We are trying to use multiple-linear discriminant analysis as a method to measure the degree to which seven independent variables contribute to a linear function that best discriminates between 3 items in one dependent variable. From my preliminary reading, I found that this method can accommodate mixed independent variables (Kohli and Devaraj, 2003) and is suitable to situations where the dependent variable is categorical (Lee 2004). It can explain relationships between multiple independent variables and one categorical dependent variable (Kohli and Devaraj, 2003). With discriminant analysis, a linear combination of independent variables that will discriminate best between predefined groups can be derived (Hair et al., 1992), which is exactly what we trying to do.
I would be thankful if anyone can give some hints on how to use this method.
Many thanks in advance
Ali
Please take help from the following:
1. Any book on Multivariate analysis
2. Any book on Advance Research Methodology
3. SPSS Package
4. Any other customized statistical package, etc.
Dear Ali,
Maybe you will find this chapter:
http://www.uk.sagepub.com/burns/website%20material/Chapter%2025%20-%20Discriminant%20Analysis.pdf
useful?
I would also recommend this book:
http://www.crcpress.com/product/isbn/9781439816806
Hope this helps.
Best,
tom
Hi Ali,
The book by Hair et al "Multivariate analysis" is definitely the best starting point. I see you have already read the 1992 edition, however I suggest you go to the last one, there are some additional info and examples that will be helpful.
This document can be helpful as well: http://spu.fem.uniag.sk/cvicenia/ksov/sojkova/Viacrozmerna%20sttatistika/Discriminant%20Analysis.pdf
hope it helps. Regards.
Dear All,
Thanks for your great suggestions, Hair's book (all editions) was like a Bible for me and still. The problem is HOW to analyze the data more than understanding the method itself.
I tried youtube videos but did not find that much there. Any suggestions would be much appriciated
Again thank you so much for your suggestions and sorry for misleading you !!
You may follow this link https://lirias.kuleuven.be/bitstream/123456789/122365/1/0575.pdf and related ones.
Thanks
Dear Ali
the following reference can help you
Tabachnick, B.G., & Fidell, L.S. (2007), Using Multivariate Statistics, 5th edn, Boston, Pearson education Inc.
Hair, J.F., Black, W.C., Babin, B.J., Anderson,R.E. and Tatham, R.L. (2006), Multivariate Data Analysis, 6th edn, Prentice Hall international, Upper Saddle River, New Jersey, USA.
Cohen, J. & Cohen, P. (1983), Applied Multiple Regression/Correlation Analysis for the Behavioural Sciences, 2nd edn, New Jersey, Lawrence Erlbaum Associates.
Malhotra, N.K. (2004), Marketing Research: an applied orientation, 4th edn, New Jersey, Pearson Prentice Hall. Upper Saddle River, USA.
Pallant, J. (2005), SPSS Survival Manual: a Step by Step Guide to Data Analysis Using SPSS Version 12, 2nd edn, Open university Press, McGraw-Hill Education, NY.
Thank you Ali for bringing out this question because this group of colleagues are superior in their generosity to provide information.
"Multivariate analysis" by Hair et al and "Marketing Research" by N K Malhotra are good books.
Dear Karuthan,
This is exactly what I am looking for !! Thank you so much :)
Maybe this is interesting to you: https://www.researchgate.net/publication/225718403_Linear_Discrimination_Ordination_and_the_Visualization_of_Selection_Gradients_in_Modern_Morphometrics
Article Linear Discrimination, Ordination, and the Visualization of ...
Hi Karuthan, is it possible that you share your notes with me too?
Thanks
Initially, you can use notes of SPSS package. That gives explanation of the concept and how to use "multiple-linear discriminant analysis" .
You may be interested in this
http://math.arizona.edu/~hzhang/math574m/Read/LogitOrLDA.pdf
which considers the circumstances under which logit modeling is preferred to discriminant analysis
and this is a short summary of the arguments
http://www.kmentor.com/socio-tech-info/2003/12/what-is-the-difference-between.html
You can use multinomial logistic model to analyze more than 2 outcomes.
To be honest - I have never used discriminant analysis and yet routinely use logit models. Indeed when I was taught the latter (in the mid 1970's !!) discriminant analysis was seen even then as rather old-fashion and superseded by the whole family of what became to be called generalized linear models [ Nelder, John; Wedderburn, Robert (1972). "Generalized Linear Models". Journal of the Royal Statistical Society. Series A (General) (Blackwell Publishing) 135 (3): 370384.doi:10.2307/2344614]
But I really did not know why one would be preferred to the other. Your question prompted me to dig a bit and I now understand that the logit models offer flexibility in analyzing non-normally distributed predictor variables and the area in which I work often has categorical predictors.
This simulation study is clear
Comparison of Logistic Regression and Linear Discriminant Analysis: A Simulation Study Maja Pohar , Mateja Blas , and Sandra Turk
http://mrvar.fdv.uni-lj.si/pub/mz/mz1.1/pohar.pdf
They conclude
"we can say that LDA is a more appropriate method when the explanatory variables are normally distributed. In the case of categorised variables, LDA remains preferable and fails only when the number of categories is really small (2 or 3). The results of LR, however, are in all these cases constantly close and a little worse than those of LDA. But whenever the assumptions of LDA are not met, the usage of LDA is not justified, while LR gives good results regardless of the distribution. As the estimates for LR are obtained by the maximum likelihood method, they have a number of nice asymptotic properties as well."
So I shall be sticking to my logit models for the analysis I undertake. So thanks for the question - I am better informed.
Thanks Kelvyn for enrishing our knowledge.
To be honest, i never used either of the two methods to anlyze my data before. I am more into using Structural Equation Modeling. But, SEM is not appropriate for situations. I attached the research-in-progress paper that we are currently extending for your reference. As you will notice we only have one dependent variable (categorical type) which make it impossible for us to use SEM. we are about to finish the data collection stage. so, based on your recommendation, LDA is more preferable when we have normal distribution then discriminant analysis . I will definitely take that into consideration before making up my mind.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics