Transcript Volunteering: Applications and Methodology

Volunteering
Applications and methodology
Applied Social Psychology
VU University Amsterdam
January 6, 2015
René Bekkers
Philanthropic Studies
VU University Amsterdam
[email protected]
Today’s story
• How can we use insights from social
psychology theories and research to get
people to give and volunteer more?
• Today’s story is not about tricks ‘nudging’ and ‘priming’.
• Instead, it is about theory, data,
methodology, and the industry called
‘social science’.
The Empirical Cycle
Context
Policy
Research
RQ
Theory
Context
1. “Participation society”: volunteers
become more important in the provision
of public goods.
2. “Trust crisis”: trust in banks, organized
religion, politics, and ultimately
generalized social trust are at risk.
3. In an ageing society, keeping people
active as long as possible may reduce
health costs.
Research Questions
1. Does volunteering make people more
trusting?
2. Does volunteering keep people healthy?
…leaving the question ‘How can we get
people to volunteer more?’ for a future
occasion.
Now here’s a social dilemma
Should I volunteer?
This is a social dilemma because
• There is an opportunity cost for me:
volunteering takes time that I could
spend working for pay.
• The benefits of volunteering accrue to a
collective.
• Personally I would be better off not
volunteering, but the group would not.
What type of social dilemma?
Trust game
Dictator game
Ultimatum game
Public goods game
Volunteer’s dilemma
Samaritan’s dilemma
Or something else?
Perhaps it is not a social dilemma at all.
Three facts
1. Volunteers live in better health.
2. Volunteers are less likely to get
depressed.
3. Volunteers ultimately live longer.
More facts
• Volunteers have larger and more diverse
social networks.
• Volunteers have more altruistic values, a
stronger sense of social responsibility,
and a stronger belief in the
trustworthiness of others.
So…
• If these are the results of voluntary
participation, there is no social dilemma
at all!
But…
• Does volunteering indeed have these
nice benefits for volunteers?
• Are networks, trust, health, subjective
well being the result of volunteering?
Whence the difference?
• Do the attitudes, values, networks, and
health change because people volunteer?
This is the common interpretation.
CAUSATION:
Networks
Voluntary action 
Values
Health
Whence the difference?
• Or do attitudes, values, networks, and
health lead people to volunteer?
This is the SELECTION Model:
Networks
Values
Health

Voluntary Action
Or perhaps..
• The relationship between volunteering,
attitudes, values, networks, and health is
confounded by OMITTED VARIABLES:
Voluntary Action
Other factors
x
Networks
e.g., education,
Values
religion
Health
Selection
• Selection: some people are more likely to
be drawn into voluntary action.
• These are the more happy, trusting,
healthy, people with larger networks.
• They are more likely to start volunteering,
and they are less likely to quit
volunteering.
A theory on selection for values
• ‘Interactionism’ in personality and social
psychology
• Individual differences in prosocial values shape
the attractiveness of situations that involve
contributions to the well being of others
• Individuals with larger networks are more likely
to be asked to start and continue volunteering
• Individuals in better health are more able to
continue volunteering
Theories on causation for values
• Group socialization theory
– People adopt the values of the groups that
they are in (family, church, work)
– ‘Social capital’ (attitude) formation
• Self-perception theory
– People adapt their values and self-identity to
their behavior
– Role identity theory: volunteer role identity is
reinforced through volunteering
Other theories on causation
• Meeting opportunities
– People gain access to new networks in
organizations through participation
– ‘Social capital’ (network) formation
• Networks protect health
– Information, social control, access to social
support, stress buffering effect
• Personality strength
– Mastery, self-efficacy, purpose in life
Data and methods of previous
studies
• Many studies use cross-sectional data,
including a limited set of controls.
• Selection and omitted variables are a huge
problem here.
• Studies using longitudinal panel data
have almost all used inadequate
regression models.
• Selection and omitted variables are still a
problem here.
Adequate Testing, Please!
• The conventional ‘change model’ includes
a lagged dependent variable Yt-1
• Halaby (2004, Annual Review of
Sociology) shows this is not enough.
• The Yt-1 does not rule out selection effects.
• Use fixed effects regression models,
eliminating variance between individuals
• Previous studies have rarely used this.
Giving in the Netherlands Panel
Study (GINPS)
• A sample of individuals fills out surveys
on the web every other year since 2002
• You are welcome to use these data.
