Using network theory to understandbehavioural aspects of energy use

Casper Albers

Introduction

Collaborators

Nitin Bhushan Linda Steg Mark VerschoorNWO #406-13-006

How to reduce (household) electricity use

1. Technological advancesMore efficient refrigerators, tv’s, etc.

2. Changing energy behaviourThis requires understanding energy behaviour

Special characteristics of psychological research

• Distinction between independent and dependent variables often not clear:everything affects everything

• Causal relations are hard. Often variables cause each other (A→ B and B→A)

• Many variables involved. Often many overlapping theories involved.• Technical background of psychologists more limited than of, e.g., biologists.Results should be interpretable for non-statisticians.

Solution: use psychological network models.

Psychological Networks

From latent variable models to network models

(Cramer et al., 2010)

From latent variable models to network models

(Cramer et al., 2010)

Example: General Anxiety Disorder & Major Depression

(Borsboom & Cramer, 2013)

Psychological network models

In essence, psychological network models are:

A set of nodes and edges, depicting random variables and their relations.

Gaussian graphical models are a specific case of psychological network models.

Psychological networks vs. social networks

Networks

• Network of things

• Examples: social networks, trafficnetworks

• Nodes represent entities• Edges are either observed ordirectly derivable from observations

• Main goal: explainpresence/absence of edges,centrality, shortest paths

Psychological networks• Network of variables

• Examples: correlation networks,depression symptom models

• Nodes represent random variables• Edges are estimated statisticalrelations, subject to uncertainty

• Main goal: explain (co)occurrence ofcertain nodes

• Conceptually improvement to latentvariable models

(Table: Sacha Epskamp)

Psychological networks vs. social networks

Networks

• Network of things• Examples: social networks, trafficnetworks

• Nodes represent entities• Edges are either observed ordirectly derivable from observations

• Main goal: explainpresence/absence of edges,centrality, shortest paths

Psychological networks• Network of variables• Examples: correlation networks,depression symptom models

• Nodes represent random variables• Edges are estimated statisticalrelations, subject to uncertainty

• Main goal: explain (co)occurrence ofcertain nodes

• Conceptually improvement to latentvariable models

(Table: Sacha Epskamp)

Psychological networks vs. social networks

Networks

• Network of things• Examples: social networks, trafficnetworks

• Nodes represent entities

• Edges are either observed ordirectly derivable from observations

• Main goal: explainpresence/absence of edges,centrality, shortest paths

Psychological networks• Network of variables• Examples: correlation networks,depression symptom models

• Nodes represent random variables

• Edges are estimated statisticalrelations, subject to uncertainty

• Main goal: explain (co)occurrence ofcertain nodes

• Conceptually improvement to latentvariable models

(Table: Sacha Epskamp)

Psychological networks vs. social networks

Networks

• Network of things• Examples: social networks, trafficnetworks

• Nodes represent entities• Edges are either observed ordirectly derivable from observations

• Main goal: explainpresence/absence of edges,centrality, shortest paths

Psychological networks• Network of variables• Examples: correlation networks,depression symptom models

• Nodes represent random variables• Edges are estimated statisticalrelations, subject to uncertainty

• Main goal: explain (co)occurrence ofcertain nodes

• Conceptually improvement to latentvariable models

(Table: Sacha Epskamp)

Psychological networks vs. social networks

Networks

• Network of things• Examples: social networks, trafficnetworks

• Nodes represent entities• Edges are either observed ordirectly derivable from observations

• Main goal: explainpresence/absence of edges,centrality, shortest paths

Psychological networks• Network of variables• Examples: correlation networks,depression symptom models

• Nodes represent random variables• Edges are estimated statisticalrelations, subject to uncertainty

• Main goal: explain (co)occurrence ofcertain nodes

• Conceptually improvement to latentvariable models

(Table: Sacha Epskamp)

Psychological networks vs. social networks

Networks

• Network of things• Examples: social networks, trafficnetworks

• Nodes represent entities• Edges are either observed ordirectly derivable from observations

