Skip to main content
Log in

Combining Revealed Preference Data with Stated Preference Data: A Latent Class Approach

  • Published:
Environmental and Resource Economics Aims and scope Submit manuscript

Abstract

A substantial literature exists combining data from revealed preference (RP) and stated preference (SP) sources, aimed either at testing for the convergent validity of the two approaches used in nonmarket valuation or as a means of drawing on their relative strengths to improve the ultimate estimates of value. In doing so, it is assumed that convergence of the two elicitation approaches is an “all or nothing” proposition; i.e., the RP and SP data are either consistent with each other or they are not. The purpose of this paper is to propose an alternative framework that allows for possible divergence among individuals in terms the consistency between their RP and SP responses. In particular, we suggest the use of a latent class approach to segment the population into two groups. The first group has RP and SP responses that are internally consistent, while the remaining group exhibits some form of inconsistent preferences. An EM algorithm is employed in an empirical application that draws on the Alberta and Saskatchewan moose hunting data sets used in earlier combined RP and SP exercises. The empirical results suggest that somewhere between one-third and one-half the sample exhibits consistent preferences. We also examine differences in welfare estimates drawn from the two classes.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Similar content being viewed by others

Notes

  1. See Randall (1994) and Azevedo et al. (2003) for alternative perspectives on the presumed reliability of RP results.

  2. The model specified here is a site selection model, rather than a model that also characterizes the participation decision, as in the repeated logit framework of Morey, Rowe and Watson Morey et al. (1993). We focus on the site selection aspect of the individual’s decision to be consistent with the earlier analyses of these same databases by Adamowicz et al. (1997), Haener et al. (2001), and von Haefen and Phaneuf (2008).

  3. Individual specific characteristics such as age, gender and education can also impact the site utilities, typically through interactions between individual and site characteristics. For now, we ignore these interaction effects for the sake of notational simplicity, but incorporate them later in both the Monte Carlo analysis and subsequent application.

  4. The emphasis in our paper is on relaxing the assumption that consistency between the RP and SP responses is an all or nothing proposition. However, as suggested by a reviewer, a natural generalization of our framework would be to allow heterogeneity within each of the consistent and inconsistent classes. This could be done using a continuous mixture (random parameters) model for each class or by introducing latent subclasses for both the consistent and inconsistent classes. In the latter case, information criteria (e.g., AIC and BIC) could be used in selecting the number of subclasses.

  5. Estimates for the alternative specific constants \(\alpha _j^{RP}\) and \(\xi ^C_j\) are not reported in Table 1 for the sake of space, but are available from the authors upon request. Also, estimates for the parameters \(\beta ^{RP}\) are obtained through a second stage regression based on the fitted alternative specific constants from the first stage and using the relationship in (3).

  6. As noted in footnote 4 above, a generalization of our modeling framework would be to allow for heterogeneous preferences within each of the consistent and inconsistent classes. As an initial exploration into this possibility, we also conducted a generated data experiment using two latent subclasses for each of the main classes. Our model was then estimated using (a) a single subclass for each of the main classes and (b) two subclasses for each of the main classes. The results from this exercise are reported in Appendix Table 10. In all of the Monte Carlo runs, the two subclass model was consistently preferred to the single subclass model on the basis of both AIC and BIC measures and the two subclass model parameter estimates were generally consistent with the underlying data generating process. This is an area for future research, but beyond the scope of the current manuscript.

  7. See McLeod et al. (1993) for additional details regarding the sampling and data collecting procedures.

  8. In the empirical setting, we include a dummy variable for ‘not hunting’ (SP dummy) to capture impact of the opt-out option.

  9. The log-likelihood values for the estimated models are as follows. For the Alberta data set: SC-C: −5377.94; SC-RP: −1868.11; SC-SP: −3468.56; LC: −4813.10. For the Saskatchewan data set: SC-C: −7482.15; SC-RP: −1162.66; SC-SP: −6301.20; LC: −6896.32.

  10. The parameter estimates reported here for the single class models have the same signs and are similar in magnitude to those reported in von Haefen and Phaneuf (2008), though the specifications differ in that von Haefen and Phaneuf incorporate a mixed logit structure.

  11. One exception is the main effect for the “unpaved” site access, since this characteristic varies across sites and individuals because individuals choose different roads to access the sites.

  12. In general, one could estimate \(\beta ^{RP}\) using a second stage regression, as suggested by Murdock Murdock (2006). Doing so would allow for a total of five welfare measures, paralleling those available for the site loss scenario. However, in the current case we have too few of sites to do so. Even with more sites, endogeneity concerns would require the use of suitable instruments, which may not always be available.

  13. Since the parameters used in constructing Measures #3 through #5 are jointly estimated, we can test for pairwise differences among the measures. In all pairwise comparisons, the differences are not statistically significant.

  14. This generalization was estimated using the Alberta Moose Hunting data, but none of the available demographic factors were found to significantly impact class membership.

