Samples
Starting Point Bias in Bidding Games
be called the exasperation effect and the
information transfer effect.
Rowe et al. tested the effects of using
three bidding starting values ($1, $5, $10)
in willingness-to-pay (WTP) and willing-
ness-to-accept-compensation (WTAC)
question formats. The impact of starting
bids on final bidding outcomes was tested
by estimating two separate bid curves to
explain variation in observed WTP and
WTAC final reported bids. In both the
WTAC and WTP regressions, estimated
coefficients associated with starting value
variables were positive, indicating a direct
relationship between final bids and start-
ing bids. However, only the starting value
term appearing in the WTP bid equation
was statistically significant at the 0.10
level.
Thayer experimented with $1 and $10
starting values using WTP question for-
mats. Mean final bids for alternative start-
ing values were compared using one-way
analysis of variance. Although they were
not shown to be significantly different,
mean final bids were found to be inversely
related to starting values used to initiate
the bidding process.
Brookshire et al. (1981) tested for start-
ing point bias in bidding games used to
measure willingness to pay for air quality
improvements. Three starting values ($1,
$10, $50) were experimented with. Mean
final bids obtained for 12 residential areas
were compared. A total of 36 pairwise
t-tests of differences between means were
conducted. In 30 of the tests (83 percent),
the hypotheses that the mean bids were
the same could not be rejected at the 0.10
level.
More recently, Desvousges et al. re-
enon. The bidding procedure they employed in-
volved raising bids of $1 rather than by the more
traditional 25 cent increments. This was done to
help ameliorate the exasperation effect by more
rapidly reaching the final bid amount. No explicit
tests of starting point bias were reported by the
authors.
ported detecting starting point bias in a
study of option prices and user benefits
associated with discrete water quality
changes. Two starting values ($25, $125)
were experimented with in WTP bidding
game question formats. Research findings
indicated that mean user benefits were 4.6
to 5.5 times higher when starting values
of $125 rather than $25 were used to ini-
tiate the bidding procedure. A positive re-
lationship between mean final bidding
outcomes and starting values was consis-
tently observed. However, statistical test-
ing to detect differences between mean
bidding outcomes yielded inconclusive re-
sults.
Further Empirical Evidence
Further testing of starting point bias was
conducted in the context of an experi-
mental study of marine recreational fish-
ing valuation in Hawaii.2 A sample of 100
offshore sport fishermen was randomly se-
lected to participate in personal inter-
views conducted by professional market
research interviewers. This group was fur-
ther randomly assigned into two subsam-
ples, each comprising 50 anglers. Anglers
comprising Subsample I were queried
about their willingness to pay a daily fuel
tax and their willingness to accept com-
pensation not to fish for a day.3 An itera-
tive bidding procedure was used to elicit
2 A complete description of survey procedures and
results can be found in SMS Research.
3 The exact wording of the questions is as follows:
“Suppose that you were going to fill up your boat’s
fuel tank to go out fishing the next day. You hear
that a new tax has been placed on fuel used for
sportfishing. Would you go ahead and buy the fuel
so that you could go fishing if the tax increased the
cost of fishing trip by $?”
“Finally, imagine that the day before you are
planning to go offshore sportfishing, you find out
that all sportfishing trips for the next day will have
to be cancelled because of top secret Navy opera-
tions. However, you will get a cash reward to make
up for the trouble caused you. Would you be sat-
isfied with a cash reward of $if you could
not go offshore sportfishing as planned?”
33