• Main goal: explainpresence/absence of edges,centrality, shortest paths

Psychological networks• Network of variables• Examples: correlation networks,depression symptom models

• Nodes represent random variables• Edges are estimated statisticalrelations, subject to uncertainty

• Main goal: explain (co)occurrence ofcertain nodes

• Conceptually improvement to latentvariable models

(Table: Sacha Epskamp)

Psychological networks

X ∼ N(µ,Σ), X : n× p

Example: grades for n = 44 students on p = 5 grades. Correlations:

mechanics vectors algebra analysis statisticsmechanics 1.00 0.32 0.41 0.38 0.27

vectors 0.32 1.00 0.43 0.23 0.22algebra 0.41 0.43 1.00 0.58 0.55analysis 0.38 0.23 0.58 1.00 0.51statistics 0.27 0.22 0.55 0.51 1.00

Psychological networks

X ∼ N(µ,Σ), X : n× pExample: grades for n = 88 students on p = 5 grades.Correlations (from Σ̂) and partial correlations (from Σ̂−1):

mch

vct

alganl

stt

mch

vct

alganl

stt

Psychological networks

X ∼ N(µ,Σ), X : n× pExample: grades for n = 88 students on p = 5 grades.Correlations (from Σ̂) and partial correlations (from Σ̂−1):

mch

vct

alganl

stt

mch

vct

alganl

stt

Psychological networks: fewer edges

I nodes→ I(I− 1)/2 potential edges. Quickly too many.

Graphical lasso. Estimate Σ−1 subject to L1-penalties. Commonly used is theGraphical lasso (Friedman, Hastie, Tibshirani, Biostatistics, 2008) approach withthe extended BIC to select tuning parameter.

mch

vct

alganl

stt

mch

vct

alganl

stt

Which nodes matter most?

mch

vct

alganl

stt

Betweenness Closeness Strength

−0.50 −0.25 0.00 0.25 0.50−1 0 1 −1 0 1

algebra

analysis

mechanics

statistics

vectors

Network Centrality Measures (standardised)

Wij: matrix with weights (∈ [−1, 1]) in the network. SDij: shortest distance fromnode i to node j.

• Node strength/centrality: Si =∑

jWij• Closeness: CLi = (

∑j SDij)−1

• Betweenness: Bi = #(paths through node i)/#(all paths)

Comparing networks

mch

vct

alganl

stt

mch

vct

alganl

stt

Below median score for mechanics Above median score for mechanics

• Structural Hamming Distance.Count number of edges that (dis)appeared/changed sign.Here: 5 out of 10.

• Network Comparison Test (Van Borculo et al.)Statistical permutation test, similar to Mantel’s test (1967).Here: p = .62.

Comparing networks

mch

vct

alganl

stt

mch

vct

alganl

stt

Below median score for mechanics Above median score for mechanics

• Structural Hamming Distance.Count number of edges that (dis)appeared/changed sign.Here: 5 out of 10.

• Network Comparison Test (Van Borculo et al.)Statistical permutation test, similar to Mantel’s test (1967).Here: p = .62.

Psychological networks: Beyond the basics

In this talk, only undirected GGM.

Network models also possible for:

• Directed graphs, when visualising temporal dynamics (e.g. VAR-models;Bringmann et al.) or causal models (DAGs).

• Ordinal data: polychoric rather than Pearson correlations.• Binary data: Ising-models.

Furthermore, all estimates (e.g. edge weights) can be equipped by bootstrap CI’s(Epskamp et al, 2017).

Environmental Applications

Buurkracht: Introduction

• Community energy initiative for promoting sustainable energy behaviour.• Research on energy behaviour usually focused on individual.• Community efforts might be more effective.• Combination of various psychological and sociological theories at play.

• N1 = 334 initiative participants, N2 = 360 right-door-neighbours.• Data on 22 variables, and 65 items, 7-point Likert scales.

Buurkracht: Introduction

• Community energy initiative for promoting sustainable energy behaviour.• Research on energy behaviour usually focused on individual.• Community efforts might be more effective.• Combination of various psychological and sociological theories at play.

