lations to estimate values. Formats that elicit more 

 than 15 percent protest responses in pretests 

 should be discarded, since a high incidence of pro- 

 test bids may indicate that some nonzero bids are 

 also distorted. 



(c) Noniterative bidding formats. (1) Noniterative 

 bidding formats are adaptable to implementation 

 with mail surveys. There are two kinds of nonitera- 

 tive formats: close-ended, which ask respondents 

 to answer "yes" or "no" to a single stated value; 

 and open-ended, which ask the respondent to write 

 down the maximum amount he would be willing to 

 pay. A variant of the open-ended format asks the 

 respondent either to select his maximum WTP from 

 a list of stated discrete values or to write down his 

 maximum WTP. Noniterative bidding formats are 

 unlikely to be as reliable as iterative formats. 



(2) Noniterative mail survey formats may be used 

 only for analysis of small projects. These formats 

 must, to the extent practicable, have the basic attri- 

 butes of the personal interview formats described 

 above. Survey instruments should include color 

 photographs and, if appropriate, other nonverbal 

 stimuli. 



(3) Open-ended bidding formats should be used 

 with one half of the sample and close-ended for- 

 mats with the other half. The bids obtained should 

 be analyzed to determine if the format influences 

 the results to a significant degree. Examples of 

 these formats are presented below. 



(4) Open-ended. "Due to pressures of population 

 growth and economic development, 10 miles of 

 trout stream such as that shown in the accompany- 

 ing photograph are likely to be converted to other 

 uses (e.g., a reservoir) and thus lost for trout fish- 

 ing. Assume that the only way to preserve this 10- 

 mile stretch for trout fishing is for trout fishermen to 

 agree to buy an annual pass to fish in that stream 

 segment. The money collected would pay for pres- 

 ervation of the stream section. If the stream seg- 

 ment was miles from your home, and you 



could expect to catch — trout in a typical day's 

 fishing there, what is the maximum amount you 

 would pay for the annual fishing pass? Answer: 

 $ per year. 



(5) Closed-ended. The information presented in 

 the open-ended format does not change, but the 

 final question reads: " ... g^id an annual 



fishing pass costs $ (assign dollar amounts 



randomly to respondents), would you buy one? 

 Answer: Yes — . No — ." 



(d) Use estimation with CVMs. (1) All of the con- 

 tingent valuation procedures described above gen- 

 erate annual value estimates directly, instead of 

 first generating values per user day and then esti- 

 mates of expected user days. The "annual value 



estimation" procedure is superior because it is 

 more reliable, it automatically corrects for the eco- 

 nomic influence of existing recreation opportunities, 

 and it is better adapted to estimating activity and 

 existence values where both are important. 



(2) Contingent valuation formats can also be de- 

 signed to estimate values per user day but can 

 have questions worded in terms of a day's activity. 

 In the case of proposed increments, great care 

 must be taken to determine the respondent's valua- 

 tion of a day at the proposed site, given the contin- 

 ued availability of existing sites. Estimates of use 

 may be made either by collecting such information 

 as part of the survey or by other approved meth- 

 ods. 



(3) To collect use information in the survey, pro- 

 ceed as follows: 



(i) For decrements in recreation opportunities, 

 ask (A): how many trips the household made (/) 

 last year or {2) in a typical year, if last year was un- 

 usual for any reason; (B) how many days the trip 

 lasted; and (C) how many household members par- 

 ticipated in each trip. 



(ii) For increments, ask (A): the same information 

 as for decrements, but about existing recreation 

 sites similar to the proposed increment. Then, if the 

 proposed increment (described with verbal and 

 nonverbal stimuli) were available, (B) how many 

 trips, for how long, and with how many family mem- 

 bers for the proposed increment; and (C) how many 

 trips, for how long, and with how many family mem- 

 bers in total for both the existing and proposed 

 sites. 



(e) Using contingent valuation methods. Contin- 

 gent valuation methods can be used to develop 

 value estimator models or to estimate recreation 

 benefits for a specific proposed project. These two 

 uses are discussed below. 



(1) Value estimator models, (i) Value estimator 

 models (VEMs) are statistical models of the rela- 

 tionships between the bid and selected characteris- 

 tics of the site(s) and user populations. A typical 

 model has the form: 



F(Eh, D,k, Ck, Ak. S,K. Q„ I,) 



Where 



V,,, is the value to household k of the specified change in recre- 

 ation opportunity at site j. 



Ek is a vector of social and demographic variables pertaining to 

 household k. typically including income, ethnicity, and education. 



D,i, IS distance from the home of k to site j. 



Ck IS a measure of the capacity use of the existing stock of rec- 

 reation facilities similar to those at site j in the market area cen- 

 tered at k's home 



81 



