Dear community,

I am trying to compare 2 groups of samples with a Mann-Whitney U test. In previous tests, I compared the same samples, grouped according to another parameter, and everything worked out. Now, when I try to follow the same procedure (Analyze > Nonparametric Tests > Independent Samples), SPSS performs automatically a Kruskal-Wallis test. When I select the Mann-Whitney test manually, I receive the notification that the results could not be computed.

What could be the reason for this? Even though I can also work with the results from Kruskal-Wallis, I would still like to understand why Mann-Whitney does not work in my case and what will be the implications for data interpretation (in terms of how reliable my conclusions will be).

I would be grateful for support!

------------------------------
Simone Trimmel
------------------------------

#SPSSStatistics

Hello, Simone. Both of these worked for me:

NPAR TESTS /M-W= y BY group(0 1).

NPTESTS
 /INDEPENDENT TEST (y) GROUP (group)
 /MISSING SCOPE=ANALYSIS USERMISSING=EXCLUDE
 /CRITERIA ALPHA=0.05 CILEVEL=95.

Could you substitute your variable for "y" and your grouping variable for "group" and try:

VARIABLE LEVEL y (SCALE) group (ORDINAL).

NPTESTS
 /INDEPENDENT TEST (y) GROUP (group)
 /MISSING SCOPE=ANALYSIS USERMISSING=EXCLUDE
 /CRITERIA ALPHA=0.05 CILEVEL=95.

------------------------------
Rick Marcantonio
Quality Assurance
IBM
------------------------------

Original Message:
Sent: Thu September 22, 2022 02:09 AM
From: Simone Trimmel
Subject: SPSS Statistics (v28) performs Kruskal-Wallis instead of Mann-Whitney U test

Dear community,

I am trying to compare 2 groups of samples with a Mann-Whitney U test. In previous tests, I compared the same samples, grouped according to another parameter, and everything worked out. Now, when I try to follow the same procedure (Analyze > Nonparametric Tests > Independent Samples), SPSS performs automatically a Kruskal-Wallis test. When I select the Mann-Whitney test manually, I receive the notification that the results could not be computed.

What could be the reason for this? Even though I can also work with the results from Kruskal-Wallis, I would still like to understand why Mann-Whitney does not work in my case and what will be the implications for data interpretation (in terms of how reliable my conclusions will be).

I would be grateful for support!

------------------------------
Simone Trimmel
------------------------------

#SPSSStatistics

Dear Rick,
Thank you very much for your reply!
I have 48 variables - do I need to copy the code 48 times in this case? I tried to perform the test without using the syntax, as it is easier to select the variables from the menu than typing the names of all of them.

------------------------------
Simone Trimmel
------------------------------

Original Message:
Sent: Thu September 22, 2022 01:06 PM
From: Rick Marcantonio
Subject: SPSS Statistics (v28) performs Kruskal-Wallis instead of Mann-Whitney U test

Hello, Simone. Both of these worked for me:

NPAR TESTS /M-W= y BY group(0 1).

NPTESTS
 /INDEPENDENT TEST (y) GROUP (group)
 /MISSING SCOPE=ANALYSIS USERMISSING=EXCLUDE
 /CRITERIA ALPHA=0.05 CILEVEL=95.

Could you substitute your variable for "y" and your grouping variable for "group" and try:

VARIABLE LEVEL y (SCALE) group (ORDINAL).

NPTESTS
 /INDEPENDENT TEST (y) GROUP (group)
 /MISSING SCOPE=ANALYSIS USERMISSING=EXCLUDE
 /CRITERIA ALPHA=0.05 CILEVEL=95.

------------------------------
Rick Marcantonio
Quality Assurance
IBM

Original Message:
Sent: Thu September 22, 2022 02:09 AM
From: Simone Trimmel
Subject: SPSS Statistics (v28) performs Kruskal-Wallis instead of Mann-Whitney U test

Dear community,

I am trying to compare 2 groups of samples with a Mann-Whitney U test. In previous tests, I compared the same samples, grouped according to another parameter, and everything worked out. Now, when I try to follow the same procedure (Analyze > Nonparametric Tests > Independent Samples), SPSS performs automatically a Kruskal-Wallis test. When I select the Mann-Whitney test manually, I receive the notification that the results could not be computed.

What could be the reason for this? Even though I can also work with the results from Kruskal-Wallis, I would still like to understand why Mann-Whitney does not work in my case and what will be the implications for data interpretation (in terms of how reliable my conclusions will be).

I would be grateful for support!

------------------------------
Simone Trimmel
------------------------------

#SPSSStatistics

You can try something like:

NPAR TESTS /M-W= y1 TO y48 BY group(0 1).

if the variables all appear in sequence in the file, or

NPAR TESTS /M-W= <my variables all listed separately> BY group(0 1).

if they don't.