• N1 = 334 initiative participants, N2 = 360 right-door-neighbours.• Data on 22 variables, and 65 items, 7-point Likert scales.

Buurkracht: Introduction

• Community energy initiative for promoting sustainable energy behaviour.• Research on energy behaviour usually focused on individual.• Community efforts might be more effective.• Combination of various psychological and sociological theories at play.

• N1 = 334 initiative participants, N2 = 360 right-door-neighbours.• Data on 22 variables, and 65 items, 7-point Likert scales.

Buurkracht: Variables in the study

Personal factors Values. Environmental self-identity. Personal importance ofsustainable energy behaviour. Outcome efficacy.

Social context Need to belong Need to be unique. Neighbourhood identification.Neighbourhood homogeneity. Neighbourhood interaction. Neighbourhoodenvironmental identity. Neighbourhood importance of sustainable energy behaviour.

Opinions on energy companies and the government Group-based anger.

Sustainable energy intentions and behaviour Household sustainable energyintentions. Collective sustainable energy intentions. Collective social intentions.Self-reported sustainable energy behaviour.

Initiative membership

Buurkracht: Findings

●●

●●

●● ●●

●●●●

●●●●●●

●●●●

●●

●●

●●●●●●●●

●●●●

●●

●● ●●●●

●●●● ●●

●●

●●●●

●●●●●●

●●●●

●●●●

●●

●●●● ●●

●●●●

●● ●●

●●

●●●●

●●●● ●●

●●●●●●

●●

●●●●

●●

●●

●●●●●●

●●

●●●●

●●

1

2

3

4

5

6

78

9

10

11

12

13

1415

16

17

18

19

20

2122

23

24

25 26

27

2829

3031

32

33

34

35

36

37

3839

40

41

42

43 44

45

46

47

48

49 50

515253

54

55

56

57

58

59

60

61

62

63

64

65

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

Altruistic valuesBiospheric ValuesCollective energy intentionsCollective social intentionsEgoistic ValuesEnvironmental Self_identityGroup−based angerHedonic ValuesHomogeneity in neighbourhoodIndividual energy intentionsInitiative outcome efficacyInteraction in neighbourhoodInteraction with neighbourhoodMembershipNeed to be UniqueNeed to belong Neighbourhood energy normsNeighbourhood environmental IdentityNeighbourhood identificationPast energy behaviorPersonal Energy NormsPersonal outcome efficacy

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

Altruistic valuesBiospheric ValuesCollective energy intentionsCollective social intentionsEgoistic ValuesEnvironmental Self_identityGroup−based angerHedonic ValuesHomogeneity in neighbourhoodIndividual energy intentionsInitiative outcome efficacyInteraction in neighbourhoodInteraction with neighbourhoodMembershipNeed to be UniqueNeed to belong Neighbourhood energy normsNeighbourhood environmental IdentityNeighbourhood identificationPast energy behaviorPersonal Energy NormsPersonal outcome efficacy

Buurkracht: Findings

12

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

Factors related to the social context5: Neighbourhood environmental Identity6: Neighbourhood identification7: Interaction with neighbourhood8: Neighbourhood energy norms9: Homogeneity in neighbourhood10: Interaction in neighbourhood13: Need to belong15: Need to be Unique20: Initiative outcome efficacy

Membership22: Membership

Opinions on energy companies21: Group based anger

Personal factors 11: Personal outcome efficacy12: Environmental Self identity14: Personal Energy Norms16: Altruistic values17: Biospheric Values18: Egoistic Values19: Hedonic Values

Sustainable energy intentions1: Collective energy intentions2: Collective social intentions3: Individual energy intentions4: Past energy behavior

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

Factors related to the social context5: Neighbourhood environmental Identity6: Neighbourhood identification7: Interaction with neighbourhood8: Neighbourhood energy norms9: Homogeneity in neighbourhood10: Interaction in neighbourhood13: Need to belong15: Need to be Unique20: Initiative outcome efficacy

