Factor Analysis Problem

[mahmoudee]

hi every

I'm a civil Engineer student and my graduation project about " Delay and Cost overrun in public construction"

i make a 30 questionnaire containing 37 variables and every variables Containing to question
1) Importance and the answer vary from (1 ...not significant) to (5....very Significant)
2) Frequency and the  answer vary From (1... not happened) to (5... Continually happened)

I use Factor Analysis in SPSS by multiplying the frequency by importance the result will vary from 1 to 25

i face some problems in SPSS like:
1) KMO test and Bartlett's test are not existing in the SPSS  i don't know why
2) under the the Correlation Matrix there are a note : This Matrix is not positive define
3) The analysis shows there are 9 significant component but i don't know to determine this 9 variables


Thanks  

[jonase]

Hi there,

It sounds to me like you would be well served by reading up on Factor Analysis procedures in general before continuing. "i make a 30 questionnaire containing 37 variables" If you are saying that you administered the questionnaire to 30 people, then that is a way too small sample for a FA.

1. KMO and Bartlett's is available under "Descriptives" in the Factor Analysis window in SPSS.

2. I suggest you read: http://www2.gsu.edu/~mkteer/npdmatri.html for some help on not positive definite error.

3. Again, I would suggest you read up on the steps that go into a Factor Analysis run, and the interpretation of your results.

J

[mahmoudee]

thank u very much Mr.jonase

you are right my survey contain 37 variables and i have asked 30 people

i don't know how many people have I asked?
it's not easy to find a lot of engineers to  to ask them to fill the questionnaire

the link u send is not helpful to me coz I'm not very good in statistics it's my first time working with SPSS and Factor Analysis

I provide the excel file which include my data please give me any not about the analysis...




Importance X Frequency Excel File

[Art Kendall]

> I use Factor Analysis in SPSS by multiplying the frequency by importance the
> result will vary from 1 to 25
I don't see why you would do this multiplication. What does this have to
do with Factor Analysis? How are you using the word "frequency"?

If you have more variables than cases, your correlation necessarily has
a problem that one or more variables is perfectly predictable from some
of the other variables.  Usually one wants to have many more cases than
variables for factor analysis.

What are you doing a factor analysis to find out?

Did you write the questionnaire or is it a pre-existing instrument?

At what level of school are you a student?

It would be a good idea for you to talk to someone who knows about
factor analysis. Where are you located?

Art Kendall
Social Research Consultant


On 4/8/2011 5:28 AM, mahmoudee wrote:

> hi every
>
> I'm a civil Engineer student and my graduation project about " Delay and
> Cost overrun in public construction"
>
> i make a 30 questionnaire containing 37 variables and every variables
> Containing to question
> 1) Importance and the answer vary from (1 ...not significant) to (5....very
> Significant)
> 2) Frequency and the  answer vary From (1... not happened) to (5...
> Continually happened)
>
> I use Factor Analysis in SPSS by multiplying the frequency by importance the
> result will vary from 1 to 25
>
> i face some problems in SPSS like:
> 1) KMO test and Bartlett's test are not existing in the SPSS  i don't know
> why
> 2) under the the Correlation Matrix there are a note : This Matrix is not
> positive define
> 3) The analysis shows there are 9 significant component but i don't know to
> determine this 9 variables
>
>
> Thanks
>
> --
> View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Factor-Analysis-Problem-tp4290412p4290412.html
> Sent from the SPSSX Discussion mailing list archive at Nabble.com.
>
> =====================
> 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

[mahmoudee]

the questionnaire has been designed by me it contain 37 variables and every variable contain two question importance of the variable and frequency of every variable

I'm a civil engineer student under graduated student

i live in Jordan and i faced problem to find any one know in SPSS and factor analysis  

[Rich Ulrich]

> Date: Fri, 8 Apr 2011 02:28:07 -0700
> From: [hidden email]
> Subject: Factor Analysis Problem
> To: [hidden email]
>
> hi every
>
> I'm a civil Engineer student and my graduation project about " Delay and
> Cost overrun in public construction"
>
> i make a 30 questionnaire containing 37 variables and every variables
> Containing to question
> 1) Importance and the answer vary from (1 ...not significant) to (5....very
> Significant)
> 2) Frequency and the answer vary From (1... not happened) to (5...
> Continually happened)

SO there are two sets of 37 variables, 74 in all...

>
> I use Factor Analysis in SPSS by multiplying the frequency by importance the
> result will vary from 1 to 25

... which (as I read it) you reduce to 37 by combining each pair,
by multiplying.  Adding would probably be better.  Choosing one of
the two as "more relevant", and dropping the other, would probably be
better.  The skewed distribution that you get from multiplying will
give extra emphasis to higher scores.  If 37 variables do have similar
averages, it may be possible to figure out some meaning for the
composites, but it won't be easy, even then.  It will be harder if their
averages are not similar.


>
> i face some problems in SPSS like:
> 1) KMO test and Bartlett's test are not existing in the SPSS i don't know
> why
> 2) under the the Correlation Matrix there are a note : This Matrix is not
> positive define

The determinant has to be zero ("not positive") whenever there are fewer
cases than variables.

> 3) The analysis shows there are 9 significant component but i don't know to
> determine this 9 variables

After you get a decent set of variables to enter, you look at the
loadings (Structure coefficients) to see which variables are extreme
on which Factors. The original structure has most variance loaded
into the earliest factors, with weights according to eigen values.
The Varimax-rotated solution is almost always better for your general
interpretation.

--
Rich Ulrich


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

[John F Hall]

With only 30 cases, I'm not sure you should be doing anything other than
descriptive analysis using simple counts and charts.  Most of this can be
done by hand rather than using SPSS, although it would be a useful learning
exercise on a student project.

The link you sent doesn't have the xls file.  Can you mail me off-list with
the Excel file attached and I can guide you through the process of setting
up a data file in SPSS: I can start you off, but you'll have to do all the
work yourself!

You can start by looking at Block 1 on my website "From questionnaire to
SPSS saved file"
(http://surveyresearch.weebly.com/block-1-from-questionnaire-to-spss-saved-f
ile.html ) and working your way through the exercises.  These start from a
short self-completion questionnaire and proceed through coding the data,
typing the data into a raw data file, reading the data into SPSS, defining
variable names, specifying variable and value labels, specifying values to
be treated as missing and saving the results in a *.sav file before
proceeding to data analysis.

Is your questionnaire in English?  If so, please send me a copy of that as
well.


John Hall
[hidden email]
www.surveyresearch.weebly.com


PS  Tell your supervisor to look at my tutorials as well!




-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
mahmoudee
Sent: 08 April 2011 19:52
To: [hidden email]
Subject: Re: Factor Analysis Problem

thank u very much Mr.jonase

you are right my survey contain 37 variables and i have asked 30 people

i don't know how many people have I asked?
it's not easy to find a lot of engineers to  to ask them to fill the
questionnaire

the link u send is not helpful to me coz I'm not very good in statistics
it's my first time working with SPSS and Factor Analysis

I provide the excel file which include my data please give me any not about
the analysis...




http://www.mediafire.com/?bm1ykahq1311vjv Importance X Frequency Excel File

--
View this message in context:
http://spssx-discussion.1045642.n5.nabble.com/Factor-Analysis-Problem-tp4290
412p4291556.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

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

[mahmoudee]

Delay_Ixf.sav
Delay_Ixf.xlsx
Delay_Ixf.spv

Questionnaire_Analysis_for_SPSS_5-4-11.xlsx

Questionnaire_Actual_Data_Full_5-4-11.xlsx

my survey English and i provide  3 excel file
1) Delay IxF.xlsx which contain the 37 variable with the 30 case and from the SPSS I import this file
2) Questionnaire Actual Data Full 5-4-11.xlsx this file contain the actual 30 survey i gathered then put it in  this file (note: in the down you will find cost overrun variables the analysis of this variables will be done separately)
3) Questionnaire Analysis for SPSS 5-4-11.xlsx this file contain the actual survey data  Importance, Frequency,  Severity "Importance x Frequency"  for each delay and cos over run

delay Ixf.sav and Delay Ixf.sov contain the out put of the analysis

thank u very much for your concern

[David Marso]

What we have here is a classic cases of GIGO waiting to happen (Garbage in Garbage out).
Factor analysis is THE classic method for running a GIGO analysis.
After all, you don't need to have a clue about anything.  The computer does all the work.
Then all you need to do is pull some fantasy crap from your bunghole and tell a great story.
Best of luck.  You need to READ up on Factor Analysis before proceeding further.
30 cases, 39 measures?  Seriously, do a little brain prep so you might be able to
understand any responses which you might be provided.

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