Academic journal article Journal of Leisure Research

Welfare Measurement Convergence through Bias Adjustments in General Population and On-Site Surveys: An Application to Water-Based Recreation at Lake Sevan, Armenia

Academic journal article Journal of Leisure Research

Welfare Measurement Convergence through Bias Adjustments in General Population and On-Site Surveys: An Application to Water-Based Recreation at Lake Sevan, Armenia

Article excerpt

(ProQuest: ... denotes formula omitted.)

Introduction

Travel cost-based demand estimation models rely on actual site visitation data. These data can be collected either on-site or through general population surveys. Each survey type has its own advantages and disadvantages. For example, while general population surveys have the potential to be more broadly representative of a population, they may suffer from known biases such as respondent recall bias if the site in question is infrequently visited or disproportionately zero visits if a large proportion of the survey sample is not visiting the site at all. On the other hand, on-site surveys have the advantage of precision regarding the time, date and certainty of visit. However, they run the risk of over-sampling only those in the population who are avid users of the site while not sampling potential participants should constraints to the participation decision be relaxed. We are aware of two studies that direcdy evaluated the convergence of benefit estimates for the same resource based on data from general population and on-site surveys (Loomis, 2003; Shaw et al., 2003). We extend this literature by evaluating the convergence of demand models based on data from general population and on-site surveys for a resource when adjustments are made for statistical biases unique to each survey mode.

A robust comparison of estimates obtained from each sample requires addressing a number of important statistical issues. In particular, recreation demand measured from a population-based household survey is typically censored due to the observation of a large number of zeros (or non-users of the site). Simply treating all zeros in the sample as users of the site may introduce an upward bias of the demand and welfare measures (Shonkwiler and Shaw, 1996; Haab and McConnell, 1996; Gurmu and Trivedi, 1996).1 On the other hand, on-site surveys have at least two separate issues. The first is that observations of visitation are truncated at one since it surveys only users at the site, while demand estimation requires observations at zero to establish a choke price. A second issue is related to the users surveyed at the site, namely endogenous stratification. On-site survey data may lead to biased standard errors and welfare measures if the sample is cndogenously stratified; i.e., avid users have higher probabilities of being sampled leading to higher trip frequencies being correlated with their characteristics (Shaw, 1988; Englin and Shonkwiler, 1995).

In the case of household surveys, it is possible to resolve the issue by separating the recreation 'participation' decision from the trip 'quantity' decision using sample selection models, thus reducing the bias introduced by non-users of the site (Haab and McConncll, 1996; Gurmu and Trivedi, 1996). In the case of on-site surveys, it is possible to correct for the potential bias by providing adjustments to the distribution function (Shaw, 1988; Englin and Shonkwiler, 1995). Loomis (2003) and Shaw et al. (2003) show that after adjusting for truncation and endogenous stratification in the on-site survey, the welfare estimates are comparable with the results from the household surveys. However, neither study accounts for the possibility of zero-inflation (or excess zeros) in the household survey sample.

In this paper, we test the hypothesis of whether the household and on-site demand estimation yield similar welfare measures after accounting for the biases discussed above. For this purpose, we construct single-site travel cost models for a household and on-site survey conducted at Lake Sevan, Armenia. The single-site travel cost model is preferred in this case as it facilitates demand estimation for visitation and associated welfare comparisons. The context of the application is also of policy relevance. Lake Sevan is a unique recreational and historically significant resource with no practical substitutes. In the past 50 years, the level of the lake has fallen 18 meters, with severe physical and ecological consequences. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.