Membership22: Membership

Opinions on energy companies21: Group based anger

Personal factors 11: Personal outcome efficacy12: Environmental Self identity14: Personal Energy Norms16: Altruistic values17: Biospheric Values18: Egoistic Values19: Hedonic Values

Sustainable energy intentions1: Collective energy intentions2: Collective social intentions3: Individual energy intentions4: Past energy behavior

Buurkracht: Findings

1

2

3

4

5

6

7

8

910

11

12

13

14

15

16

17

18

19

20

21

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

Factors related to the social context5: Neighbourhood environmental Identity6: Neighbourhood identification7: Interaction with neighbourhood8: Neighbourhood energy norms9: Homogeneity in neighbourhood10: Interaction in neighbourhood13: Need to belong15: Need to be Unique20: Initiative outcome efficacy

Opinions on energy companies21: Group based anger

Personal factors 11: Personal outcome efficacy12: Environmental Self identity14: Personal Energy Norms16: Altruistic values17: Biospheric Values18: Egoistic Values19: Hedonic Values

Sustainable energy intentions1: Collective energy intentions2: Collective social intentions3: Individual energy intentions4: Past energy behavior

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

Factors related to the social context5: Neighbourhood environmental Identity6: Neighbourhood identification7: Interaction with neighbourhood8: Neighbourhood energy norms9: Homogeneity in neighbourhood10: Interaction in neighbourhood13: Need to belong15: Need to be Unique20: Initiative outcome efficacy

Opinions on energy companies21: Group based anger

Personal factors 11: Personal outcome efficacy12: Environmental Self identity14: Personal Energy Norms16: Altruistic values17: Biospheric Values18: Egoistic Values19: Hedonic Values

Sustainable energy intentions1: Collective energy intentions2: Collective social intentions3: Individual energy intentions4: Past energy behavior

1

2

3

4

5

6

7

8

910

11

12

13

14

15

16

17

18

19

20

21

Structural Hamming Distance: 12

ESS: Introduction

• Open Data, www.europeansocialsurvey.org• “Public Attitudes to Climate Change, Energy Security, and Energy Preferences”• N = 38, 437 participants from 18 countries (880 to 2,766 per country).

Research questions:

1. What, and how strong, are the relations between the variables?2. Are these relations the same across countries?

ESS: Introduction

• Open Data, www.europeansocialsurvey.org• “Public Attitudes to Climate Change, Energy Security, and Energy Preferences”• N = 38, 437 participants from 18 countries (880 to 2,766 per country).

Research questions:

1. What, and how strong, are the relations between the variables?2. Are these relations the same across countries?

ESS: Introduction

• Open Data, www.europeansocialsurvey.org• “Public Attitudes to Climate Change, Energy Security, and Energy Preferences”• N = 38, 437 participants from 18 countries (880 to 2,766 per country).

Research questions:

1. What, and how strong, are the relations between the variables?2. Are these relations the same across countries?

ESS: the Module

The module “Public Attitudes to Climate Change, Energy Security, and EnergyPreferences” includes 32 items in the areas:

1. Beliefs on climate change2. Concerns about climate change and energy security3. Personal norms, efficacy and trust4. Energy preferences.

ESS: findings

EDM1

EDM2

EB1

ESP1

ESP2

ESP3ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

Policy supportPS1: Increasing taxes on fossil fuelsPS2: Subsidise renewable energyPS3: Ban least energy efficient appliances

Climate change beliefsCCB1: Belief in climate changeCCB2: Climate change causeCCB3: Impact of climate change

Climate change salienceCCS: Thought about climate change before today

Climate concernCC: Climate change worry

Efficacy beliefsEB1: Could use less energyEB2: Peoples impact on climate changeEB3: Likeliness of people limiting energy useEB4: Likeliness of governments limiting energy useEB5: My impact on climate change

Energy demand measuresEDM1: Buy energy efficient appliancesEDM2: Reduce your energy use

