From your description it appears that you are trying to explore whether using the Exploratory Factor Analysis (EFA) are you able to get a similar factor structure composed of 24 factors as obtained by earlier researchers. Though, you have obtained 58 factors (how?), this way of doing factor analysis can not be termed as 'confirmatory factor analysis even if you obtain 24 factors having a similar pattern of loadings across all the factors.

If one does an EFA and obtains same number of factors then there is a need to assess the factor structure congruence. And if the index of factor structure congruence is high then one may conclude that the earlier reported factor structure may be generalized across different sample (population). This procedure may be termed as cross-validation of the factor structure and such procedure of factor analysis may be classified at best as inferential exploratory factor analysis and not the confirmatory factor analysis.

Confirmatory factor analysis is is a hypothesis testing approach to factor analysis where one defines the factor structure apriori and using the structural equation modeling approach how far the data fits with this predefined factor structure. If this is your objective then as suggested by others you will have to use special software like AMOS, LISREL, EQS etc.

However, if your objective is to do a cross validation of earlier reported factor structure then EFA is advised. Use any one method of EFA (preferably the one used in earlier research) and then use some tests to determine the number of factors to retain for subsequent rotation. It would be better if one adopts several criteria to decide the number of factors such as scree plot, parallel analysis, Velicer's MAP. C-Hull method etc.

If you get same number of factors as obtained in previous research even then you can not say that you have arrived at a similar factor structure unless you demonstrate that all the factors have pattern of factor loadings similar to the earlier one. This can be done by computing factor structure congruence that has been nicely described in the Book on factor analysis by Gorsuch.

Like many others, however, I also doubt that the 24 factors may not represent first order factor related to one second order latent construct. Rather, they may be related to several second order factors. And if this is the case then you will have to follow an entirely different approach to conduct the factor analysis.

Hope this helps.

Hi

One possible explanation for the differences in factors found may be differences between yours and the prior work in the strength of the loadings each are requiring to accept a factor into the final model. What loadings were required in that prior work? What loadings did your work require. Factor differences may also result from differences in the source populations that generated the data being (factor) analyzed.

I hope these coments are helpful. Be well,

Howard

SPSS can perform true factor analysis in addition to PCA if you change the extraction method in the drop-down menu under the 'extraction' tab (for example, you can choose maximum likelihood). However, it absolutely cannot do confirmatory factor analysis - you need a structural equation modeling software for this. A compromise would be to tell SPSS to specifically extract exactly 24 factors in the extraction tab (rather than the default of extracting all factors with eigenvalues greater than 1) - you cannot control which items load onto which of these 24 factors and cannot obtain appropriate fit indices, which is why this is still exploratory, but it will show you whether the items you thought comprised the 24 factors actually do load onto them. A word of caution - you should request a scree plot from the extraction menu as well and then look up information on conducting Catell's scree test to get a better sense of how many factors you really need (the number you got using eigenvalues greater than 1 may be wrong because the eigenvalues are often inappropriate with large numbers of items, but 24 may nonetheless not be appropriate either).

SPSS uses "principal component analysis" as a choice between different possible alternatives of factor analysis. I am not using SPSS from long ago, but SPSS is somehow "obscure" on the way some analysis are performed. A couple of years ago I was trying to replicate in R the factor analysis of SPSS and was not straightforward. I would deserve to get in contact with the authors of the original paper and ask them for the details of the analysis. The problem is that many scientific publications just refer to SPSS and subsequently the "real things" behind the analysis is lost.

Hello,

to my knowledge SPSS cannot perform a CFA, only a EFA. But what do you want to do? How is your hypothesis formulated?

Best wishes,

Claas

Hi,

SPSS just run exploratory factor analysis (with principal components or common factor extraction).

AMOS is the package of SPSS that run confirmatory factor analysis, similar to LISREL, EQS and others.

If you do not have expertise with AMOS, LISREL…, one idea is:

- run one exploratory factor analysis for each original dimension (extracting just one factor)

- assess the convergent validity (factor loading, communality and variance extracted)

- assess the reliability (Cronbach’s alpha or composite reliability).

- you also could save the factor scores for each dimension and compute the correlation between them, to assess the discriminant validity.

Best regards,

Bido

I'd prefer using SPSS-AMOS for conducting CFA..

Please refer to the following research article for further information:

Article When empathy hurts: Modelling university students’ word of m...

Best,

In SPSS, you can only fix number of factors to your required number but that is not CFA. For CFA you need any SEM program like those of AMOS, Mplus, Lisrel etc.

In order to perform CFA you will need AMOS, form the SPSS "suite". Nevertheless you will find problmes in your overall fit with so many variables and such a huge number of factors. I would suggest to perform an separate confirmatory analysis (with AMOS) for each individual factor, alike Diogenes proposes.

You can't perform real CFA with SPSS, if you don't have IBM SPSS AMOS installed. It's a software package for general SEM analysis, and you can do CFA with it. You can specify the number of factors and the relationship to each item. Rule of thumb suggests that you need 3 items at least per factor. AMOS is just one example of SEM software tool, and it seems to be the best choice if you are already working with SPSS.

http://www-142.ibm.com/software/products/us/en/spss-amos/

Hi,

If you want to use SPSS (V20) you have to select your number of factors (e.g. n=8) and then test your model in terms of how much it is similar to the classification of items in previous studies. This method is not common and technically is not appropriate. The common way for CFA is using AMOS or Lisrel .

Hi,

What I deduce from your commentary is that your are computing a factor analysis or principal component analysis but no a confirmatory factor analysis.

As many colleagues say the default SPSS options do not compute confirmatory factor analysis. You need AMOS or another special software.

In confirmatory factor analysis the estimated parameters follow your apriori specification. Then you have to look the goodness of fit of the model to the data, but you do not obtain another models as a consequence of your analysis. Nonetheless you can compare the fit of different models to the data.

Hi,

As far as I know SPSS cannot performa CFA. For this purpose you have to make yourself familiar with other software such as EQS or AMOS or similar programs of structural equation modelling.

Hi

First, CFA can be conducted using AMOS software which is part of SPSS version 19. Second, the CFA is different than the EFA in that it requires a hypothesized model and fixed number of factors. Therefore, you have to run a EFA using principle component analysis first to identify your factors anf if you want a certain number of factors eg. 24 you have to increase the loading value from say 0.3 to 0.4. Then choose the appropriate rotation based on whether the produced factors were correlated or not. Finally you can perform the CFA to determine the goodness of fit.

Hope this help

Oras

Agree IBM SPSS provides SEM capability (including CFA) through the AMOS module (a SEM program developed by James Arbuckle at AMOS Development Corp. (http://www.amosdevelopment.com/ ). )

Hi

For conducting a CFA, you should use other statistical softwares such as AMOS, LISREL and or EQS

Good luck in your research,

Esmaeel

However, I also agree that a model with 24 factors is likely to be too complex to be easily interpreted.

I think you could possibly do a simple CFA or PCA in SAS or Stata or R.

As others have suggested - I believe you should perform EFA first ( and then confirmatory factor analysis using AMOS )and in the coefficient display value -suppress absolute value less than .4 you can increase it to .5 . Explore factors and group them in domains.

"Look at the table 'Communalities' and the column labelled Extraction to decide how many factors to extract. If these

values are all 0.7 or above and you have less than 30 variables then the SPSS default option for extracting factors is fine

(Kaiser’s criterion of retaining factors with eigenvalues greater than 1). Likewise, if your sample size exceeds 250 and the

average of the communalities is 0.6 or greater then the default option is fine. Alternatively, with 200 or more participants the

scree plot can be used. [ DISCOVERING STATISTICS USING SPSS by Andy Field ]"

Like the previous posters have said, a CFA is not available in SPSS, you would have to get AMOS, LISREL, or MPlus.

In regards to obtaining 58 factors rather than your 24 that you're expecting, what type of exploratory factor analysis are you using? What factor extraction method and what method of estimating communalities? I would recommend Floyd & Widaman (1995) as a great summary of all the different methods and intricacies of Factor Analysis.

Instead of looking at communalities that are above 1.0 or .7 (which isn't the most reliable way of doing so), examine the scree plot, and consider running a parallel analysis. The syntax for a parallel analysis can be found on Brian O'Connor's website: https://people.ok.ubc.ca/brioconn/nfactors/nfactors.html

Floyd & Widaman (1995) can be found here: http://psychology.ucdavis.edu/labs/widaman/mypdfs/wid079.pdf

I completely agree with Nasos Gouras. So far I know SPSS cannot performa CFA. For this purpose you have to make yourself familiar with other software such as EQS or AMOS or similar programs of structural equation modelling.

Hi,

You can also use Mplus.to run CFA. You only run CFA if you used items from an existing scale and you want to make sure that the items hold true and have similar alpha as the original scale. However, the alpha may vary depending on the population you are studying. I am not sure why you plan on running EFA? You already should know which items should be considered to formulate the scale.

Are you sure you're getting 58 factors and not 58 parameters estimates?

Hi. just seek this path: analyze-data reduction-factor-extraction-and then active the factor number buttom and then determine the number of factors.good luck

I wonder SPSS is even able to do CFA.

My idea, Amos can perform and get result easy and better than Lisrel .

Also Spss , factor analysis have problems if the variables too much , the results not comples, so the best way try to check the best result from all tools can chose the best results as most experts acceptable

As others have said, SPSS can't do CFA. I do want to give a shout-out to a great (FREE) package that can: lavaan, running in the R statistical computing environment. Although R itself has a pretty steep learning curve, the lavaan package has easy-to-understand syntax. Check it out at http://lavaan.ugent.be/.

Also, be sure to consider sample size--if you are using EFA to search for 24 factors from 240 items, you'll need a big sample size (about 1000 participants). But, if theory specifies a certain number of factors--which, from your comment, seems to be the case--then EFA is inappropriate, because it is well known that EFA can fail to recover the true factor structure. For more discussion on this point, see:

Little, T. D., Lindenberger, U., & Nesselroade, J. R. (1999). On selecting indicators for multivariate measurement and modeling with latent variables: When “good” indicators are bad and “bad” indicators are good. Psychological Methods, 4, 192-211. doi:10.1037//1082-989X.4.2.192

Just a note - if you are working with items you will likely not want to use SPSS for the EFAs. I have not looked recently but I do not believe the program handles categorical data (e.g., item responses) appropriately. SEM programs such as LISREL or Mplus would work for EFAs and CFAs. Given the number of items, you may also want to consider IRT-based programs (e.g., flexMIRT) for your CFA analyses. The SEM programs usually choke when there are a large number of items.

You can just fix the extraction to 24 factors. Factor Analysis > Extraction > Fixed number of factors. Also consider to fix the number of factors based on your judgment on the scree plot. Just a word of reminder that this is not a proper CFA, but may give some idea before venturing into proper CFA.

Factoring questionnaire items is fraught with problems of non-normal distribution of item responses, sample size in relation to number of items (as Andrew Ledbetter confirms) the use of components that magnify later factors, compared with maximum likelihood, and so one could go on.

My advice, for what it is worth, is to begin using simple and proven psychometric methods before confirming. Using a scree will let one know when variance is probably being extracted in a quasi-random manner, and one can only advise stopping factoring when the scree begins to level out. Rotate first orthogonally and then using promax. Look at the correlations of the factors.

Without seeing your data, my blind, self-indugent approach would be to look at fewer than 16 factors - Cattell needed no more; and Christal -Supes needed only five! The canon of parsimony never fails.

Keep in touch. Good Luck!

Sidney Irvine

You can establish fixed factors in SPSS that you consider necesary. However is very striking the presence of 24 factors. If you continue with this problem, we could talk trough skype to resolve that.

Just wanted to emphasize that, as Wan Arifin notes, fixing extraction at 24 hours is most emphatically NOT a substitute for CFA. At best it's an exploratory/diagnostic approach. The main difference is that the factor extraction is still data-driven, whereas true CFA is theory-driven (because the researcher must a priori specify the model--of course, that's not to say that the theoretical model is necessarily a sound one).

As previously mentioned, SPSS can force 24 factors through its normal dimension reduction interface, but I would not personally recommend SPSS for this purpose. Others have mentioned many of the popular software packages. If you have access to mplus, it is relatively easy. Please message me if you would like links to some helpful example syntax. As others have also mentioned, you may also want to consider an IRT or IFA approach if your responses are not actually continuous, which they rarely are in survey and testing data.

I wonder whether you really do have 24 independent factors, or a hierarchical structure which is made up of a small number of factors, each with subfactors. For example the NEO-PIR model of the Big Five has five main factors, each with 6 'facets.' Even models that start with large numbers of factors (e.g., Jackson's model with about 20) can be collapsed into 5 or 6, say, that are easier to understand. That may be worth trying - but you will need a structural equation modelling package to do that, such as AMOS or one of the others suggested above.

As others have said, if you want to stick with the SPSS family of products, then you have to use AMOS for confirmatory factor analysis. In AMOS, you specify the model that you are trying to confirm, so technically you do not have to extract a certain number of factors. The model presupposes the number of factors. Looks like you are assuming 24 latent variables, each with 10 items design to measure those concepts.

The bad news, from my experience anyway, is that when you are trying to confirm 24 factors using 240 items, you are very unlikely to have any luck. You will then have to start assessing potential changes to the model, as suggested by your data. At that point you are back in the realm of exploratory factor analysis.

Indeed. Back to basics seems to be the consensus.

From your description it appears that you are trying to explore whether using the Exploratory Factor Analysis (EFA) are you able to get a similar factor structure composed of 24 factors as obtained by earlier researchers. Though, you have obtained 58 factors (how?), this way of doing factor analysis can not be termed as 'confirmatory factor analysis even if you obtain 24 factors having a similar pattern of loadings across all the factors.

If one does an EFA and obtains same number of factors then there is a need to assess the factor structure congruence. And if the index of factor structure congruence is high then one may conclude that the earlier reported factor structure may be generalized across different sample (population). This procedure may be termed as cross-validation of the factor structure and such procedure of factor analysis may be classified at best as inferential exploratory factor analysis and not the confirmatory factor analysis.

Confirmatory factor analysis is is a hypothesis testing approach to factor analysis where one defines the factor structure apriori and using the structural equation modeling approach how far the data fits with this predefined factor structure. If this is your objective then as suggested by others you will have to use special software like AMOS, LISREL, EQS etc.

However, if your objective is to do a cross validation of earlier reported factor structure then EFA is advised. Use any one method of EFA (preferably the one used in earlier research) and then use some tests to determine the number of factors to retain for subsequent rotation. It would be better if one adopts several criteria to decide the number of factors such as scree plot, parallel analysis, Velicer's MAP. C-Hull method etc.

If you get same number of factors as obtained in previous research even then you can not say that you have arrived at a similar factor structure unless you demonstrate that all the factors have pattern of factor loadings similar to the earlier one. This can be done by computing factor structure congruence that has been nicely described in the Book on factor analysis by Gorsuch.

Like many others, however, I also doubt that the 24 factors may not represent first order factor related to one second order latent construct. Rather, they may be related to several second order factors. And if this is the case then you will have to follow an entirely different approach to conduct the factor analysis.

Hope this helps.

As was noted by others previously SPSS itself does not do CFA. You will use something like AMOS, Mplus or other Structural Equation Modeling packages (most likely) to do this - in most cases. There are other programs that do it however, so it is not just those. A factor structure of 24 factors with 240 items seems pretty unusual. In using the BPI for several years with it's 240 items it makes me wonder - this would mean there were 10 items per scale (assuming no repeated item use). reusing items creates discriminant validity issues for the scales/factors in most cases (and would suggest an Oblim rotation is appropriate). It makes me wonder if you do not mean there are 24 scales and not factors. However assuming you do mean 24 factors we will go with it.

If you apply CFA to the model that was initially constructed using EFA the likelihood of it fitting is very small. They each approach the question differently and the results will likely differ. In the "old days" we used to do a parallel process to modern CFA (SEM) using EFA. Now that the chi square test in SEM has become largely ignored the EFA process is not really a lot different than SEM anymore in terms of model testing. Thre are still some advantages to SEM for CFA though and that would be the way to go if you are comparing the relative fit of various models to the data.

We used to use EFA to do what you are trying to do in the days before SEM was widely available ("old days"). Douglas Jackson describes the process in the PRF manual I think, as well as in some of his other work. Whether it will suffice for your situation or not I do not know. You are really trying to see if the EFA replicates. As such you locate the exact method used by the original researchers. You then use that approach in your EFA. You set the number of factors to the number they claim using the approach they used and run it. You then see how well the items match the loading (scales and magnitude) from the original study.

As noted the factoring of items is not eh same as factors of scales. Some cautions are needed. See:

https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&ved=0CFUQFjAA&url=https%3A%2F%2Fbitbucket.org%2Feyecat%2Freadinglists%2Fsrc%2Fd8e8010f0b0d2dbb0863af3050411695254cc6b1%2FReadingList_NotreDame%2FBernsteinTeng1989FactorItemsFacotrScales.pdf&ei=XilcUeiBJeLpygGcwoDQAw&usg=AFQjCNG_D3YpnYEeXdvcPkwar6FWvOg_hg&bvm=bv.44697112,d.aWc&cad=rja

However it is not uncommon to use EFA and CFA with items. It is possible and is used as an alternate test construction method.

Or use one of the free open source R packages such as SEM, Lavaan or OpenMx. The first two are available at CRAN, the third at http://openmx.psyc.virginia.edu/

Disclaimer: I am on the core development team for OpenMx.

AMOS is available in SPSS version 20

I personally prefer MPlus, though other softwares like Lisrel and EQS would also be helpful to perform a CFA. It is possible to force the number of factors to perform something akin to a CFA using SPSS. However, this is typically not recommended. I would perform a CFA (using one of the softwares suggested above), and run one model where the factor structure is suggested by theory, and compare the fit indices to alternate models suggested by theory. MPlus has a very helpful manual, and also Bengt and Linda Muthen are very helpful with suggestions on the MPlus discussion board. Try this!

24 factors is huge enough and you are getting 50+. I think relook your factors or you can cut down intial 240 items to around 180. CFA can be done with AMOS but it does not come with SPSS. You have to take seprate license for it. You can search for its trial version.

I agree with Anant Saxena that 24 factors is just too many. What on earth you are studying that you need 240 items? It means that you are not sure what you are studying. The good side of it is that PSS is telling you that roughly there are about 50 things there, about 5items per factor on average.

My experience is that a robust factor needs about 5-7 items and a good scale should have 5-7 latent factors. Anything beyond that is very nard to confirm across contexts/samples. You need cross-validation in the long run. So you have Big Five for personality, not Big 10. Only God can handle more than 10.

I'm agree with James Ko also

As repeatedly noted, SPSS does not do a CFA on its own. While I have seen several plugs for Amos, I enjoy R for this and it is free, although it does require a bit of syntax coding on your own (r-project.org). However, there are a few other things within your post that struck me. Although it seems as though a CFA would probably be the best analysis, doing an EFA and specifying the number of extracted factors is also a possibility if you are looking to see if the factors emerge in the pattern originally described, although CFA is likely a better fit for the analysis as described since it is regularly used for pre-formulated models. If you have are looking for an alternative model for some reason, that may argue better for the EFA. I have seen some argument put forward that EFA is also a better way to address oblique measurement constructs. Although I am not entirely sold on this, I do see the strength of it in some regards.

Another option, as Diogenes Bido mentioned, is that you may also just choose to do an EFA on each independent factor. This is particularly useful in structures requiring oblique rotation because it allows for the covariance between construct areas to be minimized (a good reference for this in personality instruments is Gignac, Bates, & Lang, 2007 although they use SEM instead of EFA).

Also, depending on how you determine the threshold for inclusion of a factor, you may be conceptualizing the number of extracted factors. I am not sure if you have used Monte Carlo to take into account the sample size and number of items in order to minimize noise/extra extractions (one of the risks of EFA). If not, I would highly recommend doing that and there are some fantastic programs for it that are free/easy to use.I would advise against just using the Scree plot or taking for granted what SPSS says are extracted factors above a 1.0 eigenvalue.

Best of luck.

All of these answers are correct--SPSS cannot do CFA. But it sounds to me like you might be wanting to simply have SPSS extract just 24 factors in an EFA. If that is so, the subcommand, "/criteria factors (24)" will let you do that. Otherwise, the SPSS default is to extract all factors with eigenvalues greater than 1.

Try tis:

http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=5&ved=0CFkQFjAE&url=http%3A%2F%2Fwww.ucdenver.edu%2Facademics%2Fcolleges%2Fnursing%2FDocuments%2FPDF%2FFactorAnalysisHowTo.pdf&ei=pRCOUsCEKouEkQet0IGACQ&usg=AFQjCNEXw66VNh2kfl9sgGXYj9MsK5EtVg&sig2=GZkIAFB7E60WavjobJdipw&bvm=bv.56988011,d.eW0&cad=rja

I think you should go with SmartPLS software. Your no of items is very high. Based on EFA from SPSS, using SmarPLS you will be able to examine the strength of items in terms of loading. Based on loading values, you can omit the poor items (i.e. items with loading below 0.4 as suggested by Hulland, 1999). This will help you in reducing items as well as you can modify your factors.

Use either AMOS, LISREL, or the Hunter CFA program for CFA.

Hi Merche!

Indeed. SPSS only runs explanatory factor analysis. I would recommend using AMOS if you don't like programming yourself (syntax) as you can 'draw' the models you want to test.

Dear Merche,

If you vish to do factor analysis in SPSS and limit the number of factors you should click Extraction, then tick option Number of Factors and in blank field enter 24.

You then have to calculate Tucker coefficients of congruence to find out similarity of your factors with the factors obtained by other authors.

BUT, as it has already told by many respondents to your question it is not Confirmatory factor analysis "in the proper meaning of that term". It will be in some sense "confirmatory" using of Exploratory factor analysis. If you wish to do proper CFA you should use (as it has already told by meny respondents)specialised software for CFA , like AMOS, EQS and similar packages.

AMOS is very good for begainer, it's very friendly.

Why do you need so many items to perform a factor analysis? I think you have to perform a confirmatoy factor analysis with AMOS, for example, but I recommend you to use no more than 5 items for each factor

Even using CFA, it is often not possible to show unidimensionality with a large # of items because these items are likely measuring multiple factors. In cross cultural research the problem is accelerated and often there are severe validity issues ( See Oyserman et al's discussion of validity issues for individualism and collectivism or Levine et al., 2002 on lack of validity of self construal measurement).

Confirmatory factor analysis with Amos program.

You can slect the number of factors to retain in exploratory with not select eigen=1

I would suggest you take a look at the option of calling R from within SPSS. R is very flexible (and free) and there are a couple of options for doing structural equation modeling (the SEM and LAVAAN modules). If you only need occasional access to CFA, that may be a solution.

You're trying to test if your data will confirm the same scale structure found by others so you don't do efa (at this point), and spss doesn't doesn't let you do Cfa so you need to use different software (e.g., Amos. Sass...). But, you also have a very large number of items which means that you will need a very large sample (probably a couple thousand) to have enough power to produce reliable results.

As already stated you either need the AMOS add-on to perform a CFA in SPSS or need to rely on one of the other programs (LISREL, MPLUS, R with lavaan-package).

I would like to add that you probably will have a hard time to achieve reasonable fit for such a highly multi-factored structure (24 factors) with analysis on item-level and as much as Ø 10 items/factor. Using WLMSV (I know it’s included in MPLUS) instead of the standard ML-estimation might help a little if you are using likert-style items.

Best of luck!

1. SPSS can only do EFA, except with add-on for SPSS 19 and above.

2. The number of indicator you are using are too many.

3. If what you are doing is exploratory study, I will recommend you use one of the PLS software, especially smart PLS.

4. I will recommend you see an expert close to you on the use of these statistical softwares.

5. If you are in Malaysia then I may be of help to you.

Wish you the best.

.

Using AMOS21 or LISREL with 24 factors would be difficult.

What criteria were you using to justify retaining that many factors? You might look at Horn's Parallel Analysis and Velicer's Minimum Average Partials as ways to more reliably retain viable factors and also look at the internal consistency of the resulting dimensions and if below .70, drop it. If using PCA there will be inflated factor coefficients so after determining the salient items redo the EFA using principal factors and see how many factors meet the HPA and MAP criteria and go from there. Also, if factors are correlated you should then do a second order EFA and Schmid-Leiman (1957) orthogonalization to determine variance apportions to the higher and lower order factors. These can also be done in CFA (pick your program) and bifactor modeling might also be of help.

You check the variable before use programs so you would have good variable after you use factor analysis

To conduct CFA, it is better to have software packages such as AMOS

I'm afraid you really need AMOS (which has been acquired by SPSS so you may have access to it). SPSS does not do CFA. However you could restrict the number of factors to 24 (its an option within the SPSS menu) and see if you get the same 24 factors. However for publication purposes they will still want proper CFA.

For doing CFA with such a big number of variables, only software Statistica by StatSoft would be able to help you.

Otherwise doing it with AMOS or PLS or any other one would take a long time in drawing only.

Hi

Factor Analysis is a data dependent technique and highly related to the reliability and validity of the questionnaire. Did you check the reliability and validity of the questionnaire items you are using? Many a time the questionnaire developed in particular settings doesn't fit well in other settings. Besides this, one should always check these two parameters to ensure that the questionnaire items fulfill the requirement if in case they are adapted from the original resource.

Moreover I have a doubt whether you have taken sample from same geographical region or from a different one (many external factors related to the different geographical regions like culture, societal values etc may impact the response) . Further the responses change over a period of time even if the data is collected from the same sample (sometimes due to bias or may be due to increase in knowledge/awareness about the subject matter).

Next point is related to data itself. It seems that your data has a lot of disturbance. I suggest you to check the variability of data and variance. It may be possible that it has few outliers also.

Furthermore only reusing the questionnaire items and running factor analysis cannot be termed as confirmatory factor analysis. If you have already ran exploratory factor analysis and obtained 24 factors then fine you can move on to confirmatory factor analysis using either AMOS, LISREL, MPLUS or PLS. Before conducting the CFA please check the possibility of second or third order factors in your item set. At last if you have not conducted exploratory factor analysis yourself then first obtain a proper structure and then attempt CFA.

Best Wishes

Pl look at the Eigen Value, increase the cutoff from 1 to 1.5 as SPSS by default has 1 to be its Eigen cutoff. This would minimize your number of factors.First try EFA(Exploratory Factor analysis) and then try CFA.

If the sample is normalized then use MLE not PCA

Rotation with Vairmax or orthagonal

AMOS will perform CFA to check on the construct and identity of model fitness, although CFA is also commonly analyzed using LISRAEL

You can also try SAS for both EFA and CFA. Pl find the attachment.Hope this is helpfull

You can perform a "quasi"-confirmatory factor analysis in SPSS by (a) specifying the number of factors to retain to 24, and (b) rotating the factors to a predefined target that is based on the measurement model of the test/scale/questionnaire. Target rotation (also referred to as Procrustes rotation) is not one of the menu options in SPSS, but syntax has been written for it, which makes it possible to do it in SPSS (see Van de Vijver & Leung (1997). Methods and data analysis for cross-cultural research. Sage). This approach is popular among personality psychologists. The idea is to use previous theory to guide the factor analysis. A full explanation is given in: McCrae et al. (1996). Evaluating replicability of factors ..... Journal of Personality and Social Psychology, 70, 552-566.

As far as i know SPSS only helps your to figure out the domains using EFA(Exploratory Factor Analysis).Pl try using SAS which is very robust in estimating CFA.

Pl find the following document .

SPSS run exploratory factor analysis based on principal components and ML.

The pathway for this analysis:

Data reduction / factor analysis

Best regard

1.0 Generally, SPSS runs EFA. But outputs from factor or component analyses can be used to calculate AVE (average variance extracted) and other stats to 'advance' EFA into CFA.

2.0 54 instead of 28 factors - share with our community here your scree plot and Eigenvalue. That shall give us some clues whether that 54-factor solution is generally acceptable. Governed by a good theory, you can 'force' SPSS to extract 28 factors.

3.0 AMOS is normally bundled with SPSS. You can use this app to run CFA. Given that your preliminary extraction yields 54 instead of predicted 28 factors, my concern is your data points are widely scattered which can create identification issue (your model cannot converge) in AMOS.

You can also use R to conduct CFA. Has better math than CFA on AMOS.

I agree that Amos is the better solution. Given that CFA is part of the larger family of SEM, it usually plays an essential role in evaluating the measurement model before a structural analysis is conducted. Structural analysis is then used for specifying and estimating models of linear relationships between both observed and latent variables.

I actually don't know Smart PLS , but Amos is very popular in all the good journals so if you want to publicize your work, Amos is better.

You can also use R which is a free ware. In fact R is more reliable for CFA than AMOS. There is also a free student version of LISREL which is more reliable for CFA compared to AMOS.

Hello researchers,

I have a very basic question :

how to run exploratory factor analysis in SPSS?

Thanks.

You have to enter all the items from your SPSS data file that you are factoring.  There are several excellent tutorials for how to run SPSS EFA on YOUTUBE which show you step by step how to do this and then how to interpret your results.

 there is no way to run CFA within  SPSS. However you can run EFA and restrict the number of factors to be extracted to 24 and see if you reach a satisfactory solution. To run CFA you can use AMOS (which belongs to the same company as SPSS but requires a separate and expensive license) or use shareware like R.

There is a free student download version of LISREL but it handles limited factors.

there is a free 14-day trial version on AMOS, fully functional.

efa_class is available for free, if you're looking for free software that would work. Description is here: http://cran.r-project.org/web/packages/semTools/semTools.pdf

You can use statistica for CFA.

lavaan is free and easy to use:

LV =~ x1 + x2 + x3

see:  

ROSSEEL, Y. The lavaan tutorial. Belgium: Ghent University, Department of Data Analysis, 2015. Disponível em: . 

lavaan_site: 

Great discussion !! Please also answer my queries...

@Rakesh: I couldn't completely understand "It would be better if one adopts several criteria to decide the number of factors such as scree plot, parallel analysis, Velicer's MAP. C-Hull method etc."

For factor extraction in EFA, I am aware of the following criteria: eigen vales (generally >1), TVE (>60 %), communities (>0.5), anti-image variance diagonal elements > 0.6,  less than 50 % residuals to be > 0.5, apriori knowledge/preference of #factors supported by ROL (forced EFA)

Plz explain parallel analysis, Velicer's MAP. C-Hull method.

Also, someone mentioned: SPSS can be used only for EFA (NOT CFA) except for version 19 or higher. Plz share some resource on how to use SPSS for CFA. Do I need some plug-in?

@Paul Ingram: which package would you recommend for factor analysis in R, I am considering FactomineR or rcmdr

But I am getting trouble interpreting the output (it's different from SPSS). Would you please share some resource for the same. Thanks

Hello. I am currently performing confirmatory factor analysis for my research work. 

I am using software R for the same.  There exists a number of  helpful literature on performing CFA with R. I have attached one paper I found extremely helpful. Hope this helps. To what I know of, CFA can't be performed on SPSS. I may be incorrect here. AMOS student version is also available online but there is a time limit to its use. 

Best wishes with your work 

Confirmatory factor analysis with Amos program.

I think you need to install Amos in order to perform CFA

Please see the attached Video below

https://www.youtube.com/watch?v=JkZGWUUjdLg