Re: How many items in a scale before facctor analysis can be used.

[John F Hall]

Faiz

 

I suspect 3 items is far too few, but I may be wrong.  I'm no statistician, so I've copied your question on minimum number of items for factor analysis to the SPSS list.  Queries and answers on the list are archived so if anyone wants to help they can dig out our earlier exchanges from the following subjects:

"Categorizing study participants based on their scores on summated scales (non-SPSS question)."

"When to treat variables as continuous variables?"

Good to see you are making progress and have got as far as a *.sav file.  I can give you more help if you send me as attachments (off-list and in confidence) your *.sav file and a copy of your questionnaire in *.doc or *.pdf format.

Yesterday I came across an excellent tutorial site http://calcnet.mth.cmich.edu/org/spss/index.htm from Central Michigan University aimed at students with no previous experience of statistics or SPSS. 

SPSS (PASW) On-Line Training Workshop uses examples based on simple questionnaires and has step-by-step tutorials in SPSS and data analysis comprising video clips of the SPSS  data editor (in both Data View and Variable View) syntax editor and output viewer screenshots.  You won't be able to see the video, but you will find the running commentary extremely detailed and helpful.

Best

John Hall
[hidden email]
http://surveyresearch.weebly.com

 

----- Original Message -----

From: [hidden email]

To: [hidden email]

Sent: Saturday, January 22, 2011 6:47 AM

Subject: How many items in a scale before facctor analysis can be used.

Hi John,

 

You may remember me from our  prior correspondence. I'm a Blind Sociology student in University of Karachi, Karachi, Pakistan. I'm trying to improve my skills and knowledge of data analysis. You may also recall that I have no appropriate guidance available at my department. I have a dataset that I have been able to save in SPSS, and my aim is to publish a paper in a local journal. However there are many questions that are coming in my mind and sometimes because of my visual impairment and some time due to lack of guidance I'm unable to find satisfactory  answers. So if you do not mind I plan to ask some questions to you in days ahead.

 

The question that I have today is how many items should be in a scale before one should think about running a factor analysis. Obviously, it can not be run on two items, but what about scales having three items? Is there a criteria on how many items in a scale before factor analysis should be used.

 

Thank you for your help.

 

Regards,

Faiz.

[Tesiny, Ed]

Off the top of my head I'd say probably a minimum of 10-12 but you'd be better off with 20-30 or more depending on the construct you're trying to measure.  You also need a sufficient number of responses to have reasonable degrees of freedom.  As you said John, need a little more information.

________________________________

From: SPSSX(r) Discussion on behalf of John F Hall
Sent: Sat 1/22/2011 4:40 AM
To: [hidden email]
Subject: Re: How many items in a scale before facctor analysis can be used.


Faiz

I suspect 3 items is far too few, but I may be wrong.  I'm no statistician, so I've copied your question on minimum number of items for factor analysis to the SPSS list.  Queries and answers on the list are archived so if anyone wants to help they can dig out our earlier exchanges from the following subjects:
"Categorizing study participants based on their scores on summated scales (non-SPSS question)."

"When to treat variables as continuous variables?"

Good to see you are making progress and have got as far as a *.sav file.  I can give you more help if you send me as attachments (off-list and in confidence) your *.sav file and a copy of your questionnaire in *.doc or *.pdf format.

Yesterday I came across an excellent tutorial site http://calcnet.mth.cmich.edu/org/spss/index.htm from Central Michigan University aimed at students with no previous experience of statistics or SPSS.

SPSS (PASW) On-Line Training Workshop uses examples based on simple questionnaires and has step-by-step tutorials in SPSS and data analysis comprising video clips of the SPSS  data editor (in both Data View and Variable View) syntax editor and output viewer screenshots.  You won't be able to see the video, but you will find the running commentary extremely detailed and helpful.

