I ran multiple regression using normalized (ZCDiff obtained by HB/OLS) individual utilities and am getting odd results.
For instance, people with positive attitude toward organic products significantly prefer imported products.
People with positive attitude toward organic products presented less preference toward locally grown ingredients.
I was wondering whether this happened as the normalized utilities for certain individuals are negative values?
I initially ran correlation on multiple IDV (measured using 7 point likert but averaged them) for utility values per levels within each attribute, 13 levels.
Upon finding only 4 levels had correlation, I ran MRA individually for each utility value.
For instance, my survey has 5 attributes with total of 13 levels.
What would be the best way to ensure that the results are valid or not?
As individual utilities are interval scale, if I end up adding the same value to all the ZCDiff utility values to get rid of negative values, and re-run regression? Is this wrong?
I deeply appreciate your feedback.