References

  • Abidoye B, Herriges JA, Tobias J (2012) Controlling for observed and unobserved site characteristics in RUM models of recreation demand. Am J Agric Econ 94(5):1070–1093

    Article  Google Scholar 

  • Adamowicz W, Louviere J, Williams M (1994) Combining revealed and stated preference methods for valuing environmental amenities. J Environ Econ Manag 26:271–292

    Article  Google Scholar 

  • Adamowicz W, Swait J, Boxall P, Louviere J, Williams M (1997) Perceptions vs. objective measures of environmental quality in combined revealed and stated preference models of environmental valuation. J Environ Econ Manag 32:65–84

    Article  Google Scholar 

  • Azevedo C, Herriges JA, Kling C (2003) Combining revealed and stated preference: consistency tests and their interpretations. Am J Agric Econ 85:525–537

    Article  Google Scholar 

  • Ben-Akiva M, Bradley M, Morikawa T, Benjamin J, Novak T, Oppewal H, Rao V (1994) Combining revealed and stated preferences data. Mark Lett 5(4):335–350

    Article  Google Scholar 

  • Ben-Akiva M, Morikawa T (1990) Estimation of switching models from revealed preferences and stated intentions. Transp Res A 24A(6):485–495

    Article  Google Scholar 

  • Bishop RC, Heberlein TA (1979) Measuring values of extra market goods: Are indirect measures biased? Am J Agric Econ 61(5):926–930

    Article  Google Scholar 

  • Boxall PC, Adamowicz WL (2002) Understanding heterogeneous preferences in random utility models: a latent class approach. Environ Res Econ 23:421–446

    Article  Google Scholar 

  • Breffle W, Morey E, Thacher J (2011) A joint latent-class model: combining likert-scale preference statements with choice data to harvest preference heterogeneity. Environ Res Econ. doi:10.1007/s10640-011-9463-0

    Google Scholar 

  • Bujosa A, Riera A, Hicks RL (2010) Combining discrete and continuous representations of preference heterogeneity: a latent class approach. Environ Res Econ 47:477–493

    Article  Google Scholar 

  • Cameron TA (1992) Combining contingent valuation and travel cost data for the valuation of nonmarket goods. Land Econ 68(3):302–317

    Article  Google Scholar 

  • Dosman D, Adamowicz W (2006) Combining stated and revealed preference data to construct an empirical examination of intrahousehold bargaining. Rev Econ Househ 4:15–34

    Article  Google Scholar 

  • Eom Y-S, Larson DM (2006) Improving environmental valuation estimates through consistent use of revealed and stated preference information. J Environ Econ Manag 52:501–516

    Article  Google Scholar 

  • Evans K, Herriges JA (2010) Rounding in recreation demand model: a latent class count model, ISU working paper

  • Greene WH, Hensher DA (2010) Revealing additional dimensions of preference heterogeneity in a latent class mixed multinomial logit model, Working paper

  • Haener MK, Boxall PC, Adamowicz WL (2001) Modeling recreation site choice: do hypothetical choices reflect actual behavior. Am J Agric Econ 83(3):629–642

    Article  Google Scholar 

  • Huang J-C, Haab TC, Whitehead JC (1997) Willingness to pay for quality improvements: Should revealed and stated preference data be combined? J Environ Econ Manag 34(3):240–255

    Article  Google Scholar 

  • Jeon Y, Herriges JA (2010) Convergent validity of contingent behavior responses in models of recreation demand. Environ Res Econ 45:223–250

    Article  Google Scholar 

  • Kuriyama K, Michael Hanemann W, Hilger JR (2010) A latent segmentation approach to a Kuhn–Tucker Model: an application to recreation demand. J Environ Econ Manag 60:209–220

    Article  Google Scholar 

  • McFadden D, Train K (2000) Mixed MNL models for discrete response. J Appl Econom 15(5):447–470

    Article  Google Scholar 

  • McLeod K, Boxall PC, Adamowicz WL, Williams M, Louviere JJ (1993) The incorporation of non-timber goods and services in integrated resource management: an introduction to the Alberta Moose Hunting Study. Interim project report. Project Report No. 93-12. Department of Rural Economy, University of Alberta, Edmonton, Alberta, 60 p

  • Morey E, Rowe RD, Watson M (1993) A repeated nested-logit model of Atlantic salmon fishing. Am J Agric Econ 75(3):578–592

    Article  Google Scholar 

  • Morey E, Thacher J, Breffle W (2006) Using angler characteristics and attitudinal data to identify environmental preference classes: a latent–class model. Environ Res Econ 34(1):91–115

    Article  Google Scholar 

  • Murdock J (2006) Handling unobserved site characteristics in random utility models of recreation demand. J Environ Econ Manag 51:1–25

    Article  Google Scholar 

  • Randall A (1994) A difficulty with the travel cost method. Land Econ 70:88–96

    Article  Google Scholar 

  • Train K (2009) Discrete choice methods with simulation, 2nd edn. Cambridge University Press, Cambridge

    Book  Google Scholar 

  • von Haefen RH (2003) Incorporating observed choice into construction of welfare measures from random utility models. J Environ Econ Manag 45:145–165

    Article  Google Scholar 

  • von Haefen RH, Phaneuf DJ (2008) Identifying demand parameters in the presence of unobservables: a combined revealed and stated preference approach. J Environ Econ Manag 56:19–32

    Article  Google Scholar 

  • Whitehead JC, Pattanayak S, Van Houtven G, Gelso B (2008) Combining revealed and stated preference data to estimate the nonmarket value of ecological services: an assessment of the state of the science. J Econ Surv 22:872–908

    Article  Google Scholar 

  • Whitehead JC, Phaneuf DJ, Dumas CF, Herstine J, Hill J, Buerger B (2010) Convergent validity of revealed and stated behavior with quality change: a comparison of multiple and single site demands. Environ Res Econ 45:91–112

    Article  Google Scholar 

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Joseph A. Herriges.

Appendix

Appendix

See Tables 8, 9 and 10.

Table 8 Generated data experiments—Model 2 parameter estimates
Table 9 Generated data experiments—Model 3 parameter estimates
Table 10 Generated data experiment: 2 subclasses in each consistent and inconsistent class (N \(=\) 2000)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Jeon, H., Herriges, J.A. Combining Revealed Preference Data with Stated Preference Data: A Latent Class Approach. Environ Resource Econ 68, 1053–1086 (2017). https://doi.org/10.1007/s10640-016-0060-0

Download citation

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10640-016-0060-0

Keywords

Navigation