Take the 2-minute tour ×
Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. It's 100% free, no registration required.

I am currently doing my MBA and I am attempting to find out what variables affect new student intake at the university which I am working at. My model is based on the 7Ps of the marketing mix (Product, Place, Price, Promotion, People, Process and Physical Evidence) and non-marketing mix factors (such as influence of people, influence of media, influence of the internet, etc). Essentially I want to find out

  1. what impact the 7Ps have on new student intake,
  2. what impact non-marketing mix factors have on new student intake,
  3. what impact both the 7Ps AND non-marketing mix factors combined have on new student intake.

I will be sending out questionnaires and for each dependent variable (e.g. each 7P and non-marketing mix factors) there will be a variety of 'sub' variables. For example, under price, the 'sub' variables are: tuition fees, development fees, cost of study materials, cost of field trips, cost of university provided food, cost of study related clothing, payment arrangements, method of payment and early bird payment discounts. Each variable has at least 4 'sub' variables.

Can multiple regression analysis find out which 'sub' variables are valid or not? I.e. which are significantly correlated to the main variable? Going back to price, maybe only three or four of the variables are sufficiently correlated to have any impact on price, which then impacts on new student intake.

From what I understand, multiple regression is best suited if the main variables do not have 'sub' variables. Or should I use SEM (AMOS)? Please describe when multiple regression should be used and when SEM (AMOS) should be used. I do not want to use an inappropriate data analysis technique!

share|improve this question
    
This is not a direct answer to your question, but I'd recommend you pay very close attention to issues of measurement, of reliability and validity. Even before, or at least parallel to, your investigation of mult. regression and SEM, you'll want to look into scale development, factor analysis, and principal component analysis. To get accurate estimates of a relationship, you'll need to have the most accurate measurement possible of each construct. This can be an intricate and fascinating process when you have many constructs, each potentially consisting of many others in turn. –  rolando2 May 9 '12 at 1:59
add comment

1 Answer

Before delving into any data collection I would recommend you to read at least a basic chapter on what factors and/or latent variables are. Then I'd suggest you to read what SEM does do!

Take a look at the latest edition of Hair et al Multivariate Analysis http://www.amazon.com/Multivariate-Data-Analysis-5th-Edition/dp/0138948585

share|improve this answer
    
To do some reading is what I was going to recommend. If you have an opportunity, get a short course on structural equation modeling. What you are describing is a textbook application of latent variables, but you would have to figure out how people in this field talk, and what the common ideas are. –  StasK May 9 '12 at 2:28
    
Ok. Thanks. Will look into latent variables. –  James May 9 '12 at 11:41
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.