Angler choice experiment and a RPL

New Zealand Trout Angler
Choice Experiment:
A RPL-Error component model
By Stephen Beville
Supervisors: Geoff Kerr & Ken Hughey
Presentation Structure
► Context
► Survey
for the Choice experiment
Design and Implementation
► RPL-Error
component model
► RPL-Error
component with decompositions
New Zealand’s recreational trout
fisheries
1. Changing fishery conditions
2. Changing patterns in fishery usage
Backcountry Rivers,
Lakes
Lowland Streams
Mainstem Rivers
Experimental Design
-Bayesian D-efficient
-Unbalanced
-Constrained
-16 blocks of 6 choice scenarios
-randomized choice sets
Response
► Survey
link sent to 6500 email addresses
with 1 reminder email.
► 813
responses in 2.5 weeks
RPL-Error component
r-sq=.167
Explaining sources of Heterogeneity
in and RPL-error component
► Re-parameterized
RP means, RP variances and
latent effects variances using variables which
identified anglers level of specialisation
(Greene & Hensher 2006)
► Variables
created via an index which included:
-behavioural, skill, cognitive, affective dimensions of the
recreational specialisation construct.
Recreation Specialization Theory
Bryan (1977)
► As
anglers specialize:
-preferences shift from consumption toward
conservation of fish stocks
-representations of the activity may become more
complex
-preferences may become more heterogeneous
“Multiple trajectories” from a single starting point
Sources of Heterogeneity
Conclusions
► Error
component extension can add valuable
information
► Decompositions
of parameter means,
variances can add valuable insights.
However, loss of parameter richness and
sign change may occur.