How do you compute a paired t-test on a question with four or five parts using a Likert Scale?

[Annette Hawkins]

A co-worker needs help on a thesis in Criminal Justice.  She is
comparing the mean responses of 80 men and 73 female correctional
officers on their training.  There are eight questions with four or five
parts.  Would you treat each subpart individually and run a t-test on
each part?  How would you treat the complete question?  She used Excel
and I told her Excel is not for statistical analyses.  I have listed
question one for you to see.  Thanks.


1.Firearms
a. Do you feel training is necessary to do the
job                     1  2  3  4  5
b. I learned how to retain a weapon during
confrontation           1  2  3  4  5
c. I have allotted time to effectively learn duty
weapon                  1  2  3  4  5
d. I am satisfied with firearm training
curriculum                      1  2  3  4  5
e Do you feel training should be twice a year rather once a
year            1  2  3  4  5

Strongly Agree: 1
Agree:2
Neutral:3
Disagree:4
Strongly Disagree:5

Thanks.



Annette D. Hawkins, Ed. D.
Math Instructor
Wayne Community College
3000 Wayne Memorial Drive
Goldsboro, NC 27534

"Nearly all men can stand adversity, but if you want to test a man's
character, give him power."
Abraham Lincoln

E-̄Mail correspondence to and from this sender may be subject to the
North Carolina public records law and may be disclosed to third parties.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Adam Martin-7]

Hi Annette

Excel can be used for statistical analyses, the only thing is that it's much easier in SPSS. Excel can be used for quick & dirties.

Looking at your situation, I would suggest running some validity tests such reliability analyses / factor analyses in order to see if you can group the questions together to make dimensions (e.g. Firearms).

Thereafter a t-test will be ample.

Regards
Adam

2009/2/24 Annette Hawkins <[hidden email]>

A co-worker needs help on a thesis in Criminal Justice.  She is
comparing the mean responses of 80 men and 73 female correctional
officers on their training.  There are eight questions with four or five
parts.  Would you treat each subpart individually and run a t-test on
each part?  How would you treat the complete question?  She used Excel
and I told her Excel is not for statistical analyses.  I have listed
question one for you to see.  Thanks.


1.Firearms
a. Do you feel training is necessary to do the
job                     1  2  3  4  5
b. I learned how to retain a weapon during
confrontation           1  2  3  4  5
c. I have allotted time to effectively learn duty
weapon                  1  2  3  4  5
d. I am satisfied with firearm training
curriculum                      1  2  3  4  5
e Do you feel training should be twice a year rather once a
year            1  2  3  4  5

Strongly Agree: 1
Agree:2
Neutral:3
Disagree:4
Strongly Disagree:5

Thanks.



Annette D. Hawkins, Ed. D.
Math Instructor
Wayne Community College
3000 Wayne Memorial Drive
Goldsboro, NC 27534

"Nearly all men can stand adversity, but if you want to test a man's
character, give him power."
Abraham Lincoln

E-̄Mail correspondence to and from this sender may be subject to the
North Carolina public records law and may be disclosed to third parties.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

--
Cell: [phone]
Website: http://www.sigmasurveys.co.za
Blog: http://www.sigmasurveys.co.za/resources

[David Greenberg]

Judging from the questionnaire, the individual items are ordinal variables. A test assumes interval-level variables. The distinction is often overlooked, and doing so may be harmless, but technically speaking,  Student's t test is not appropriate here. David Greenberg, Sociology Department, New York University

----- Original Message -----
From: Annette Hawkins <[hidden email]>
Date: Monday, February 23, 2009 10:13 pm
Subject: How do you compute a paired t-test on a question with four or five parts using a Likert Scale?
To: [hidden email]