Energy security concernESC1: Power cutsESC2: ExpensiveESC3: Energy importsESC4: Fossil fuelsESC5: Natural disastersESC6: Insufficient powerESC7: Technical failuresESC8: Terrorist attacks

Energy supply source preferenceESP1: CoalESP2: Natural gasESP3: Hydroelectric powerESP4: Nuclear powerESP5: Sun or solar powerESP6: Wind powerESP7: Biomass energy

Personal normsPN: Personal responsibility to reduce climate change

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

Policy supportPS1: Increasing taxes on fossil fuelsPS2: Subsidise renewable energyPS3: Ban least energy efficient appliances

Climate change beliefsCCB1: Belief in climate changeCCB2: Climate change causeCCB3: Impact of climate change

Climate change salienceCCS: Thought about climate change before today

Climate concernCC: Climate change worry

Efficacy beliefsEB1: Could use less energyEB2: Peoples impact on climate changeEB3: Likeliness of people limiting energy useEB4: Likeliness of governments limiting energy useEB5: My impact on climate change

Energy demand measuresEDM1: Buy energy efficient appliancesEDM2: Reduce your energy use

Energy security concernESC1: Power cutsESC2: ExpensiveESC3: Energy importsESC4: Fossil fuelsESC5: Natural disastersESC6: Insufficient powerESC7: Technical failuresESC8: Terrorist attacks

Energy supply source preferenceESP1: CoalESP2: Natural gasESP3: Hydroelectric powerESP4: Nuclear powerESP5: Sun or solar powerESP6: Wind powerESP7: Biomass energy

Personal normsPN: Personal responsibility to reduce climate change

ESS: findings

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

EDM1

EDM2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

No major between-network differences. K-means clustering yields 1 cluster.

ESS: findings

−2

0

2

EDM

1ED

M2

EB1

ESP1

ESP2

ESP3

ESP4

ESP5

ESP6

ESP7

ESC

1ES

C2

ESC

3ES

C4

ESC

5ES

C6

ESC

7ES

C8

CC

B1C

CS

CC

B2 PN CC

CC

B3 EB2

EB3

EB4

EB5

PS1

PS2

PS3

Cen

tral

ityCountry

Austra

Belgium

Czech Republic

Estonia

Finland

France

Germany

Iceland

Ireland

Israel

Netherlands

Norway

Poland

Russia

Slovenia

Sweden

Switzerland

United Kingdom

ESS: zooming in

EDM1

EDM2

EB1

ESP1

ESP2

ESP3ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

Policy supportPS1: Increasing taxes on fossil fuelsPS2: Subsidise renewable energyPS3: Ban least energy efficient appliances

Climate change beliefsCCB1: Belief in climate changeCCB2: Climate change causeCCB3: Impact of climate change

Climate change salienceCCS: Thought about climate change before today

Climate concernCC: Climate change worry

Efficacy beliefsEB1: Could use less energyEB2: Peoples impact on climate changeEB3: Likeliness of people limiting energy useEB4: Likeliness of governments limiting energy useEB5: My impact on climate change

Energy demand measuresEDM1: Buy energy efficient appliancesEDM2: Reduce your energy use

Energy security concernESC1: Power cutsESC2: ExpensiveESC3: Energy importsESC4: Fossil fuelsESC5: Natural disastersESC6: Insufficient powerESC7: Technical failuresESC8: Terrorist attacks

Energy supply source preferenceESP1: CoalESP2: Natural gasESP3: Hydroelectric powerESP4: Nuclear powerESP5: Sun or solar powerESP6: Wind powerESP7: Biomass energy

Personal normsPN: Personal responsibility to reduce climate change

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

Policy supportPS1: Increasing taxes on fossil fuelsPS2: Subsidise renewable energyPS3: Ban least energy efficient appliances

Climate change beliefsCCB1: Belief in climate changeCCB2: Climate change causeCCB3: Impact of climate change

Climate change salienceCCS: Thought about climate change before today

Climate concernCC: Climate change worry

Efficacy beliefsEB1: Could use less energyEB2: Peoples impact on climate changeEB3: Likeliness of people limiting energy useEB4: Likeliness of governments limiting energy useEB5: My impact on climate change