Best

John Hall
[hidden email]
http://surveyresearch.weebly.com <http://surveyresearch.weebly.com/>




----- Original Message -----

        From: Faiz Rasool <mailto:[hidden email]>
        To: John F Hall <mailto:[hidden email]>
        Sent: Saturday, January 22, 2011 6:47 AM
        Subject: How many items in a scale before facctor analysis can be used.

        Hi John,

        You may remember me from our  prior correspondence. I'm a Blind Sociology student in University of Karachi, Karachi, Pakistan. I'm trying to improve my skills and knowledge of data analysis. You may also recall that I have no appropriate guidance available at my department. I have a dataset that I have been able to save in SPSS, and my aim is to publish a paper in a local journal. However there are many questions that are coming in my mind and sometimes because of my visual impairment and some time due to lack of guidance I'm unable to find satisfactory  answers. So if you do not mind I plan to ask some questions to you in days ahead.

        The question that I have today is how many items should be in a scale before one should think about running a factor analysis. Obviously, it can not be run on two items, but what about scales having three items? Is there a criteria on how many items in a scale before factor analysis should be used.

        Thank you for your help.

        Regards,
        Faiz.

=====================
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

[Art Kendall]

A lot depends on the circumstances and what you are trying to achieve.  Off the cuff questions such as these are among the things that can make a difference.

Are you just doing a quality assurance check that you have correctly written syntax that uses the key for a pre-existing instrument?

Do you have a new instrument that was written to measure several constructs?

Do you have a set of items intended to measure a single construct?

How many values can the items have?

If you are creating (a) new instrument(s), did you strive to have the response scale have as many response values as practical under the circumstances?

What is the nature of the constructs you are dealing with?

What is the population that you wish to make projections or generalizations about?

How was the data generated?

How were the respondents selected?



On 1/22/2011 4:40 AM, John F Hall wrote:

Faiz

 

I suspect 3 items is far too few, but I may be wrong.  I'm no statistician, so I've copied your question on minimum number of items for factor analysis to the SPSS list.  Queries and answers on the list are archived so if anyone wants to help they can dig out our earlier exchanges from the following subjects:

"Categorizing study participants based on their scores on summated scales (non-SPSS question)."

"When to treat variables as continuous variables?"

Good to see you are making progress and have got as far as a *.sav file.  I can give you more help if you send me as attachments (off-list and in confidence) your *.sav file and a copy of your questionnaire in *.doc or *.pdf format.

Yesterday I came across an excellent tutorial site http://calcnet.mth.cmich.edu/org/spss/index.htm from Central Michigan University aimed at students with no previous experience of statistics or SPSS. 

SPSS (PASW) On-Line Training Workshop uses examples based on simple questionnaires and has step-by-step tutorials in SPSS and data analysis comprising video clips of the SPSS  data editor (in both Data View and Variable View) syntax editor and output viewer screenshots.  You won't be able to see the video, but you will find the running commentary extremely detailed and helpful.

Best

John Hall
[hidden email]
http://surveyresearch.weebly.com

 

----- Original Message -----

From: [hidden email]

To: [hidden email]

Sent: Saturday, January 22, 2011 6:47 AM

Subject: How many items in a scale before facctor analysis can be used.

Hi John,

 

You may remember me from our  prior correspondence. I'm a Blind Sociology student in University of Karachi, Karachi, Pakistan. I'm trying to improve my skills and knowledge of data analysis. You may also recall that I have no appropriate guidance available at my department. I have a dataset that I have been able to save in SPSS, and my aim is to publish a paper in a local journal. However there are many questions that are coming in my mind and sometimes because of my visual impairment and some time due to lack of guidance I'm unable to find satisfactory  answers. So if you do not mind I plan to ask some questions to you in days ahead.

 

The question that I have today is how many items should be in a scale before one should think about running a factor analysis. Obviously, it can not be run on two items, but what about scales having three items? Is there a criteria on how many items in a scale before factor analysis should be used.

 

Thank you for your help.

 

Regards,

Faiz.

===================== 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