> A co-worker needs help on a thesis in Criminal Justice.  She is
> comparing the mean responses of 80 men and 73 female correctional
> officers on their training.  There are eight questions with four or five
> parts.  Would you treat each subpart individually and run a t-test on
> each part?  How would you treat the complete question?  She used Excel
> and I told her Excel is not for statistical analyses.  I have listed
> question one for you to see.  Thanks.
>
>
> 1.Firearms
> a. Do you feel training is necessary to do the
> job                     1  2  3  4  5
> b. I learned how to retain a weapon during
> confrontation           1  2  3  4  5
> c. I have allotted time to effectively learn duty
> weapon                  1  2  3  4  5
> d. I am satisfied with firearm training
> curriculum                      1  2  3  4  5
> e Do you feel training should be twice a year rather once a
> year            1  2  3  4  5
>
> Strongly Agree: 1
> Agree:2
> Neutral:3
> Disagree:4
> Strongly Disagree:5
>
> Thanks.
>
>
>
> Annette D. Hawkins, Ed. D.
> Math Instructor
> Wayne Community College
> 3000 Wayne Memorial Drive
> Goldsboro, NC 27534
>
> "Nearly all men can stand adversity, but if you want to test a man's
> character, give him power."
> Abraham Lincoln
>
> E-̄Mail correspondence to and from this sender may be subject to the
> North Carolina public records law and may be disclosed to third parties.
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Matthew Pirritano]

I believe that ordinal items can be treated as interval if the test can be shown to be valid, meaning that it works, it predicts what it should predict, is related to what it should be related to, based on theory. I believe Shadish, Cook, & Campbell (2002) take this position, as do pretty much all social scientists that do research.

The real question is what would you combine? What can you combine? To answer that question you'd really need to do a factor analysis unless the questionnaire already has a established factor structure.

my2cents
matt

 

Matthew Pirritano, Ph.D.
Email: [hidden email]

---

From: David Greenberg <[hidden email]>
To: [hidden email]
Sent: Tuesday, February 24, 2009 6:23:02 PM
Subject: Re: How do you compute a paired t-test on a question with four or five parts using a Likert Scale?

Judging from the questionnaire, the individual items are ordinal variables. A test assumes interval-level variables. The distinction is often overlooked, and doing so may be harmless, but technically speaking,  Student's t test is not appropriate here. David Greenberg, Sociology Department, New York University

----- Original Message -----
From: Annette Hawkins <[hidden email]>
Date: Monday, February 23, 2009 10:13 pm
Subject: How do you compute a paired t-test on a question with four or five parts using a Likert Scale?
To: [hidden email]

> A co-worker needs help on a thesis in Criminal Justice.  She is
> comparing the mean responses of 80 men and 73 female correctional
> officers on their training.  There are eight questions with four or five
> parts.  Would you treat each subpart individually and run a t-test on
> each part?  How would you treat the complete question?  She used Excel
> and I told her Excel is not for statistical analyses.  I have listed
> question one for you to see.  Thanks.
>
>
> 1.Firearms
> a. Do you feel training is necessary to do the
> job                    1  2  3  4  5
> b. I learned how to retain a weapon during
> confrontation          1  2  3  4  5
> c. I have allotted time to effectively learn duty
> weapon                  1  2  3  4  5
> d. I am satisfied with firearm training
> curriculum                      1  2  3  4  5
> e Do you feel training should be twice a year rather once a
> year            1  2  3  4  5
>
> Strongly Agree: 1
> Agree:2
> Neutral:3
> Disagree:4
> Strongly Disagree:5
>
> Thanks.
>
>
>
> Annette D. Hawkins, Ed. D.
> Math Instructor
> Wayne Community College
> 3000 Wayne Memorial Drive
> Goldsboro, NC 27534
>
> "Nearly all men can stand adversity, but if you want to test a man's
> character, give him power."
> Abraham Lincoln
>
> E-̄Mail correspondence to and from this sender may be subject to the
> North Carolina public records law and may be disclosed to third parties.
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Linda Bruce]

Although you didn't ask....I see a bit of a problem with the way Q5 is
worded.
It is a yes/no (nominal)  response as it is now, but if you word it as "I
feel ..." it would allow for ordinal responses.
_________

Linda Bruce
Senior Analyst / Analyste Principale
Business Integration Office / Bureau d'Intégration des Affaires
Biologics and Genetic Therapies Directorate / Direction des Produits
Biologiques et des Thérapies Génétiques
Health Products and Food Branch / Direction Générale des Produits de Santé
et des Aliments
Health Canada / Santé Canada

