FW: another multiplication of items questions

[Josh Wenk]

Title: Best way to analyze: Importance & Satisfaction data!

 

-----Original Message-----
From: Josh Wenk
Sent: Wednesday, August 01, 2001 10:18 AM
To: [EMAIL PROTECTED]
Subject: another multiplication of items questions

let's say I have a big list of objects, each rated on how well repondents KNOW that object [0=don't know, 1=kinda know, 2=know, 3=know well], and, if they know, how well they LIKE that object (-2 to +2).

 

 

				LIKE
		-2	-1	0	1	2
	3
KNOW	2
	1
	0

 

One issue, is that the variables are on diff scales, so I would want to "normalize" (excuse the misnomer) the scale to each range in abs val from 0 -1, so then  the scale would be like:

 

				LIKE
		-1	-0.5	0	0.5	1
	1
KNOW	2/3
	1/3
	0

 

so that each variable has equal potential effect on resultant score.

 

 

If i want to come up with a single (that is, 1-D) Score, how much does it make sense to do an avg of the products????

 

ie, weighted avg of products = (the product of the Row and Column value for know and like) * (% of respondents who chose that combination).

 

 

 

 

 

other potential analyses under consideration include doing 2-D graphs, tests, etc, BUT it would be really nice to have this 2-D info "densified" to 1-D if methodologically poss/valid.

 

 

Thanks in advance

 

 

 

 

-----Original Message-----
From: Seth, Sanjay [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, August 01, 2001 7:01 AM
To: [EMAIL PROTECTED]
Subject: Best way to analyze: Importance & Satisfaction data!

Hello Everybody,

I am wondering if you could give me your input in terms of best strategy to analyze data that contains rating (1 to 5) scores on both "Importance" and "satisfaction" with 30 attributes for four brands.

                Brand A                         Brand B
Item             Importance  Satisfaction           Importance    Satisfaction
1                         2             5               3               1
2               3               2               5               5
3               1               5               2               3
4               5               1               1               5

Does it make sense to multiply "importance" score with "satisfaction" before analyzing the data (i.e.,  factor analysis and regression etc.)?

Look forward to hearing from you all.

Best regards,
Sanjay

Japan