank Sum test in R and SPSS
Dear Bharat Rawlley,
What you tried to do appears to be nonsense. That is, you're treating
PFD_n and drug_code as if they were scores for two different groups.
I assume that what you really want to do is to treat PFD_n as a vector
of scores and drug_code as defining two groups. If that's correct, and
with your data into Data, you can try the following:
------snip ------
> wilcox.test(PFD_n ~ drug_code, data=Data, conf.int=TRUE)
Wilcoxon rank sum test with continuity correction
data: PFD_n by drug_code
W = 197, p-value = 0.05563
alternative hypothesis: true location shift is not equal to 0
95 percent confidence interval:
-2.000014e+00 5.037654e-05
sample estimates:
difference in location
-1.000019
Warning messages:
1: In wilcox.test.default(x = c(27, 26, 20, 24, 28, 28, 27, 27, 26, :
cannot compute exact p-value with ties
2: In wilcox.test.default(x = c(27, 26, 20, 24, 28, 28, 27, 27, 26, :
cannot compute exact confidence intervals with ties
------snip ------
You can get an approximate confidence interval by specifying exact=FALSE:
------snip ------
> wilcox.test(PFD_n ~ drug_code, data=Data, conf.int=TRUE, exact=FALSE)
Wilcoxon rank sum test with continuity correction
data: PFD_n by drug_code
W = 197, p-value = 0.05563
alternative hypothesis: true location shift is not equal to 0
95 percent confidence interval:
-2.000014e+00 5.037654e-05
sample estimates:
difference in location
-1.000019
------snip ------
As it turns out, your data are highly discrete and have a lot of ties
(see in particular PFD_n = 28):
------snip ------
> xtabs(~ PFD_n + drug_code, data=Data)
drug_code
PFD_n 0 1
0 2 0
16 1 1
18 0 1
19 0 1
20 2 0
22 0 1
24 2 0
25 1 2
26 5 2
27 4 2
28 5 13
30 1 2
------snip ------
I'm no expert in nonparametric inference, but I doubt whether the
approximate p-value will be very accurate for data like these.
I don't know why wilcox.test() (correctly used) and SPSS are giving you
slightly different results -- assuming that you're actually doing the
same thing in both cases. I couldn't help but notice that most of your
data are missing. Are you getting the same value of the test statistic
and different p-values, or is the test statistic different as well?
I hope this helps,
John
John Fox, Professor Emeritus
McMaster University
Hamilton, Ontario, Canada
web: https://socialsciences.mcmaster.ca/jfox/
On 2021-01-19 5:46 a.m., bharat rawlley via R-help
wrote:> Thank you for the reply and suggestion, Michael!
> I used dput() and this is the output I can share with you. Simply
explained, I have 3 columns namely, drug_code, freq4w_n and PFD_n. Each column
has 132 values (including NA). The problem with the Wilcoxon Rank Sum test has
been described in my first email.
> Please do let me know if you need any further clarification from my side!
Thanks a lot for your time!
> structure(list(drug_code = c(0, 0, 1, 1, 1, 0, 0, 0, 1, 1, 0,?1, 0, 1, 0,
1, 1, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0, 0, 0, 0, 1,?0, 1, 0, 0, 1, 1, 0, 1, 1, 1,
1, 1, 0, 0, 1, 0, 1, 0, 0, 0, 1,?1, 0, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 0, 1, 1,
0, 0, 0, 0, 1,?0, 0, 1, 1, 1, 0, 1, 1, 0, 1, 1, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0,?1,
1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 0, 1, 0, 0, 1,?1, 1, 1, 0, 0, 1, 0,
1, 0, 0, 1, 1, 1, 1, 0, 0), freq4w_n = c(1,?NA, NA, 0, NA, 4, NA, 10, NA, 0, 6,
NA, NA, NA, NA, NA, 10, NA,?0, NA, NA, NA, NA, 0, NA, 0, NA, NA, NA, 0, NA, 0,
NA, NA, NA,?NA, NA, NA, NA, NA, 0, 0, 12, 0, NA, 1, 2, 1, 2, 2, NA, 28, 0,?NA,
4, NA, 1, NA, NA, NA, NA, NA, 0, 3, 1, NA, NA, NA, NA, 4,?28, NA, NA, 0, 2, 12,
0, NA, NA, NA, 0, NA, 0, NA, NA, NA, NA,?NA, NA, NA, NA, NA, 3, NA, NA, NA, NA,
NA, NA, 6, 1, NA, NA,?NA, 0, NA, NA, NA, 0, 0, NA, 0, NA, 2, 8, 3, NA, NA, NA,
0, NA,?NA, NA, 9, NA, NA, NA, NA, NA, NA, NA, NA), PFD_n = c(27, NA,?NA, 28, NA,
26, NA, 20, NA, 30, 24, NA, NA, NA, NA, NA, 18, NA,?28, NA, NA, NA, NA, 28, NA,
28, NA, NA, NA, 28, NA, 28, NA, NA,?NA, NA, NA, NA, NA, NA, 28, 28, 16, 28, NA,
27, 26, 27, 26, 26,?NA, 0, 30, NA, 24, NA, 27, NA, NA, NA, NA, NA, 28, 25, 27,
NA,?NA, NA, NA, 26, 0, NA, NA, 28, 26, 16, 28, NA, NA, NA, 28, NA,?28, NA, NA,
NA, NA, NA, NA, NA, NA, NA, 25, NA, NA, NA, NA, NA,?NA, 22, 27, NA, NA, NA, 28,
NA, NA, NA, 28, 28, NA, 28, NA, 26,?20, 25, NA, NA, NA, 30, NA, NA, NA, 19, NA,
NA, NA, NA, NA, NA,?NA, NA)), row.names = c(NA, -132L), class =
c("tbl_df", "tbl",?"data.frame"))
>
> Yours sincerely?Bharat Rawlley On Tuesday, 19 January, 2021, 03:53:27 pm
IST, Michael Dewey <lists at dewey.myzen.co.uk> wrote:
>
> Unfortunately your data did not come through. Try using dput() and then
> pasting that into the body of your e-mail message.
>
> On 18/01/2021 17:26, bharat rawlley via R-help wrote:
>> Hello,
>> On running the Wilcoxon Rank Sum test in R and SPSS, I am getting the
following discrepancies which I am unable to explain.
>> Q1 In the attached data set, I was trying to compare freq4w_n in those
with drug_code 0 vs 1. SPSS gives a P value 0.031 vs R gives a P value 0.001779.
>> The code I used in R is as follows -
>> wilcox.test(freq4w_n, drug_code, conf.int = T)
>>
>>
>> Q2 Similarly, in the same data set, when trying to compare PFD_n in
those with drug_code 0 vs 1, SPSS gives a P value 0.038 vs R gives a P
value?< 2.2e-16.
>> The code I used in R is as follows -
>> wilcox.test(PFD_n, drug_code, mu = 0, alternative =
"two.sided", correct = TRUE, paired = FALSE, conf.int = TRUE)
>>
>>
>> I have tried searching on Google and watching some Youtube tutorials, I
cannot find an answer, Any help will be really appreciated, Thank you!
>> ______________________________________________
>> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
>> 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.
>>
>
bharat rawlley
2021-Jan-20 18:45 UTC
[R] Different results on running Wilcoxon Rank Sum test in R and SPSS
Dear Professor John,? Thank you very much for your reply!? I agree with you that the non-parametric tests I mentioned in my previous email (Moods median test and Median test) do not make sense in this situation as they treat PFD_n and drug_code as different groups. As you correctly said, I want to use PFD_n as a vector of scores and drug_code to make two groups out of it. This is exactly what the Independent samples median test does in SPSS. I wish to perform the same test in R and am unable to do so. Simply put, I am asking how to perform the Independent samples median test in R just like it is performed in SPSS?? Secondly, for the question you are asking about the test statistic, I have not performed the Wilcoxon Rank sum test in SPSS for the PFD_n and drug_code data. I have said something to the contrary in my first email, I apologize for that.? Thank you very much for your time!? Yours sincerelyBharat Rawlley On Wednesday, 20 January, 2021, 04:47:21 am IST, John Fox <jfox at mcmaster.ca> wrote: Dear Bharat Rawlley, What you tried to do appears to be nonsense. That is, you're treating PFD_n and drug_code as if they were scores for two different groups. I assume that what you really want to do is to treat PFD_n as a vector of scores and drug_code as defining two groups. If that's correct, and with your data into Data, you can try the following: ------snip ------ > wilcox.test(PFD_n ~ drug_code, data=Data, conf.int=TRUE) ??? Wilcoxon rank sum test with continuity correction data:? PFD_n by drug_code W = 197, p-value = 0.05563 alternative hypothesis: true location shift is not equal to 0 95 percent confidence interval: ? -2.000014e+00? 5.037654e-05 sample estimates: difference in location ? ? ? ? ? ? ? -1.000019 Warning messages: 1: In wilcox.test.default(x = c(27, 26, 20, 24, 28, 28, 27, 27, 26,? : ? cannot compute exact p-value with ties 2: In wilcox.test.default(x = c(27, 26, 20, 24, 28, 28, 27, 27, 26,? : ? cannot compute exact confidence intervals with ties ------snip ------ You can get an approximate confidence interval by specifying exact=FALSE: ------snip ------ > wilcox.test(PFD_n ~ drug_code, data=Data, conf.int=TRUE, exact=FALSE) ??? Wilcoxon rank sum test with continuity correction data:? PFD_n by drug_code W = 197, p-value = 0.05563 alternative hypothesis: true location shift is not equal to 0 95 percent confidence interval: ? -2.000014e+00? 5.037654e-05 sample estimates: difference in location ? ? ? ? ? ? ? -1.000019 ------snip ------ As it turns out, your data are highly discrete and have a lot of ties (see in particular PFD_n = 28): ------snip ------ > xtabs(~ PFD_n + drug_code, data=Data) ? ? ? drug_code PFD_n? 0? 1 ? ? 0? 2? 0 ? ? 16? 1? 1 ? ? 18? 0? 1 ? ? 19? 0? 1 ? ? 20? 2? 0 ? ? 22? 0? 1 ? ? 24? 2? 0 ? ? 25? 1? 2 ? ? 26? 5? 2 ? ? 27? 4? 2 ? ? 28? 5 13 ? ? 30? 1? 2 ------snip ------ I'm no expert in nonparametric inference, but I doubt whether the approximate p-value will be very accurate for data like these. I don't know why wilcox.test() (correctly used) and SPSS are giving you slightly different results -- assuming that you're actually doing the same thing in both cases. I couldn't help but notice that most of your data are missing. Are you getting the same value of the test statistic and different p-values, or is the test statistic different as well? I hope this helps, ? John John Fox, Professor Emeritus McMaster University Hamilton, Ontario, Canada