Energy demand measuresEDM1: Buy energy efficient appliancesEDM2: Reduce your energy use

Energy security concernESC1: Power cutsESC2: ExpensiveESC3: Energy importsESC4: Fossil fuelsESC5: Natural disastersESC6: Insufficient powerESC7: Technical failuresESC8: Terrorist attacks

Energy supply source preferenceESP1: CoalESP2: Natural gasESP3: Hydroelectric powerESP4: Nuclear powerESP5: Sun or solar powerESP6: Wind powerESP7: Biomass energy

Personal normsPN: Personal responsibility to reduce climate change

ESS: zooming in

Detailed look at five variables

Werfel, 2017; Noblet & McCoy, 2017:

Engaging inEnergy Behaviour

Support forEnergy Policies

Steinhorst & Matthies, 2016; Thøgersen & Noblet, 2012:

Engaging inEnergy Behaviour

Support forEnergy Policies

Desire for consistency:Stronger relation between related concepts

Detailed look at five variables

B1B2

P1

P2

P3

●

●

●

●

●

BehaviourB1: Buying an energy−efficient applianceB2: Energy−saving behaviour

Policy supportP1: Taxes on fossil fuelsP2: Subsidies for renewable energyP3: Ban on inefficient appliances

●

●

●

●

●

BehaviourB1: Buying an energy−efficient applianceB2: Energy−saving behaviour

Policy supportP1: Taxes on fossil fuelsP2: Subsidies for renewable energyP3: Ban on inefficient appliances

B1B2

P1

P2

P3

●

●

●

●

●

BehaviourB1: Buying an energy−efficient applianceB2: Energy−saving behaviour

Policy supportP1: Taxes on fossil fuelsP2: Subsidies for renewable energyP3: Ban on inefficient appliances

●

●

●

●

●

BehaviourB1: Buying an energy−efficient applianceB2: Energy−saving behaviour

Policy supportP1: Taxes on fossil fuelsP2: Subsidies for renewable energyP3: Ban on inefficient appliances

Country comparison

Country comparison

ESS: zooming in

EDM1

EDM2

EB1

ESP1

ESP2

ESP3ESP4

ESP5

ESP6

ESP7

ESC1

ESC2

ESC3

ESC4

ESC5

ESC6

ESC7

ESC8

CCB1

CCS

CCB2

PN

CC

CCB3

EB2

EB3

EB4

EB5

PS1

PS2

PS3

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

Policy supportPS1: Increasing taxes on fossil fuelsPS2: Subsidise renewable energyPS3: Ban least energy efficient appliances

Climate change beliefsCCB1: Belief in climate changeCCB2: Climate change causeCCB3: Impact of climate change

Climate change salienceCCS: Thought about climate change before today

Climate concernCC: Climate change worry

Efficacy beliefsEB1: Could use less energyEB2: Peoples impact on climate changeEB3: Likeliness of people limiting energy useEB4: Likeliness of governments limiting energy useEB5: My impact on climate change

Energy demand measuresEDM1: Buy energy efficient appliancesEDM2: Reduce your energy use

Energy security concernESC1: Power cutsESC2: ExpensiveESC3: Energy importsESC4: Fossil fuelsESC5: Natural disastersESC6: Insufficient powerESC7: Technical failuresESC8: Terrorist attacks

Energy supply source preferenceESP1: CoalESP2: Natural gasESP3: Hydroelectric powerESP4: Nuclear powerESP5: Sun or solar powerESP6: Wind powerESP7: Biomass energy

Personal normsPN: Personal responsibility to reduce climate change

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

Policy supportPS1: Increasing taxes on fossil fuelsPS2: Subsidise renewable energyPS3: Ban least energy efficient appliances

Climate change beliefsCCB1: Belief in climate changeCCB2: Climate change causeCCB3: Impact of climate change

Climate change salienceCCS: Thought about climate change before today

Climate concernCC: Climate change worry

Efficacy beliefsEB1: Could use less energyEB2: Peoples impact on climate changeEB3: Likeliness of people limiting energy useEB4: Likeliness of governments limiting energy useEB5: My impact on climate change

Energy demand measuresEDM1: Buy energy efficient appliancesEDM2: Reduce your energy use

Energy security concernESC1: Power cutsESC2: ExpensiveESC3: Energy importsESC4: Fossil fuelsESC5: Natural disastersESC6: Insufficient powerESC7: Technical failuresESC8: Terrorist attacks

Energy supply source preferenceESP1: CoalESP2: Natural gasESP3: Hydroelectric powerESP4: Nuclear powerESP5: Sun or solar powerESP6: Wind powerESP7: Biomass energy

Personal normsPN: Personal responsibility to reduce climate change

ESS: zooming in

ESS: zooming in

ESS: zooming in

Strong relation between the various correlations per country.

Conclusions

Summary

Psychological networks are useful for:

• Exploratory analyses of (relatively) high-dimensional data• Validation of scales of various items• Testing hypotheses on the absence/presence, direction or strength ofrelations

• Testing hypotheses on the co-occurrence of relations

With these models, we explored and confirmed theories on energy-relatedbehaviour and found differences and similarities between countries.

References & credits

Papers in preparation on this topic:

• N. Bhushan, F. Mohnert, C. J. Albers, L. Jans, L. Steg. The value of Gaussian graphicalmodels to explore relationships between environmental psychological constructs.

• M. Verschoor, C. J. Albers, L. Steg. A network model of the environmental module inthe European Social Survey 2016

• T. Bouman, M. Verschoor, L. Steg, G. Böhm, S. D. Fisher, W. Poortinga, L. Whitmarsh, C.J. Albers. General personal factors predicting energy-saving behaviours and climatepolicy support

• E. J. Sharpe, M. Verschoor, L. Steg, G. Perlaviciute, C. J. Albers. Similarity encouragesconsistency across energy-saving behaviour and energy policy support in Europeand beyond

Further reading – Environmental Psychology

• Abrahamse, W. (2007). Energy conservation through behavioural change. Universityof Groningen

• Van der Werff, E., Steg, L., & Keizer, K. (2013). The value of environmental self-identity:The relationship between biospheric values, environmental self-identity andenvironmental preferences, intentions and behaviour. Journal of EnvironmentalPsychology.

• Steg, L., Bolderdijk, J.W., Keizer, K., & Perlaviciute, G.(2014). An integrated frameworkfor encouraging pro-environmental behaviour: the role of values, situational factorsand goals. Journal of Environmental Psychology ofvalues,situationalfactorsan

• Bamberg, S., Rees, J., & Seebauer, S. (2015). Collective climate action: Determinants ofparticipation intention in community-based pro-environmental initiatives. Journal ofEnvironmental Psychology.

• Steg, L., Perlaviciute, G., & van der Werff, E. (2015). Understanding the humandimensions of a sustainable energy transition. Frontiers in Psychology.

Further reading – Psychological Networks

• Dempster, A.P. (1972). Covariance selection. Biometrics.

• Lauritzen, S.L. (1996). Graphical Models. Oxford University Press.

• Snijders, T.A.B. (2002). The statistical evaluation of social network dynamics.Sociological methodology.

• Opsahl, T., Skvoretz, J. (2010). Node centrality in weighted networks. Social networks.

• Cramer, A.O.J., Waldorp, L., van der Maas, H., & Borsboom, D. (2010). Comorbidity: ANetwork Perspective. Behavioral and Brain Sciences.

• Borsboom, D., & Cramer, A. O. (2013). Network analysis: an integrative approach tothe structure of psychopathology. Annual review of clinical psychology.

• Bringmann, L.F. (2016). Dynamical networks in psychology: More than a prettypicture? University of Leuven.

• Epskamp, S., Borsboom, D., & Fried, E. I. (2017). Estimating psychological networksand their accuracy: a tutorial paper. Behavioural Research.