Combining Choice Experiments with Psychometric Scales to assess the social acceptability of wind energy projects: a Latent Class approach

Transcription

1 Combining Choice Experiments with Psychometric Scales to assess the social acceptability of wind energy projects: a Latent Class approach Elisabetta Strazzera a, Marina Mura a, Davide Contu a a Department of Social Sciences and Institutions, University of Cagliari, Viale S.Ignazio 78, I Cagliari, Italy Forthcoming in Energy Policy

2 OBJECTIVE Combine Choice Experiments and Psychometric Scales in order to: Identify salient factors explaining attitudes toward a proposed wind energy project.

3 THE NEXT 15 MINUTES 1. The literature 2. The case study 3. Methods 4. The qualitative study 5. The survey 6. Results 7. Conclusion

4 1. The Literature Literature The case study Methods The qualitative study The survey Results Conclusion

5 Choice modeling analyses of public acceptance of wind farm projects have typically focused on technological and environmental impacts dimension and density of turbines their location (e.g. on/off shore, on mountains, hills, or flat land, etc.) impacts on wildlife economic impacts (employment, financial costs/benefits)

6 However, there are other factors that need to be taken into account: The effect of demographic and socio-economic characteristics Socio-psychological factors Contextual factors

7 The effect of demographic and socio-economic characteristics the effect of gender, age, education, income, has been examined in several studies (for example, Borchers et al., 2007; Ek, 2002; Ek and Söderholm, 2008; Krueger, 2007; Longo et al., 2008) with mixed results.

8 The effect of Socio-psychological factors (I) : awareness: more informed individuals may be more favorable toward green energy technologies, although there is limited empirical evidence in this sense; environmental concern, which may play a role on both sides of the conflict. (Warren et al., 2005).

9 The effect of Socio-psychological factors (II) : place attachment: people who feel an emotional attachment and identification with some place may oppose the construction of green energy plants in that location, if the project is perceived as being a threat to the integrity of the local environment (Cass and Walker, 2009).

10 The effect of contextual factors : Procedural justice and levels of trust (Gross, 2007; Zoellner et al., 2008 Zoellner et al. 2005, Upham and Shackley 2006); information should be clear and detailed (Kaldellis, 2005; Dimitropoulos and Kontoleon, 2009; Jones, 2009). Fairness in the distribution of benefits: Jobert et al. (2007) and Warren and McFadyen (2010) find that community ownership leads to higher social acceptance of wind power installations; this result is confirmed by Maruyama et al. (2007), and Walker et al. (2010).

11 2. The case study Literature The case study Methods The qualitative study The survey Results Conclusion

12 The case study The research has been conducted in two bordering counties located in the South-West of Sardinia (Italy): Sulcis-Iglesiente (SI) and Medio Campidano (MC). Both areas are characterized by critical socio-economic conditions, due to deindustrialization processes and loss of economic value of agricultural products. Both counties have been important mining areas since the Roman era; the mining activity being replaced by energy intensive industry in the past 40 years. Global competition, and high costs of energy in Italy, are now leading the holding companies to disinvest and relocate in more favorable economic environments. The largest company operating in the territory has envisaged a possible, although partial, solution in the development of a wind farm which would provide low-cost energy for its production.

13

14 3. Methods Literature The case study Methods The qualitative study The survey Results Conclusion

15 3. Methods Choice experiments+ latent class analysis CE draws from the Lancaster s Theory of value and the Random utility theory, which provide the theoretical framework. Often the model used to estimate the parameters is the MNL. However this models assumes homogeneity: the coefficients are the same for each individual. But taking into account heterogeneity seems to be crucial in order to shape respondents attitudes properly. One way is that of estimating a latent class model, according to which our respondents are divided into s segments according to their preferences, and it is possible to condition the probability of belonging to a given segment on socio-economic variables and psychometric variables. Also the RPL approach could have been used, but it would have not helped us understanding the determinants of heterogeneity.

