Ordinal data+ independent factors+Marketing analysis+SPSS= HELP!!!!

#1
Hi,

I desperately need help analyisng my survey data.

Which tests do i use in order to test the relationship between my variables, and to see if my dependent variable can predict bahaviours in others.

Here is the deal:
I am investigating 6 products (independent) (i think this is cathegorical). They are individually rated on a 5 point Likter scale (1:very unattractive- 5:very attractive) by the participants in the survey.

I want to know if there is a relationship between the rating of attractiveness and consumers willingness to purchase. Participants indicate which (one) product they are most willing to purchase by selecting that product. (product 1 coded numerically as 1)

Furthermore I want to se the relationship between the products attractiveness and their percieved quality. The percieved quality is rated on a likter scale (much worse- much better) for each product in the same question.


So far I have understood, I cannot use a factor analysis because the rating is independent according to product
I cannot use correlation or regression analysis, beacause apparently my data is ORDINAL.
Is data still ordinal when they have been assigned numeric values??

I want to determine a relationship and test if product attractiveness is a predictor of the following two.

PLEASE help me!!!!


Or better, if you are a marketing research specialist and want to analyse my data for me I will not let it go unnoticed!

cheers!
 

spunky

Can't make spagetti
#2
I want to know if there is a relationship between the rating of attractiveness and consumers willingness to purchase. Participants indicate which (one) product they are most willing to purchase by selecting that product. (product 1 coded numerically as 1)
whether you are working this one out as 1 = willigness to purchase and 0 = unwilligness to purchase or as an ordered response set (1 = most willing to be purchased, 2 = second most willing to be purchased, etc.) you can try some sort of grouping method here, like binary or multinomial logistic regression.

Furthermore I want to se the relationship between the products attractiveness and their percieved quality. The percieved quality is rated on a likter scale (much worse- much better) for each product in the same question.
do you want to see whether they are just related or if quality can predict attractiveness (or vice-versa). if it's just relationship, you can do a spearman's rho if the likert-scale thing is killing you, but these things get treated as continuous so often nowadays that i wouldnt worry too much about using traditional parametric methods to analyse your data.


Or better, if you are a marketing research specialist and want to analyse my data for me I will not let it go unnoticed!
you're just too funny... :)
 
#3
Did you know that there are survey data analysis solutions out there that actually select the right stats test for you? This is great if you don't have a heavy stat background and don't have the time to learn how to setup an application. Also - many of these solutions are web-based, so you don't need to load software on any one specific computer. Worth investigating if you want to streamline the analysis process and get to your summary/recommendations faster!
 

spunky

Can't make spagetti
#5
An analysis completely devoid of any thought? yuck.
welcome to my world... but at one point i think you just stop caring... besides, those webpages... ha! depending on which page you go to you can get different analyses so...
 

noetsi

No cake for spunky
#6
welcome to my world... but at one point i think you just stop caring... besides, those webpages... ha! depending on which page you go to you can get different analyses so...
Dason must not work much with politicians :) My world too much of the time.
 

noetsi

No cake for spunky
#8
I used to be a purist (in my own research area which was not statistics). But I have concluded that 1) sometimes some knowledge even if imperfect is better than none (probable reduction in error :) ) and 2) academics are so uncertain of so many things (including their methods) that data without well established theory is acceptable. We create our own theories for it in interpretation. Its impossible not to.

Anyone here can tell that I just got done with a massive SEM project :)