a question on raw data eigenvalues of parallel analysis

[fufu]

I've a question on parallel analysis. I used O'Connor's syntax to do the parallel analysis in SPSS. The result is as follows:


PAF/Common Factor Analysis & Raw Data Permutation

Raw Data Eigenvalues, & Mean & Percentile Random Data Eigenvalues
         Root     Raw Data        Means     Prcntyle
     1.000000     4.647325      .639979      .728634
     2.000000     2.627683      .552915      .617100
     3.000000      .906230      .488962      .543583
     4.000000      .724413      .434583      .484906
     5.000000      .646384      .385350      .426445
     6.000000      .487371      .340357      .385823
     7.000000      .443986      .293844      .338639
     8.000000      .346904      .256452      .294767
     9.000000      .233345      .216896      .255271
.......

It seems that the first eight factors should be retained for the eigenvalues of the raw data are larger than the random data eigenvalues. The raw data eigenvalue from the third factor is smaller than 1. I  wondering whether it did matter or not. Is there any criterion that the raw data eigenvalue should exceed 1? Should I retain the first 8 factors or only 2 factors?

Thank you.

[Bruce Weaver]

It might help if you gave people a link to the syntax, as not everyone will know what you're talking about.  I assume this is it:

https://people.ok.ubc.ca/brioconn/nfactors/nfactors.html

fufu wrote

I've a question on parallel analysis. I used O'Connor's syntax to do the parallel analysis in SPSS. The result is as follows:


PAF/Common Factor Analysis & Raw Data Permutation

Raw Data Eigenvalues, & Mean & Percentile Random Data Eigenvalues
         Root     Raw Data        Means     Prcntyle
     1.000000     4.647325      .639979      .728634
     2.000000     2.627683      .552915      .617100
     3.000000      .906230      .488962      .543583
     4.000000      .724413      .434583      .484906
     5.000000      .646384      .385350      .426445
     6.000000      .487371      .340357      .385823
     7.000000      .443986      .293844      .338639
     8.000000      .346904      .256452      .294767
     9.000000      .233345      .216896      .255271
.......

It seems that the first eight factors should be retained for the eigenvalues of the raw data are larger than the random data eigenvalues. The raw data eigenvalue from the third factor is smaller than 1. I  wondering whether it did matter or not. Is there any criterion that the raw data eigenvalue should exceed 1? Should I retain the first 8 factors or only 2 factors?

Thank you.

[Art Kendall]

Why are you doing the factor analysis?
Is you are trying to create summative scales you have one or two possible scales if you can come up with enough clean loading items.

take a look at the archives of this list for "number of factors"
On 7/28/2012 12:06 PM, Bruce Weaver wrote:

It might help if you gave people a link to the syntax, as not everyone will
know what you're talking about.  I assume this is it:

https://people.ok.ubc.ca/brioconn/nfactors/nfactors.html



fufu wrote

I've a question on parallel analysis. I used O'Connor's syntax to do the
parallel analysis in SPSS. The result is as follows:


PAF/Common Factor Analysis & Raw Data Permutation

Raw Data Eigenvalues, & Mean & Percentile Random Data Eigenvalues
         Root     Raw Data        Means     Prcntyle
     1.000000     4.647325      .639979      .728634
     2.000000     2.627683      .552915      .617100
     3.000000      .906230      .488962      .543583
     4.000000      .724413      .434583      .484906
     5.000000      .646384      .385350      .426445
     6.000000      .487371      .340357      .385823
     7.000000      .443986      .293844      .338639
     8.000000      .346904      .256452      .294767
     9.000000      .233345      .216896      .255271
.......

It seems that the first eight factors should be retained for the
eigenvalues of the raw data are larger than the random data eigenvalues.
The raw data eigenvalue from the third factor is smaller than 1. I
wondering whether it did matter or not. Is there any criterion that the
raw data eigenvalue should exceed 1? Should I retain the first 8 factors
or only 2 factors?

Thank you.

-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/a-question-on-raw-data-eigenvalues-of-parallel-analysis-tp5714515p5714516.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

[fufu]

Weaver, thanks for your reminding. I used the syntax from O'Connor, which is the link you mentioned.

https://people.ok.ubc.ca/brioconn/nfactors/nfactors.html  (rawpar.sps)


To answer Kendall's question
"Why are you doing the  factor analysis?
        Is you are trying to create summative scales you have one or two possible scales if you can come up with enough clean loading items."

Actually, I tried to retain some factors from about 30 items. That's why I choose factor analysis.

[fufu]

Yes, that's the link I used. Thank you for your reminding.