Injustice - Developers Among Us (SciFiDevCon 2024)
Structural Implications of Destination Value System Networks
1. ENTER 2017 Research Track Slide Number 1
Structural Implications of
Destination Value System Networks
Jason L. Stienmetza and Daniel R. Fesenmaierb
aUniversity of Surrey, UK
j.stienmetz@surrey.ac.uk
bUniversity of Florida, USA
drfez@ufl.edu
2. ENTER 2017 Research Track Slide Number 2
Destination Value System
A tourism destination can be
considered as a constellation
of visitor-centric networks
which represent core value
creation processes:
• exist in both physical and
virtual space
• are focused on capacity
• are constantly evolving
• occurring simultaneously
as part of an integrated
system
• Supply meets demand
Separate networks for each core value
creation process: marketing,
sales/distribution, collaboration,
activities/experience co-creation
3. ENTER 2017 Research Track Slide Number 3
• Destination touchpoints (value creating interactions) are represented by
nodes in the network.
• The sharing of resources between touchpoints (such as the movement of
travelers from one activity to another) are represented by ties connecting
nodes in the network.
• Network structure (i.e. the ways in which the distinct components and
processes of a destination are connected and organized) constrain the
actions of the actors in the network.
Network structures of a destination will influence how value is created
within a destination.
• Previous studies have been conducted at the level of analysis of individual
actors within simple destination systems
Key Concepts -
Network Theory
Main Research Question
What is the relationship between the structure of the
network and total value creation within a destination?
4. ENTER 2017 Research Track Slide Number 4
1. Establish a conceptual framework
of destination value systems
2. Provide empirical evidence that
DVS network structure influences
the value created within a
destination and identify ideal
network structures that can
maximize tourism expenditures.
3. Validate emerging methods for
measuring technology mediated
travel and the structure of
destination value systems.
Research Objectives
5. ENTER 2017 Research Track Slide Number 5
Summary of Hypotheses
H1: DVS network density will have a significant positive effect on destination
value creation.
H2: The out-degree centralization of a DVS network will have a significant
positive effect on destination value creation.
H3: The in-degree centralization of a DVS network will have a significant
negative effect on destination value creation.
H4: The betweenness centralization of a DVS network will have a significant
positive effect on destination value creation.
H5: Global clustering coefficient of a DVS network will have a significant
positive affect on destination value creation.
H6: The number of network clusters (i.e. subcommunities) in a DVS network
will have a significant positive effect on destination value creation.
H7: Seasonality, measured by quarterly time period, will have a significant
and variable impact on total value creation within a DVS network.
6. ENTER 2017 Research Track Slide Number 6
• Data was created utilizing
metadata from 4.6 million Florida
photographs with GPS
coordinates shared on Flickr from
2002 to 2015
• Photo metadata were
downloaded using the Flickr API
between March 1, 2016 and April
15, 2016
• Volunteered Geographic
Information (timestamp, latitude,
and longitude) were aggregated
to form network structures of
traveler activities within Florida
destinations
Research Methods
7. ENTER 2017 Research Track Slide Number 7
• Touchpoint (node) = place where photos were
taken by at least 5 Flickr users
• Trip segment (tie) = same user taking photos at
two separate touchpoints during the same week
(Wednesday through Tuesday).
• Traveler = “someone who moves between
different geographic locations for any purpose and
any duration” (UNWTO, 2010, p. 9)
– No exclusions based on residency or length of
stay
• Destination = Florida counties, by quarter
• Value = total taxable tourism related sales (Florida
Department of Revenue, 2015)
Operational Definitions
8. ENTER 2017 Research Track Slide Number 8
Creating DVS Networks
1. Photos Uploaded
2. VGI data
3. Individual Activated Paths
4. Path Aggregation
9. ENTER 2017 Research Track Slide Number 9
Big Data Reliability
Flickr vs Survey Data
• Survey: St. Johns County Tourism Development Council Visitor
Survey 2012-2013
– 6,334 Responses
– Demographics
– 34 attractions listed
• Created an undirected (no sequence) activities network
• 11,330 photos (from 966 users) obtained from Flickr were
geolocated to the same 34 attractions measured in the
survey.
10. ENTER 2017 Research Track Slide Number 10
Network Comparisons
Survey-based
Traveler Activities Network
Flickr-based
Traveler Activities Network
11. ENTER 2017 Research Track Slide Number 11
VGI Data
Quality Diagnostics
Network Metrics
Flickr
VGI
Visitor
Survey
Network Density .319 .977
Average Number of Touchpoints Visited per Person 2.0 3.2
Spearman Correlation: Nodal Weighted Degree rho = .668, p < .001
Spearman Correlation: Nodal Weighted Betweenness rho = -.365, p = .034
Spearman Correlation: Nodal Weighted Cluster Coefficient rho = .324, p = .062
Overall Network Q.A.P. Correlation r = .640, p < .001
12. ENTER 2017 Research Track Slide Number 12
Attendance Comparison
10
12
14
16
18
20
22
24
15
20
25
30
35
40
45
x10000
NPS Visitor Counts vs Flickr Photos
Flickr NPS
13. ENTER 2017 Research Track Slide Number 13
• Flickr VGI data must be sufficiently large to
reliably measure traveler activities networks
• Aggregated touchpoints to 2.5 acre grid
(about 1 city block)
– Increased quality of overall network
measurement
• Only included photos taken in 2007 or later
(too few observations in years prior) for 36
quarterly observation periods
• Eliminated destinations with less that 10
quarterly networks
• 43 out of 67 Florida counties represented
• igraph for R used to calculate metrics for
1,364 DVS networks
Quality Assessment
& Cleaning
14. ENTER 2017 Research Track Slide Number 14
DVS Network Examples
Gulf 4Q2014 Flagler 1Q2011 Walton 2Q2013
Okaloosa 2Q2014 Hillsborough 1Q2012 Orange 1Q2008
15. ENTER 2017 Research Track Slide Number 15
Model Estimation (n=1364)
Feasible Generalized Least Squares (FGLS) Estimator – fixed effects that control for the
unique characteristics of each destination (unique intercept for each destination)
𝐿 𝑇𝑅𝑆𝑖𝑡
= 𝛼𝑖 + 𝛽1 𝐷𝑖𝑡 + 𝛽2 𝐼𝐷𝐶𝑖𝑡 + 𝛽3 𝑂𝐷𝐶𝑖𝑡 + 𝛽4 𝐵𝐶𝑖𝑡 + 𝛽5 𝐺𝐶𝐶𝑖𝑡 + 𝛽6 𝐶𝐶𝑖𝑡
+ 𝛽7 𝑄2 𝑡 + 𝛽8 𝑄3 𝑡 + 𝛽9 𝑄4 𝑡 + 𝑢𝑖𝑡
16. ENTER 2017 Research Track Slide Number 16
• Network Structure does matter!
– Statistically significant relationships
found for all structure variables
– Foundation for further DVS model
development
• Unexpected results for Density, In-Degree
Centralization, Out-degree Centralization,
and Global Clustering Coefficient
– Further researcher needed to better
understand both cooperation and
competition effects
• Network dynamics play an important part
in destination value creation and must be
part of future DVS model development
• Foundation for “Smarter” Destination
management
Discussion
17. ENTER 2017 Research Track Slide Number 17
• Destinations can be defined by visitor
behavior rather than administrative
boundaries (Beritelli et al., 2013)
• Aggregation of destination
touchpoints assumes homogeneity
• Destinations with fewer VGI
observations remain less understood
• Structuralist approach does not
explain antecedents of network
structure nor how to best manage
destination networks (Borgatti &
Foster, 2003)
– Future research is needed
Limitations