------------------------------
Rick Marcantonio
Quality Assurance
IBM
------------------------------

Original Message:
Sent: Wed September 28, 2022 02:19 AM
From: Simone Trimmel
Subject: SPSS Statistics (v28) performs Kruskal-Wallis instead of Mann-Whitney U test

Dear Rick,
Thank you very much for your reply!
I have 48 variables - do I need to copy the code 48 times in this case? I tried to perform the test without using the syntax, as it is easier to select the variables from the menu than typing the names of all of them.

------------------------------
Simone Trimmel

Original Message:
Sent: Thu September 22, 2022 01:06 PM
From: Rick Marcantonio
Subject: SPSS Statistics (v28) performs Kruskal-Wallis instead of Mann-Whitney U test

Hello, Simone. Both of these worked for me:

NPAR TESTS /M-W= y BY group(0 1).

NPTESTS
 /INDEPENDENT TEST (y) GROUP (group)
 /MISSING SCOPE=ANALYSIS USERMISSING=EXCLUDE
 /CRITERIA ALPHA=0.05 CILEVEL=95.

Could you substitute your variable for "y" and your grouping variable for "group" and try:

VARIABLE LEVEL y (SCALE) group (ORDINAL).

NPTESTS
 /INDEPENDENT TEST (y) GROUP (group)
 /MISSING SCOPE=ANALYSIS USERMISSING=EXCLUDE
 /CRITERIA ALPHA=0.05 CILEVEL=95.

------------------------------
Rick Marcantonio
Quality Assurance
IBM

Original Message:
Sent: Thu September 22, 2022 02:09 AM
From: Simone Trimmel
Subject: SPSS Statistics (v28) performs Kruskal-Wallis instead of Mann-Whitney U test

Dear community,

I am trying to compare 2 groups of samples with a Mann-Whitney U test. In previous tests, I compared the same samples, grouped according to another parameter, and everything worked out. Now, when I try to follow the same procedure (Analyze > Nonparametric Tests > Independent Samples), SPSS performs automatically a Kruskal-Wallis test. When I select the Mann-Whitney test manually, I receive the notification that the results could not be computed.

What could be the reason for this? Even though I can also work with the results from Kruskal-Wallis, I would still like to understand why Mann-Whitney does not work in my case and what will be the implications for data interpretation (in terms of how reliable my conclusions will be).

I would be grateful for support!

------------------------------
Simone Trimmel
------------------------------

#SPSSStatistics

Dear Rick,
Thank you for your support. My issue is solved now - I sent my datafile to Jon and he discovered that there were blank spaces in my file which were counted as a third group. I deleted them and now I'm able to perform a Mann-Whitney U test.

------------------------------
Simone Trimmel
------------------------------

Original Message:
Sent: Wed September 28, 2022 11:27 AM
From: Rick Marcantonio
Subject: SPSS Statistics (v28) performs Kruskal-Wallis instead of Mann-Whitney U test

You can try something like:

NPAR TESTS /M-W= y1 TO y48 BY group(0 1).

if the variables all appear in sequence in the file, or

NPAR TESTS /M-W= <my variables all listed separately> BY group(0 1).

if they don't.

------------------------------
Rick Marcantonio
Quality Assurance
IBM

Original Message:
Sent: Wed September 28, 2022 02:19 AM
From: Simone Trimmel
Subject: SPSS Statistics (v28) performs Kruskal-Wallis instead of Mann-Whitney U test

Dear Rick,
Thank you very much for your reply!
I have 48 variables - do I need to copy the code 48 times in this case? I tried to perform the test without using the syntax, as it is easier to select the variables from the menu than typing the names of all of them.

------------------------------
Simone Trimmel

Original Message:
Sent: Thu September 22, 2022 01:06 PM
From: Rick Marcantonio
Subject: SPSS Statistics (v28) performs Kruskal-Wallis instead of Mann-Whitney U test

Hello, Simone. Both of these worked for me:

NPAR TESTS /M-W= y BY group(0 1).

NPTESTS
 /INDEPENDENT TEST (y) GROUP (group)
 /MISSING SCOPE=ANALYSIS USERMISSING=EXCLUDE
 /CRITERIA ALPHA=0.05 CILEVEL=95.

Could you substitute your variable for "y" and your grouping variable for "group" and try:

VARIABLE LEVEL y (SCALE) group (ORDINAL).

NPTESTS
 /INDEPENDENT TEST (y) GROUP (group)
 /MISSING SCOPE=ANALYSIS USERMISSING=EXCLUDE
 /CRITERIA ALPHA=0.05 CILEVEL=95.

