2

Consider a project in which you have 120 participants answering a questionnaire about perceived quality and perceived price of several variations of a product.

Each participants gets introduced to a picture of 12 different variations of a product (this could be a color change, geometric change etc).

We therefore have 120*12 = 1440 observations.

Can I run a linear regression on a parameter such as "perceived price" and then product variation 1-12 to see if there is a significant difference in perceived price?

I feel I am violating some law of regression seeing as there are only 120 people participating and their answers may be biased for the rest of their observations, while the observation count is 1440.

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
Paze
  • 1,751
  • 7
  • 21

1 Answers1

1

You should not use linear regression, the assumption of independence would be violated. Look into repeated measurements analysis, in R this can be done with the lme4 package (and others.)

Examples and references can be found here Example reports for mixed-model analysis using lmer in biology, psychology and medicine? and otherwise at this site.

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467