dynamics of social interaction frequency: role of geography and accessibility
DESCRIPTION
This presentation won the postgraduate paper prize for a TGRG-sponsored submission at the RGS-IBG conference 2013.TRANSCRIPT
Dynamics of Social Interaction Frequency: Role of Geography and Accessibility
Fariya Sharmeen
Background
• People are space bound.
• Social networks also are embedded in a spatial setting.
• Studies on social networks recognize distance in • Studies on social networks recognize distance in geographical space as a determinant of distance in network links in terms of tie strength between an ego and alters.
• Thus geography matters for social interactions as much as anything else
27-8-2013 2/ department of the Built Environment
Background
Which geographical features?
• Distance to alters
• Accessibility to Transport infrastructures
• Accessibility to Facilities
• Degree of Urbanization
History of interaction
Dynamics of distance
27-8-2013 3/ department of the Built Environment
Background
Why this is important?
1. Acknowledge the role of geographical features in explaining social interaction
2. Comprehensive modelling approach2. Comprehensive modelling approach
3. can provide useful direction to sustainable community design and can strengthen social cohesion policies.
27-8-2013 4/ department of the Built Environment
Theory and Concepts
a
c
c
e
s
s
i
b
h
o
m
o
p
Social Interaction
Accessibility to
facilities and
transport
infrastructures
Ego-Alter tie
characteristics
Degree of
urbanization
Ego’s socio
demographics
/ department of the Built Environment 27-8-2013 5
path dependency life course approach
b
il
it
y
p
h
il
y Geographical
distance with alter History of social
interaction
Changes in
geographical
distance with alter
Research Questions
• To what extent are social interaction/activity frequencies path dependent?
• Does a change in geographical distance with alters influence social interaction/activity frequency?
• What are the effects of urbanization and accessibility to social/recreational facilities on social interaction/activity frequency?
• What are the effects of life cycle-events on social interaction/activity frequency?
27-8-2013 6/ department of the Built Environment
Data
Event based retrospective survey
1.Socio demographics
2.Present social network
3.Dynamics in social network
4.Dynamics in Activity Travel schedule4.Dynamics in Activity Travel schedule
Netherlands
700 respondents
September 2011
8458 ties
27-8-2013 7/ department of the Built Environment
Data
Cetral Bureau of Statistics, Netherlands
Degree of urbanization and
Accessibility to facilities
within neighbourhood – defined by 4 digit postcode
20062006
Open street map
calculate distance to city/town/village center
27-8-2013 8/ department of the Built Environment
Methodology
Ordered Logit model
Extension 1: Random effects model
1 1
1
n
ijt itk itk ij
k
y xβ ε=
= +∑
n n n
y x x xβ β β α ε= + + + +∑ ∑ ∑
Extension 2: Path dependence model
1 1 2 2 3 3
1 1 1
n n n
ijt itk itk ijtk ijtk itk itk i ij
k k k
y x x xβ β β α ε= = =
= + + + +∑ ∑ ∑
ijitij
n
k
ktijktij
n
k
itkitk
n
k
ijtkijtk
n
k
itkitkijt yxxxxy εαββββ ++++++= −
=
−−
===
∑∑∑∑ )1(
1
)(2)1(2
1
33
1
22
1
11
27-8-2013 9/ department of the Built Environment
Descriptives
/ department of the Built Environment 1027-8-2013
Results
Model 1:Basic
Model 2: Random
Model 3: Random effects with land use
Model 4: Random effects with
Model 5:Random effects with
Ordered logit model of face-to-face interaction frequency
Dependent variable face o face interaction frequency
(0=none, 1=less, 2=once a month, 3=2-3 times a month, 4=once a week,
5=2-3 times a week, 6=daily)
Basic model
Random effects model
with land use and accessibility indicators
effects with state dependence
effects with interaction variables
Log likelihood function -14589.34 -14262.48 -14245.97 -12276.56 -12262.65
Restricted log likelihood
-15514.46 -14589.34 -14562.30 -12582.74 -12567.12
Chi square 1850.24*** 653.71*** 632.65*** 612.36*** 608.93***
# of Parameters 34 35 44 49 54
AIC/N 3.458 3.381 3.379 2.915 2.912
∆i=Exp(-∆AIC /2) 6.70 4.85 3.30 1.34 1
27-8-2013 11/ department of the Built Environment
Results
6 groups of variables
I1: Ego’s socio demographicsI2: Ego-alter tie characteristicsI3: Life cycle eventsI4: Density and Aceessibility indicatorsI5: History of social interactions and distanceI6: interaction effects
I1: Ego’s socio demographics β t
AGE21-30 0.352 *** 7
Gender: male -0.007 -0.25
No car 0.319 *** 7.91
Driving license: yes 0.144 *** 4.19
HH with children -0.081 ** -2.17
HH without children -0.151 *** -3.67
HH with friends -0.080 -1.6
Work hr per week 0.001 1.33
Size of Social network -0.103 *** -7.45
# club membership 0.082 *** 3.45
27-8-2013 12/ department of the Built Environment
Results
I2: Ego-alter tie characteristics β t
Same age group -0.119 *** -4.57
Same education level 0.218 *** 5.07
Same gender 0.169 *** 6.24
Age difference: upto 2 category 0.011 0.46
Education level difference: upto 2 category -0.028 -0.73
Relation: neighbour 0.203 *** 4.69Relation: neighbour 0.203 *** 4.69
Relation: friend 0.239 *** 8.2
Relation: family 0.272 *** 7.49
Relationship strength: weak -0.893 *** -19.13
Relationship strength: medium -0.315 *** -12.46
Distance: 0-1 km 2.739 *** 46.67
Distance: 2-5 km 0.284 *** 7.21
Distance: 6-30 km 0.257 *** 6.98
27-8-2013 13/ department of the Built Environment
Results
I3: Life cycle events β t
Event: change in work/study 0.056 0.81
Event: children start school 0.092 1.12
Event: change in residence -0.108 *** -2.71
Event time: 6-12 months -0.161 *** -4.6
Event time: 0-6 months -0.202 *** -6.1
I4: Density and accessibility indicators β t
Distance to city centre (log) -0.011 * 1.68Distance to city centre (log) -0.011 * 1.68
Rural area 0.064 * 1.92
Suburban area -0.057 * -1.86
#Shopping within 1 km 0.014 * 1.67
#Café/Restaurant within 1 km -0.002 *** -6.53
#Attractions within 10 km -0.015 *** -2.41
Distance to highway (log) 0.034 * 1.79
Distance to train station (log) -0.031 * -1.76
Distance to public green (log) 0.001 -0.62
27-8-2013 14/ department of the Built Environment
Results
I5: History of social interaction and distance β t
F2F before: daily 3.342 *** 89.85
F2F before: weekly 2.200 *** 60.84
F2F before: monthly 0.812 *** 17.39
NOWFAR -0.425 ** -2.13
NOWCLOSE -0.059 -1.01
I6: Interaction effects β t
Neighbour*change in residence -0.347 *** -5.07
NOWFAR*change in residence 0.412 *** 2.35NOWFAR*change in residence 0.412 *** 2.35
NOWFAR*change in work/study 0.104 0.56
NOWCLOSE*change in work/study -0.144 *** -3.48
NOWCLOSE*change in residence 0.251 *** 5.27
β t
Mu(01) 1.632 66.96
Mu(02) 2.483 96.24
Mu(03) 3.116 114.0
Mu(04) 3.938 142.4
Mu(05) 5.003 163.2
β t
Sigma 0.482 20.82
Threshold parameter
Std Deviation random effect
27-8-2013 15/ department of the Built Environment
Answers to Research Questions
• Face-to-face social interaction frequencies are strongly path dependent.
• Findings support that change in geographical distance affect social interaction frequency.
• Varied effects of degree of urbanization, accessibility to facilities and transport infrastructures were observed and reported.
• Social interaction/activity frequency dynamics are subject to the type of event.
27-8-2013 16/ department of the Built Environment
Conclusion
• The study aims to model face to face social interaction frequency taking the effects of accessibility, degree of urbanization, and path dependency.
• It shows that inclusion of local geographical and accessibility indicators and long term dynamics accessibility indicators and long term dynamics improves model performance.
• The findings add to the long debated body of literature in sociology about distance decay in social relationships.
• The novel contribution is the inclusion of the effects of changes induced by life-cycle events.
27-8-2013 17/ department of the Built Environment
Next?
• In depth analysis of purpose and location-type choice
• Alter’s home location
• Preferences, limitations of individuals• Preferences, limitations of individuals
• Ethnic and migrated neighbourhoods
27-8-2013 18/ department of the Built Environment