web: https://socialsciences.mcmaster.ca/jfox/ On 2021-01-19 5:46 a.m., bharat rawlley via R-help wrote:>? Thank you for the reply and suggestion, Michael! > I used dput() and this is the output I can share with you. Simply explained, I have 3 columns namely, drug_code, freq4w_n and PFD_n. Each column has 132 values (including NA). The problem with the Wilcoxon Rank Sum test has been described in my first email. > Please do let me know if you need any further clarification from my side! Thanks a lot for your time! > structure(list(drug_code = c(0, 0, 1, 1, 1, 0, 0, 0, 1, 1, 0,?1, 0, 1, 0, 1, 1, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0, 0, 0, 0, 1,?0, 1, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 0, 0, 1, 0, 1, 0, 0, 0, 1,?1, 0, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 0, 1, 1, 0, 0, 0, 0, 1,?0, 0, 1, 1, 1, 0, 1, 1, 0, 1, 1, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0,?1, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 0, 1, 0, 0, 1,?1, 1, 1, 0, 0, 1, 0, 1, 0, 0, 1, 1, 1, 1, 0, 0), freq4w_n = c(1,?NA, NA, 0, NA, 4, NA, 10, NA, 0, 6, NA, NA, NA, NA, NA, 10, NA,?0, NA, NA, NA, NA, 0, NA, 0, NA, NA, NA, 0, NA, 0, NA, NA, NA,?NA, NA, NA, NA, NA, 0, 0, 12, 0, NA, 1, 2, 1, 2, 2, NA, 28, 0,?NA, 4, NA, 1, NA, NA, NA, NA, NA, 0, 3, 1, NA, NA, NA, NA, 4,?28, NA, NA, 0, 2, 12, 0, NA, NA, NA, 0, NA, 0, NA, NA, NA, NA,?NA, NA, NA, NA, NA, 3, NA, NA, NA, NA, NA, NA, 6, 1, NA, NA,?NA, 0, NA, NA, NA, 0, 0, NA, 0, NA, 2, 8, 3, NA, NA, NA, 0, NA,?NA, NA, 9, NA, NA, NA, NA, NA, NA, NA, NA), PFD_n = c(27, NA,?NA, 28, NA, 26, NA, 20, NA, 30, 24, NA, NA, NA, NA, NA, 18, NA,?28, NA, NA, NA, NA, 28, NA, 28, NA, NA, NA, 28, NA, 28, NA, NA,?NA, NA, NA, NA, NA, NA, 28, 28, 16, 28, NA, 27, 26, 27, 26, 26,?NA, 0, 30, NA, 24, NA, 27, NA, NA, NA, NA, NA, 28, 25, 27, NA,?NA, NA, NA, 26, 0, NA, NA, 28, 26, 16, 28, NA, NA, NA, 28, NA,?28, NA, NA, NA, NA, NA, NA, NA, NA, NA, 25, NA, NA, NA, NA, NA,?NA, 22, 27, NA, NA, NA, 28, NA, NA, NA, 28, 28, NA, 28, NA, 26,?20, 25, NA, NA, NA, 30, NA, NA, NA, 19, NA, NA, NA, NA, NA, NA,?NA, NA)), row.names = c(NA, -132L), class = c("tbl_df", "tbl",?"data.frame")) > > Yours sincerely?Bharat Rawlley? ? On Tuesday, 19 January, 2021, 03:53:27 pm IST, Michael Dewey <lists at dewey.myzen.co.uk> wrote: >? >? Unfortunately your data did not come through. Try using dput() and then > pasting that into the body of your e-mail message. > > On 18/01/2021 17:26, bharat rawlley via R-help wrote: >> Hello, >> On running the Wilcoxon Rank Sum test in R and SPSS, I am getting the following discrepancies which I am unable to explain. >> Q1 In the attached data set, I was trying to compare freq4w_n in those with drug_code 0 vs 1. SPSS gives a P value 0.031 vs R gives a P value 0.001779. >> The code I used in R is as follows - >> wilcox.test(freq4w_n, drug_code, conf.int = T) >> >> >> Q2 Similarly, in the same data set, when trying to compare PFD_n in those with drug_code 0 vs 1, SPSS gives a P value 0.038 vs R gives a P value?< 2.2e-16. >> The code I used in R is as follows - >> wilcox.test(PFD_n, drug_code, mu = 0, alternative = "two.sided", correct = TRUE, paired = FALSE, conf.int = TRUE) >> >> >> I have tried searching on Google and watching some Youtube tutorials, I cannot find an answer, Any help will be really appreciated, Thank you! >> ______________________________________________ >> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see >> 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. >> >[[alternative HTML version deleted]]