16 3. Methods Principal component analysis It is a technique to reduce data dimension while finding meaningful patterns in the data. We apply this technique to the data obtained from the Likert-scale questions in order to obtain our psychometric variables.

17 Latent Class model Identity scale and other issues scales Choice Exp.

18 4. The qualitative study Literature The case study Methods The qualitative study The survey Results Conclusion

19 The qualitative study The focus groups participants (about 10 people in each group) were conducted by two moderators to discuss issues in five main areas: Spatial factors (visual characteristics: dimension of turbines, number and density; choice of sites: close or far away from urban centers, offshore or on-shore; if on-shore, close to the coast or in the interior; on hills or flat land; close to valued environmental/archeological sites or to less valuable sites) Environmental concerns (climate change and pollution: renewable energy technologies/ other energy sources; conservation/ destruction of landscape)

20 The qualitative study The focus groups participants (about 10 people in each group) were conducted by two moderators to discuss issues in five main areas: Place attachment (economic development: wind farm helps economy/ hinder tourist development; wind farm spoils emotional and identity feelings for the location) Fairness (fairness in the distribution of benefits: private/public property, economic advantages for the community/private individuals) Procedural justice (information; trust in political decision makers; participation in planning decisions)

21 5. The survey Literature The case study Methods The qualitative study The survey Results Conclusion

22 The survey Based on the outcome of the qualitative stage of the research, a choice experiment was designed in order to assess the trade-offs between relevant attributes of the choice involving a wind farm project. We decided to implement a design where the development project would take place in any case (no status quo option was included as a choice alternative), so that the choice was between different levels of costs (visual impacts) and benefits generated by the project.

23 The survey Building the CE: Table 1. Attributes and levels of choice experiments. Attributes Visual impact in the SI county Levels Installation close to the coast and well visible; far away from the coast and not well visible; far away inland not visible from the seaside Visual impact in the MC county Installation close to the coast and well visible; far away from the coast and not well visible; far away inland not visible from the seaside Visual impact on a generic site of archeological interest Installation close to the site; installation far away from the site Property of the plant Private; Public Regional; Public Local Public benefits: additional services No additional public services; training and formation for young residents; training and formation for young residents, plus microcredit to small enterprises Private benefits: reduction in energy bill No decrease, 10%, 30%, 50% reduction

24 Attribute: visual impact in the SI county Not visible Well visible Not well visible

25 Attribute: visual impact in the MC county Well visible Not visible Not well visible

26 Attribute: impact on a archeological site Close to the site Far away from the site

27 The Choice Experiment:

28 The Likert-scale questions: Table A2. Identity scales. Complet ely disagree Quite disagree Disagree Neither agree nor disagree Agree Quite agree Complet ely agree STATEMENTS: I like to spend some time in this place I feel attachment to this place I wouldn't like to go away from this place In this place I feel myself at home When I am away, I miss this place This is my favorite place This place is part of my identity I feel that I belong to this place This place is really different from other places I like this place This is one of my favorite places I have a lot in common with people coming often to this place I identify myself with those who come often to this place

29 The Likert-scale questions: Table A3. Statements concerning various issues. Completely Completely disagree Disagree Neither agree nor disagree Agree agree STATEMENTS: I think wind energy is useful I am against wind energy I wish more investments in wind energy Wind turbines do not give benefits to community Wind turbines produce little energy I like wind turbines I would like wind turbines far away from home I'd like more wind farms I think my economic situation may improve with wind farms Territory where I live is in economic difficulty Some places of the territory where I live are degraded The economic development of this area depends on energy availability The economic development of this area depends on wind farms projects I wish more investments on renewables in Sardinia I wish investments on nuclear energy in Sardinia I'm worried of criminal organization interests in wind farms in Sardinia I search the market for local products I search the market for good prices I always check the product origin I like to consume out of season fruits and vegetables

30 6. The Results Literature The case study Methods The qualitative study The survey Results Conclusion

31 MNL model (without using data on observed heterogeneity) Table 3. Multinomial logit model (MNL). Variable Coeff. MRS WTA ( ) a Beach SI 0.340*** (0.027) Beach MC 0.476*** (0.034) Archsite 0.339*** (0.046) Property 0.09*** (0.03) Services 0.162*** (0.028) *** (0.021) *** (0.021) *** (0.304) *** (0.016) *** (0.016) Bill Reduction 1.677*** (0.145) Choice observations: 2592 Individuals: 432 Log Likelihood LL(0) Pseudo R a WTA computed as - MRS*average electric bill. *** 1% significance, standard errors in parenthesis.

32 PCA (IDENTITY SCALES) A Principal Component Analysis (PCA) was applied to the identity scales, and based on standard statistical criteria (amount of explained variance, value of eigenvalues) only one component has been extracted for each scale, which can be easily interpreted as : Identification with site-si and Identification with site-mc.

33 PCA (ATTITUDINAL BEHAVIORAL SCALES) Analogously, a PCA was applied to the attitudinal/behavioral psychometric scale. Based on the statistical criteria cited above, three components have been extracted: 1. RES FRIENDS (in favor of renewable energy sources) 2. LOCAL DEVOTED (concern over the economic and environmental conditions of the territory) 3. CONSUMERISTS (individual mostly concerned about individual well being) Notice that for each respondent we have 5 psychometric variables.

34 Latent class model + psychometric variables Table 6. Latent Class Model (4 classes). Variable Beach SI Beach MC Archsite Class1 Class2 Class3 Class *** (0.177) 2.009*** (0.272) 1.136*** (0.286) Utility Model Coefficients 2.077*** (0.223) 0.648*** (0.127) ** (0.198) 0.447*** (0.094) 0.382*** (0.084) ** (0.133) 0.092*** (0.032) 0.235*** (0.037) 0.618*** (0.049) Variable Class1 Class2 Class3 Class4 Constant Local Devoted Consumerists Land owners ID_SI Beach Class Probability Model (0.801) (0.206) (0.209) (0.427) (0.218) ** (1.044) (0.255) (0.244) (0.679) 0.755** (0.296) (1.069) *** (0.294) 1.236*** (0.327) 2.081*** (0.586) (0.287) Property (0.160) 0.417** (0.131) *** (0.149) 0.290*** (0.035) ID_MC Beach 0.474** (0.217) (0.262) *** (0.346) 0 Services (0.126) 0.372*** (0.083) 1.272*** (0.171) (0.030) Wind Farm 0.658* (0.361) (0.526) ** (0.662) 0 Bill Reduction (0.757) 3.346*** (0.584) 8.721*** (0.944) 1.208*** (0.158) Average class probabilities

35 Summarizing... Class 1: there is no compensation that would induce acceptance of the proposed project, individuals are more likely to be identified with the MCbeach Class 2: highly value the SI site; willing to accept compensations; individuals identified with the SI beach Class 3: ok wind mills close to the archaeological site, private firm preferred, public benefits highly valued; this segment is positively associated with consumerists Class 4: relevant attributes are the archaeological site, the public ownership of the firm and the impact on the MC-site. Locally devoted individuals are most probably located in this segment.

36 7. Conclusion Literature The case study Methods The qualitative study The survey Results Conclusion

37 7. Conclusion A case study was selected, where the local economy could possibly benefit from a wind farm development project, and a choice experiment study was designed in order to analyze social acceptance of impacts and to assess possible compensation measures. The survey administered for the choice experiment study included a series of psychometric scales, also defined after the qualitative stage results, that could be used in order to better understand the factors behind certain patterns of choice The results have shown that on average the sampled individuals are willing to accept some impact on valuable sites, provided that social costs can be compensated by means of private (reduction of the electric bill) or public (distribution of profits in the territory, public services) benefits. However, a deeper analysis reveals that respondents had indeed very different valuations of the proposed project.

38 Thank You!