------------------------------
Rick Marcantonio
Quality Assurance
IBM

Original Message:
Sent: Thu September 22, 2022 02:09 AM
From: Simone Trimmel
Subject: SPSS Statistics (v28) performs Kruskal-Wallis instead of Mann-Whitney U test

Dear community,

I am trying to compare 2 groups of samples with a Mann-Whitney U test. In previous tests, I compared the same samples, grouped according to another parameter, and everything worked out. Now, when I try to follow the same procedure (Analyze > Nonparametric Tests > Independent Samples), SPSS performs automatically a Kruskal-Wallis test. When I select the Mann-Whitney test manually, I receive the notification that the results could not be computed.

What could be the reason for this? Even though I can also work with the results from Kruskal-Wallis, I would still like to understand why Mann-Whitney does not work in my case and what will be the implications for data interpretation (in terms of how reliable my conclusions will be).

I would be grateful for support!

------------------------------
Simone Trimmel
------------------------------

#SPSSStatistics

Original Message:
Sent: 9/22/2022 2:09:00 AM
From: Simone Trimmel
Subject: SPSS Statistics (v28) performs Kruskal-Wallis instead of Mann-Whitney U test

Dear community,

I am trying to compare 2 groups of samples with a Mann-Whitney U test. In previous tests, I compared the same samples, grouped according to another parameter, and everything worked out. Now, when I try to follow the same procedure (Analyze > Nonparametric Tests > Independent Samples), SPSS performs automatically a Kruskal-Wallis test. When I select the Mann-Whitney test manually, I receive the notification that the results could not be computed.

What could be the reason for this? Even though I can also work with the results from Kruskal-Wallis, I would still like to understand why Mann-Whitney does not work in my case and what will be the implications for data interpretation (in terms of how reliable my conclusions will be).

I would be grateful for support!

------------------------------
Simone Trimmel
------------------------------

#SPSSStatistics

Dear Jon,
Thank you, no, my group has indeed only two values.

------------------------------
Simone Trimmel
------------------------------

Original Message:
Sent: Thu September 22, 2022 02:09 PM
From: Jon Peck
Subject: SPSS Statistics (v28) performs Kruskal-Wallis instead of Mann-Whitney U test

My guess is that your group variable actually has more than two values.  M-W only works with two.  In automatic mode, the dialog would have generated a K-W in that case.

--

Jon K Peck
jkpeck@gmail.com

Original Message:
Sent: 9/22/2022 2:09:00 AM
From: Simone Trimmel
Subject: SPSS Statistics (v28) performs Kruskal-Wallis instead of Mann-Whitney U test

Dear community,

I am trying to compare 2 groups of samples with a Mann-Whitney U test. In previous tests, I compared the same samples, grouped according to another parameter, and everything worked out. Now, when I try to follow the same procedure (Analyze > Nonparametric Tests > Independent Samples), SPSS performs automatically a Kruskal-Wallis test. When I select the Mann-Whitney test manually, I receive the notification that the results could not be computed.

What could be the reason for this? Even though I can also work with the results from Kruskal-Wallis, I would still like to understand why Mann-Whitney does not work in my case and what will be the implications for data interpretation (in terms of how reliable my conclusions will be).

I would be grateful for support!

------------------------------
Simone Trimmel
------------------------------

#SPSSStatistics

Original Message:
Sent: 9/28/2022 2:20:00 AM
From: Simone Trimmel
Subject: RE: SPSS Statistics (v28) performs Kruskal-Wallis instead of Mann-Whitney U test

Dear Jon,
Thank you, no, my group has indeed only two values.

------------------------------
Simone Trimmel
------------------------------

Original Message:
Sent: Thu September 22, 2022 02:09 PM
From: Jon Peck
Subject: SPSS Statistics (v28) performs Kruskal-Wallis instead of Mann-Whitney U test

My guess is that your group variable actually has more than two values.  M-W only works with two.  In automatic mode, the dialog would have generated a K-W in that case.

--

Jon K Peck
jkpeck@gmail.com

Original Message:
Sent: 9/22/2022 2:09:00 AM
From: Simone Trimmel
Subject: SPSS Statistics (v28) performs Kruskal-Wallis instead of Mann-Whitney U test

Dear community,

I am trying to compare 2 groups of samples with a Mann-Whitney U test. In previous tests, I compared the same samples, grouped according to another parameter, and everything worked out. Now, when I try to follow the same procedure (Analyze > Nonparametric Tests > Independent Samples), SPSS performs automatically a Kruskal-Wallis test. When I select the Mann-Whitney test manually, I receive the notification that the results could not be computed.

What could be the reason for this? Even though I can also work with the results from Kruskal-Wallis, I would still like to understand why Mann-Whitney does not work in my case and what will be the implications for data interpretation (in terms of how reliable my conclusions will be).

I would be grateful for support!

------------------------------
Simone Trimmel
------------------------------

#SPSSStatistics