Contignet valuation - when negative mean WTP values occur in model

[Eileen Mitchell]

Hello all,

thanks for viewing.

I was wondering could anyone suggest a good model to use or advice generally what to do when you get negative WTP mean values?


When no control variables are included in the question the average WTP is -0.92 euros (bid amounts were initial 0.5-5.00 euros),

answer1 = 1 when the respondent answers yes to the initial bid amount

When i ran the probit model for only the bid amount : The bid variable whilst significant - as the bid goes up the probability of a positive answer decreases.

command: probit answer1_cntl bid1_cntl

bid1_cntl: coef= -0.16 se= 0.04 z=-3.69 p=0.00 (-0.24, 0.075)
_cons: coef=-0.149, se=0.111 z=-1.35 p=0.178 (-0.36, 0.068)


However, when no control variables are included in the question the average WTP is -0.92 euros (bid amounts were initially between 0.5-5.00 euros)

command: nlcom (WTP_cntl:- _b[_cons]/_b[bid1_cntl]), noheader

WTP coef= -0.91
se =0.89
z = -1.04
p = 0.3
(-2.68, 0.82)


This was based on Lopez contingent valuation using stata method:
https://mpra.ub.uni-muenchen.de/4101...aper_41018.pdf


I am not sure how to interpret this when the outcome is negative.

[Carlo Lazzaro]

Eileen:
the guru in this field is Arne Risa Hole (https://www.sheffield.ac.uk/economics/people/hole).
Skimming through his Stata threads and web page is my advice.

[Eileen Mitchell]

Thanks Carlo