2nd Floor, Room 128, Health Protection Building
200 promenade Tunney's Pasture Driveway, Tunney's Pasture
Ottawa, ON  K1A 0K9

tel/tél :  [phone]
fax/téléc :  [phone]
email/courriel :  [hidden email]

Address Locator / Indice de l'adresse : 0702E



David Greenberg <[hidden email]>
Sent by: "SPSSX(r) Discussion" <[hidden email]>
2009-02-24 09:23 PM
Please respond to
David Greenberg <[hidden email]>


To
[hidden email]
cc

Subject
Re: How do you compute a paired t-test on a question with four or     five
parts using a Likert Scale?






Judging from the questionnaire, the individual items are ordinal
variables. A test assumes interval-level variables. The distinction is
often overlooked, and doing so may be harmless, but technically speaking,
Student's t test is not appropriate here. David Greenberg, Sociology
Department, New York University

----- Original Message -----
From: Annette Hawkins <[hidden email]>
Date: Monday, February 23, 2009 10:13 pm
Subject: How do you compute a paired t-test on a question with four or
five parts using a Likert Scale?
To: [hidden email]

> A co-worker needs help on a thesis in Criminal Justice.  She is
> comparing the mean responses of 80 men and 73 female correctional
> officers on their training.  There are eight questions with four or five
> parts.  Would you treat each subpart individually and run a t-test on
> each part?  How would you treat the complete question?  She used Excel
> and I told her Excel is not for statistical analyses.  I have listed
> question one for you to see.  Thanks.
>
>
> 1.Firearms
> a. Do you feel training is necessary to do the
> job                     1  2  3  4  5
> b. I learned how to retain a weapon during
> confrontation           1  2  3  4  5
> c. I have allotted time to effectively learn duty
> weapon                  1  2  3  4  5
> d. I am satisfied with firearm training
> curriculum                      1  2  3  4  5
> e Do you feel training should be twice a year rather once a
> year            1  2  3  4  5
>
> Strongly Agree: 1
> Agree:2
> Neutral:3
> Disagree:4
> Strongly Disagree:5
>
> Thanks.
>
>
>
> Annette D. Hawkins, Ed. D.
> Math Instructor
> Wayne Community College
> 3000 Wayne Memorial Drive
> Goldsboro, NC 27534
>
> "Nearly all men can stand adversity, but if you want to test a man's
> character, give him power."
> Abraham Lincoln
>
> E-̄Mail correspondence to and from this sender may be subject to the
> North Carolina public records law and may be disclosed to third parties.
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[mpirritano]

Hello all,

Any idea why I'm being asked for a password to search the spssx
archives? I've never seen that before.

Also, the reason why I was searching the archives, I've got an old
laptop that I'm using for a task that has SPSS 12.0 on it. I use 17.0 on
my desktop. When was the filename limit changed? With what version was
the 9 character limit done away with. My filenames in 17.0 are often
longer than 9 characters.

Thanks
matt


Matthew Pirritano, Ph.D.
Research Analyst IV
Medical Services Initiative (MSI)
Orange County Health Care Agency
[phone]

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Parry, James]

Hi Matt,

That was version 12 where it went from 8 bytes to 64 bytes for var names. Also a good tip to find out what changed in which version is to go to help, topics and type in 'what's new' and it brings up all the features that were new in each version- not that you'd want to search every version, just good to know. I don't know why you need a password for the archive however.

-hth

James Parry

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Pirritano, Matthew
Sent: Wednesday, February 25, 2009 2:45 PM
To: [hidden email]
Subject: archives

Hello all,

Any idea why I'm being asked for a password to search the spssx
archives? I've never seen that before.

Also, the reason why I was searching the archives, I've got an old
laptop that I'm using for a task that has SPSS 12.0 on it. I use 17.0 on
my desktop. When was the filename limit changed? With what version was
the 9 character limit done away with. My filenames in 17.0 are often
longer than 9 characters.

Thanks
matt


Matthew Pirritano, Ph.D.
Research Analyst IV
Medical Services Initiative (MSI)
Orange County Health Care Agency
[phone]

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD