basic que-Factor Analysis
Locked
 
|
Hi,
I am trying to run (for the first time) factor analysis in SPSS.18 and would appreciate any feedback and recommendations from seasoned SPSS users and statisticians on the following I started with 93 continuous variables (5-point, 6-point and 7-point rating scale type questions) and trying to reach to a reasonable factor solution. Here are key questions I am struggling with. 1) I started with 93 variables. Are these too many variables to start with? Is there any guideline about (minimum and maximum) number of variables? 2) Though all variables I am using are continuous (rating scale) they are of different scale points and types (e.g. 5-point ratings where 1=Strongly Disagree and 5=Strongly Agree; 6-point scale where 1=Will not influence at all and 6=Will strongly influence etc.). I understand that this will have very different variance and therefore can possible affect the results. Having said that do I need to standardize these variables or factors analysis will automatically take care of it? 3) I selected an option for Rotated Component Matrix but it failed to produce that with the message “Rotation failed to converge in 25 iterations. (Convergence = .001).” Surprisingly, I ran the factor analysis with the same data couple days before (with 92 variables, I guess) and it did produce Rotated Component Matrix. Any thoughts on why it is showing that error and how to handle this? 4) My ultimate goal is to identify broader dimensions based on factor analysis and then use these composite variables for the segmentation by doing cluster analysis. Is this the right approach for the segmentation? I would appreciate any help on this. Regards, Mehul ===================== 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 |
|
|
Mehul
A few suggestions: 1) The maximum number of variables you can analyse to give a robust solution depends on the number of cases (subjects) and partly on how clear the factor structure is. A common suggestion is at least 10 cases per variable, though 5 can work fine in some situations. 2) If you choose Analyze Correlation Matrix on the Extraction submenu, the standardization will be done automatically. 3) Your 93rd variable may be problematic. Eg if it has many missing values, and you are using the default Listwise (on the Options submenu) it will reduce the number of cases available for analysis. Or it may simply be a few more iterations are needed. Try increasing the number of iterations on the Rotation submenu to say 100. If the solution still doesn't converge, it is likely your extra variable is causing the problem. 4) Yes you can cluster on the factors, provided your factor solution "makes sense" (i.e. is substantively interpretable). HTH Garry Gelade -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Mehul Pajwani Sent: 17 August 2011 03:17 To: [hidden email] Subject: basic que-Factor Analysis Hi, I am trying to run (for the first time) factor analysis in SPSS.18 and would appreciate any feedback and recommendations from seasoned SPSS users and statisticians on the following I started with 93 continuous variables (5-point, 6-point and 7-point rating scale type questions) and trying to reach to a reasonable factor solution. Here are key questions I am struggling with. 1) I started with 93 variables. Are these too many variables to start with? Is there any guideline about (minimum and maximum) number of variables? 2) Though all variables I am using are continuous (rating scale) they are of different scale points and types (e.g. 5-point ratings where 1=Strongly Disagree and 5=Strongly Agree; 6-point scale where 1=Will not influence at all and 6=Will strongly influence etc.). I understand that this will have very different variance and therefore can possible affect the results. Having said that do I need to standardize these variables or factors analysis will automatically take care of it? 3) I selected an option for Rotated Component Matrix but it failed to produce that with the message "Rotation failed to converge in 25 iterations. (Convergence = .001)." Surprisingly, I ran the factor analysis with the same data couple days before (with 92 variables, I guess) and it did produce Rotated Component Matrix. Any thoughts on why it is showing that error and how to handle this? 4) My ultimate goal is to identify broader dimensions based on factor analysis and then use these composite variables for the segmentation by doing cluster analysis. Is this the right approach for the segmentation? I would appreciate any help on this. Regards, Mehul ===================== 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 |
|
|
In reply to this post by Mehul Pajwani
Mehul,
You might look at Anna B. Costello and Jason W. Osborne: Best Practices in Exploratory Factor Analysis: Four Recommendations for Getting the Most From Your Analysis http://pareonline.net/pdf/v10n7.pdf the tutorials of A.J.Schwab , U of Texas : http://www.utexas.edu/courses/schwab/sw388r7/SolvingProblems/PrincipalComponentAnalysis.ppt http://www.utexas.edu/courses/schwab/sw388r7/SolvingProblems/PrincipalComponentAnalysis_Outliers_Validation_Reliability.ppt Jae-On Kim, Charles W. Mueller : Factor Analysis. Statistical Methods and Practical Issues Quantitative Applications in the Social Sciences Volume 14 Sage Jae-On Kim, Charles W. Mueller: Introduction to Factor Analysis. What It Is and How To Do It Quantitative Applications in the Social Sciences 1978 to get some ideas of the conditions and limits. HTH Dr Frank Thomas FTR Internet Research Rosny-sous-Bois France On 17/08/2011 04:16, Mehul Pajwani wrote: > Hi, > > I am trying to run (for the first time) factor analysis in SPSS.18 and > would appreciate any feedback and recommendations from seasoned SPSS > users and statisticians on the following > > I started with 93 continuous variables (5-point, 6-point and 7-point > rating scale type questions) and trying to reach to a reasonable > factor solution. Here are key questions I am struggling with. > > 1) I started with 93 variables. Are these too many variables to start > with? Is there any guideline about (minimum and maximum) number of > variables? > > 2) Though all variables I am using are continuous (rating scale) they > are of different scale points and types (e.g. 5-point ratings where > 1=Strongly Disagree and 5=Strongly Agree; 6-point scale where 1=Will > not influence at all and 6=Will strongly influence etc.). I > understand that this will have very different variance and therefore > can possible affect the results. Having said that do I need to > standardize these variables or factors analysis will automatically > take care of it? > > 3) I selected an option for Rotated Component Matrix but it failed to > produce that with the message “Rotation failed to converge in 25 > iterations. (Convergence = .001).” Surprisingly, I ran the factor > analysis with the same data couple days before (with 92 variables, I > guess) and it did produce Rotated Component Matrix. Any thoughts on > why it is showing that error and how to handle this? > > 4) My ultimate goal is to identify broader dimensions based on factor > analysis and then use these composite variables for the segmentation > by doing cluster analysis. Is this the right approach for the > segmentation? > > I would appreciate any help on this. > > Regards, > > Mehul > > ===================== > 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 |
|
In reply to this post by Mehul Pajwani
Do you mean Q Factor analysis which is an old way
of clustering cases?
Art Kendall Social Research Consultants On 8/16/2011 10:16 PM, Mehul Pajwani wrote: ===================== 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 REFCARDHi, I am trying to run (for the first time) factor analysis in SPSS.18 and would appreciate any feedback and recommendations from seasoned SPSS users and statisticians on the following I started with 93 continuous variables (5-point, 6-point and 7-point rating scale type questions) and trying to reach to a reasonable factor solution. Here are key questions I am struggling with. 1) I started with 93 variables. Are these too many variables to start with? Is there any guideline about (minimum and maximum) number of variables? 2) Though all variables I am using are continuous (rating scale) they are of different scale points and types (e.g. 5-point ratings where 1=Strongly Disagree and 5=Strongly Agree; 6-point scale where 1=Will not influence at all and 6=Will strongly influence etc.). I understand that this will have very different variance and therefore can possible affect the results. Having said that do I need to standardize these variables or factors analysis will automatically take care of it? 3) I selected an option for Rotated Component Matrix but it failed to produce that with the message “Rotation failed to converge in 25 iterations. (Convergence = .001).” Surprisingly, I ran the factor analysis with the same data couple days before (with 92 variables, I guess) and it did produce Rotated Component Matrix. Any thoughts on why it is showing that error and how to handle this? 4) My ultimate goal is to identify broader dimensions based on factor analysis and then use these composite variables for the segmentation by doing cluster analysis. Is this the right approach for the segmentation? I would appreciate any help on this. Regards, Mehul ===================== 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
Social Research Consultants |
|
|
by first three que, I am referring to R Factor Analysis (grouping variables) and once I am done with that, I intend to use those composite variables for cluster analysis.
Thanks,
Mehul
On Wed, Aug 17, 2011 at 11:22 AM, Art Kendall <[hidden email]> wrote:
|
|
|
We are looking for experienced freelance quantitative research analysts to write quantitative reports for an upcoming health care study. Must have your own copy of SPSS, PowerPoint, and Excel. Health care experience a plus. Experience with correlation, factor/regression helpful. Timeframe is Sept/Oct. For more information, please send resume to [hidden email] |
|
|
In reply to this post by Mehul Pajwani
"Failure to converge in 25 interations" is not fatal.
It might be an indication that you need 50 or 100 iterations, when two competing structures seem to fit. Of course, it could also indicate that var 93 had a lot of missing (as someone suggested), so that the analysis is a lot less stable than the analysis that worked, using 92 vars. -- Rich Ulrich |
|
|
Makes sense! Thanks very much to Frank, Farry and Rich!
Finally, i ran the factor analys and (Hope, I haven't missed anything) here is what I did. I would appreciate if you can take a quick look and let me know if I have missed anything or could have done this in a better way. I can send you my output in Excel or SPSS file, if required.
Iteration-1 (98 variables)
- Ran factor analysis with 98 items. Prior to performing PCA, the sutability of data for factor analysis was assessed. Inspection of r-matrix revealed the presence of many coefficients of 0.3 and above. KMO value was 0.961 and the Bartlett's Test of Sphericity as significant.
- Variables were validated based on communalities and 19 variables with loadings of less than 0.55 were dropped. 2nd Iteration was ran with 79 variables (98-19=79)
Iteration-2 (79 variables)
-Communalities- All 79 variables have communilities more than 0.50
-Measures of Sampling Adequacy (KMO- on the diagonal of the anti-image correlation matrix) was checked. KMO for all 79 variables were greater than 0.50. -After removing variables with low communalities and checking KMOs, the factor solution was examined to remove any components that have only a single variable loading on them. Component 13 was removed since it had only one variable. Componont 14 has no variables at all. Iteration-3 (78 variables)
-Communalities- All 78 variables have communilities more than 0.50 - The measures of sampling adequacy (KMO) for all the variables are greater than 0.50. (actually, greater than 0.7, which is good) - All 13 componants have 2 to 24 variables
Final Solution:
13 compononts with variables loadings as below. I got 24 variables for Componant-1 and 15 for compononant-2. Are these too many? Componant Number of Variables
1 -24 2 -15 3 -6 4 -7 5 -5 6 -6 7 -3 8 -2 9 -2 10 - 2 11 - 2 12 - 2 13 - 2 Total- 78 On Thu, Aug 18, 2011 at 6:30 PM, Rich Ulrich <[hidden email]> wrote:
|
|
|
What are you using the factor analysis for? (to
create a summative score?) Is this a pre-existing instrument or
did you create it? If you are creating summative scales did you
remove "splitters" from the final key?
What did you use for a stopping rule? Do the factors make sense? try running parallel analysis. With moderate Ns (several hundred to a few thousand), I have usually retained the number of factors where the obtained eigenvalue is one more than what would come from random data or random permutations of the same data. http://flash.lakeheadu.ca/~boconno2/nfactors.html Art Kendall Social Research Consultants On 8/19/2011 5:20 PM, Mehul Pajwani wrote: ===================== 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
Social Research Consultants |
|
|
The ultimate objective is to segment the respondents based on their perception about current and future economy and how they are/will be dealing with the changes in the economy.
I intend to define the broad dimensions based on the factor analysis and then to use the same for cluster analysis. Yes, factors do make sense and are seem to be meaningful. BTW, all these 14 factors have eigenvalues more than 1. Would parallel analysis help me reduce the number of variables loading under a given factor? Would it be a good idea to use the current factor solution? Please let me know your thoughts on this Thanks, Mehul On Sat, Aug 20, 2011 at 7:00 AM, Art Kendall <[hidden email]> wrote:
|
|
|
It would not be a good idea to use the current
factor solution.
The Kaiser criterion is used to limit the number of factors to extract. It is based on the idea that one would not be interested in a new dimension that accounted for on one "variables worth" of the variance. It merely avoids inefficient use of computer time. The parallel analysis test and scree test will give you a ball park number of factors to retain. The final number of retained factors is based on substantive interpretation of the constructs that might underlie set of clean loading items. You need to look at the interpretability of the overall factor solution and of each candidate factor. Art Kendall Social Research Consultants On 8/21/2011 12:40 AM, Mehul Pajwani wrote: The ultimate objective is to segment the respondents based on their perception about current and future economy and how they are/will be dealing with the changes in the economy.===================== 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
Social Research Consultants |
|
|
Thanks for your advise Art!
May I ask you the main reason why it is not a good idea to use the current factor solution? I mean, is it because of too many variables on first two factors? I also looked at the scree plot and it tells to retain 4-5 factors Would you like to see the scree plot and/or Rotated Component Matrix? I would appreciate your help on this. thanks, Mehul On Tue, Aug 23, 2011 at 4:26 PM, Art Kendall <[hidden email]> wrote:
|
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |