How to JOINTLY perform multiple imputation with normal dataset and multiple response sets in SPSS? And right kind of Cluster analysis for binary data?

#1
down votefavorite
(Unfurtuntely i can't upload the excel file ,can't find the button)
Please im begging anybody’s help because im an mba student in global hospitality and tourism management and my teacher is unable to help I have purchase the spss premium version so I can use all its features I carried out a survey of Chinese tourists at the airport in order to perform pattern recognition and analysis about them, so I created this small-sized database whose cases are 130 tourists which departed from Milano, Italy in December. There is quite a number of variables Here you have the attached file, my dataset, made from multiple sheets. The first one if the raw datasheet, with variables not split. The second one is the main file actually supposed to be imported into spss. The remaining are multiple response data sets which cannot can be imported as well, but I want them to me analyzed JOINTLY alongside the main sheet. I wish I could merge the sheets into a single sheet to be entered into spss, I have 6 multiple response sets, whose missing values have to be imputed through multiple imputation indeed, but I need to bind the results and the relationships to the main dataset which will be multiply imputed according to the normal spss procedure. They will have to be dummy coded into spss I believe
Ive browsed the web all over searching for papers to help me, i even red lots of discussions on this, it took me more than 1 month to read all of it and for subsequently analyzing those. But so found I found nothing helpful in my case. I have almost no background in statistics that’s why I’m struggling I wish I could merge the sheets into a single sheet to be entered into spss, or at least find a way to bind the algorithm so as to include all my sheets and the procedure to take into account PATTERN At first I thought I had to use subscales, but then I found out that these do refer do multiple non-adjacent items, actually belonging to different questions Does anybody know of any string or manually entered code which will allow me to do this? Or a way to enter the excel sheet into spss or to merge them? How to set up the spreadsheet? If not the multiple imputation will think that different excel sheets are different datasets when they are actually the same I’m afraid these data are not jointly analyzable, I really hope to be wrong, otherwise this will result in data loss on my part

I’ve red these paper which seem not to be helping -Addressing Item-Level Missing Data: A Comparison of Proration and Full Information Maximum Likelihood Estimation -Multiple Imputation by Chained Equations: What is it and how does it work? And on top of that, I can’t find this paper online and authors are unreponsive -A Comparison of Item-Level and Scale-Level Multiple Imputation for Questionnaire Batteries So I supposed that I should go on analyzing the main data set which is made from nominal, scale, dichotomous and ordinal variables.
After the data being solved that I will have to run the multiple imputation itself, but perhaps another kind of analysis is best here. Perhaps factor analysis, but it seems to me that its actually useful for classify variables, not cases Does missing value analysis tell me whether the data are MCAR or MAR MNAR? An user pointed out that data at the end of the day there is always some pattern , and thus no data are truly missing at random If the data are MCAR then listwise deletion is a feasible option, but in order to tell that I need a missing value analysis first. But if the data on the other hand are MAR, then the expectation-maximization could be just as good as multiple imputation. But my teacher told me to perform cluster analysis

After that I will run descriptive statistics, and afterwards I will test the null hypothesis to check whether some hypothesis I will have developed hold true. This will be a lengthy process I guess I’m unsure whether I have to weight cases, but I red it must be done. Perhaps some variables hold a greater weight than others, but im clueless on how to continue. Please help me in this too. It

After that I will have to set the right kind of cluster analysis (since these multiple response sets are indifferent excel sheets), which I reckon can’t jointly be performed on the first dataset either, but just like the multiple imputation it needs to. The same problem explained in the first heading above And also, im unsure on what kind of spss cluster analysis to use, since you can see my data are binary, categorical, nominal and more. It seems to me that k-means clustering can only handle scale/continuous data and is therefore useless here. So it’s either hierarchical clustering or 2-step clustering Hierarchical or TwoStep cluster analysis for binary data?? As this knowledgeable spss developed pointed out, despite what spss website spss, the hierarchical cluster analysis is actually good for binary data I fail to understand the difference between nominal binary and ordinal binary data he points out, which according to him call for different actions, but you can see that since ordinal binary data like mine cannot be analyzed by the 2-step clustering solution, then I don’t know what to do

And what variables should I use for clustering? Will weighting cases affect them? Lastly, I will have to make sure my clusters will be valid, and what should I do for knowing that? If anyone among you professionals could be so kind as to help me out it would be great and, for what is worth it, mentioned in the acknowledgements section of my thesis I’m even willing to hire an expert among you to help me get out this problem, it’s been more than 2 months since im stuck in this situation I’m pretty confused and I therefore ask for people to help me, the procedure is lengthy and I need to graduate
All the best
 
#2
Start by combining your various data into a single place. You can do this in either Excel or SPSS. I suggest Excel. Use the "Combine Worksheets" wizard. (You may need to install the add-in if it is not already installed.) You can also use Syntax do accomplish this in SPSS but it is more complex.

As far as analysis, you are making it too complex. Unless you are assigned to use things like clustering and weighting, do not bother with them. If none of your data are continuous, then stick to something like a Chi square analysis.