I am going to measure the caring level of the educated people about the food wastage when they do dining.My dependent variable is caring level for food wastage and independent variables are frequency of food waste,awareness about the food wastage prevention program (Yes/no question), pre idea about the amount of meal before dine.
Yes, Discriminant Analysis technique can be used along with Principal Components Analysis. The approach is as follows: The dependent variable in Likert scale can be treated as a categorical variable while your two Likert scale independent variables can be combined into one interval data using the scores from a Principal Components Analysis of those two variables.
On this premise, discriminant analysis becomes appropriate. Discriminant Analysis requires a categorical dependent variable and interval data independent variables.
In this situation the independent variable is used as the basis for classifying or discriminating between the food wastage levels which is the dependent variable. That is in fact your stated objective.
Dear Yasantha,
Multivariate analysis is not a pre-determinant of your tool utilized in the study. Likert scales are often analyzed as continuous variables or could be dichotomized to find associations between independent variables. If your dependent variable from your likert scale score is analyzed as a continuous variable, you may use multivariate linear regression analysis. If it's analyzed as a categorical variable, use multivariate logistic regression analysis. Three principles that qualifies for multivariate analysis are:
1. Principle of statistical significance at univariate level
2. Principle of parsimony
3. Principle of biological availability
Good luck!
Yasanthia, I believe that your question is in order.
The issue of converting your Liket scale data to a higher order continuous variable is inappropriate statistically. In any case ; if you eventually do logistic regression; you will still be involved in multivariate analysis. So I see no problem with your objective of utilizing the powerful attributes of mutivariate analysis.
Dear Yasantha, please note that since you actually have three independent variables and not two as I had mentioned, you may or may not end up with just one set of PCA scores of interval data independent variables. Whatever the number of sets of scores, the powerful discriminant analysis technique will achieve your objective. It is in fact the statistical analysis that will best solve your problem. Regression analysis is not usually used for classification, grouping or discrimination. It is mainly used for prediction. Thanks
- Badges
- Science topic
- Similar topics
- Medicine
- Nutrition and Dietetics
- Nutritional Medicine
- Food Science
- Food