See the user manual at
http://geveninnederland.nl/file/208/ginps_codebook.pdf
Development of generalized social
trust (‘most people can be trusted’)
3,5
3,4
3,3
never
quit
joined
sustained
3,2
3,1
3
2,9
2002
2004
2006
Source: GINPS
Bekkers, R. (2012). ‘Trust and Volunteering: Selection or Causation? Evidence From a 4 Year Panel Study’. Political
Behaviour, 34: 225-247, DOI 10.1007/s11109-011-9165-x (open access)
Volunteering effects on values
Trust Altruistic
Social
values responsibility
Constant
Random effects
Fixed effects
Hausman test
3.125
3.460
3.727
***.096
***.203
***.147
-.034
**.058
.035
***22.72 ***96.62
***30.30
Estimates on effects of volunteering (m=40%) from random and fixed effects regression models. Source: GINPS, 20022006 (n=4,754; 2,783). Bekkers, R. (2007). ‘Values and Volunteering. A Longitudinal Study of Reciprocal Influences in the
Giving in the Netherlands Panel Study’. Paper presented at the 36th Arnova conference, Atlanta, 2007.
Similar results in other countries
Random effects
Fixed effects
First difference
Switzerland
(SHP)
UK
(BHPS)
Australia
(HILDA)
***.957
***.182
***.485
-.053
***.033
.034
.025
***.049
.039
Estimates on effects of volunteering in Switzerland (m=32%), the UK (21%) and Australia (18%) from between and
fixed effects logistic regression models of trust. Van Ingen, E. & Bekkers, R. (2013). Trust Through Civic Engagement?
Evidence from Five National Panel Studies. Political Psychology. DOI: 10.1111/pops.12105
Effects of trust
A one SD increase in trust is associated with a 2.3% increase in the
likelihood of starting to volunteer, 3.5% decrease in quitting, and 8.3%
increase in the likelihood of being asked to volunteer
Bekkers, R. (2012). ‘Trust and Volunteering: Selection or Causation? Evidence From a 4 Year Panel Study’. Political
Behaviour, 34: 225-247, DOI 10.1007/s11109-011-9165-x (open access)
Selection and omitted variables
+ Start t0-2
Altruistic
values t0
Trust t0
- Quit t0-2
+
+
+ Confidence t1 + Confidence t2
Based on Bekkers, R. & Bowman, W. (2009). The Relationship Between Confidence in Charitable Organizations
and Volunteering Revisited. Nonprofit & Voluntary Sector Quarterly, 38 (5): 884-897.
Laurence (2015)
• How does job displacement affect trust?
• The NCDS was a sample of all individuals
born in a single week in March, 1958.
• What happened to trust in the UK
between 1991 and 2008?
• What happened to the trust of those
Britons whose jobs were displaced
between 1991 (age 33) and 2008 (age 50)?
Changes from age 33 to 50
N=
0
500
1000
1500
2000
2500
3000
3500
to distrust (at 50)
distrust (at 33)
to 'depends' (at 50)
to trust (at 50)
FROM
to distrust (at 50)
trust (at 33)
to 'depends' (at 50)
to trust (at 50)
Changes from age 33 to 50
N=
0
500
1000
1500
2000
2500
3000
3500
to distrust (at 50)
distrust (at 33)
to 'depends' (at 50)
to trust (at 50)
FROM
LOSS OF TRUST
to distrust (at 50)
trust (at 33)
to 'depends' (at 50)
to trust (at 50)
Questions for Laurence (2015)
• The article uses longitudinal data, right?
• How are selection effects taken into account?
• How can you check for selection effects by
inspecting changes in ‘X’ as a result of changes
in ‘Y’ in this case?
• How is the lagged dependent model different
from a fixed effects model?
Volunteering Effects on Health
Depression
(CESD)
Subjective
health
8.809
3.540
***-1.569
***.163
Fixed effects
***-.814
***.097
Hausman test
***36.96
***19.00
Constant
Random effects
Estimates on effects of volunteering on depression and subjective health from random and fixed effects models. Source:
LASA, 1992-2002 (n=7,864; 2,362). Bekkers, R., Van Tilburg, T.G., Aartsen, M., Brown, S. & Wilson, J. (2007).
‘Volunteering and Health: A Prospective Study of Mediating Mechanisms’. Unpublished manuscript.
So…
• Use longitudinal data and then still do not claim
too much about effects of voluntary action.
• Selection effects are a big part of the reason why
voluntary action is related to networks, values
and health.
• Always check for selection effects by inspecting
changes in ‘X’ as a result of changes in ‘Y’.
• Estimate fixed effects and first difference
models.
Questions on Konrath et al.
• The article uses longitudinal data, right?
• How are selection effects taken into account?
• How can you check for selection effects by
inspecting changes in ‘X’ as a result of changes
in ‘Y’ in this case?
• How to estimate a fixed effects model of
mortality?
• How to estimate a first difference model?
Contact
• ‘Geven in Nederland’, Philanthropic
Studies, Faculteit Sociale Wetenschappen,
Vrije Universiteit Amsterdam:
www.geveninnederland.nl
• René Bekkers, [email protected]
• Blog: renebekkers.wordpress.com
• Twitter: @renebekkers
• Please do get in touch if you want to write
your thesis on charitable giving,
volunteering, blood donation, etc.!