This is an openaccess article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Interactive Journal of Medical Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.ijmr.org/, as well as this copyright and license information must be included.
In Canada, 11.5% to 15.7% of couples suffer from infertility. Anovulation, or failed ovulation, is one of the main causes of infertility in women. In Quebec, the treatment for ovulation induction and other services related to assisted reproductive technology (ART) have been partially reimbursed by the government since 2010.
This study aimed to compare the willingness to pay (WTP) of women of childbearing age to receive drug treatment in the event of failed ovulation according to 3 different contingent valuation methods.
The following elicitation techniques were used: simple bid price dichotomous choice (DC), followed by an openended question (DCOE), and a simplified multiplebounded discrete choice (MBDC). Each participant was randomly assigned to 1 of 3 elicitation techniques. Bid prices ranged from Can $200 to Can $5000. Of the 7 bid prices, 1 was randomly proposed to each participant in the DC and DCOE groups. For the DCOE group, if the answer to the DC bid price was
The survey was conducted from 2009 to 2010 with a total sample of 680 women. Analyses were performed on 610 respondents (199 DC, 230 DCOE, and 181 MBDC). Of the 70 respondents who were excluded, 6 did not meet the age criterion, 45 had an annual income less than Can $2500, and 19 did not respond to the WTP question. Mean WT
A positive WTP for ovulation induction was found in Quebec. Adding a followup question resulted in more accurate WT
According to the World Health Organization, infertility is defined as the inability to conceive after 12 months of unprotected sex [
The benefit of the drug treatment to induce ovulation in adult female patients with infertility is generally measured by the proportion of women who ovulate as a result of such treatment [
Several methods can be used to estimate this monetary value, including the contingent valuation method (CVM) [
Although different variants exist, the 4 main techniques reported in the literature are bidding game (BG), payment card (PC), openended (OE) questions, and dichotomous choice (DC) [
In this study, 3 CVMs were compared: DC, DC followed by an OE question (DCOE), and a simplified multiplebounded discrete choice (MBDC), which is very similar to a BG. These 3 methods were chosen because of their simplicity and because they are widely used in the literature. The main objective of this study was to assess the WTP of women of childbearing age to receive a drug treatment in the event of failed ovulation according to the 3 different CVMs. More specifically, this study aimed to assess whether these 3 techniques generate statistically different WTPs and, if so, to determine which method is the most accurate.
The data used in this study were from a survey conducted in Quebec between January 2009 and February 2010. Inclusion criteria required participants to be a woman aged 18 to 45 years and to agree to complete the survey in French. Women were excluded if they had an annual income less than or equal to Can $2500 (this amount corresponded to the middle of the lowest bracket proposed for annual income and because it is unlikely that women can afford an infertility treatment with this income) or if they did not respond to the WTP question. No sample size was calculated, but 200 patients per elicitation method were targeted, which is the usual number for this type of study [
The data were collected through a Web survey, which were first distributed using an email listing from previous studies (ie, respondents from previous studies who accepted to be contacted for future research) and were then distributed by a Web survey company. Participants were randomly allocated to 1 of the 3 elicitation methods tested (DC, DCOE, or MBDC).
Each questionnaire had 3 main components: introduction, socioeconomic variables, and WTP questions. The introduction presented a definition of infertility, gave the prevalence of infertility (including infertility related to ovulation failure), the type of treatment associated, the probability of success, and the associated risks. The socioeconomic variables included age (years), weight (kilograms), height (centimeters), employment status, is the job stressful (yes or no), individual annual income (using brackets), educational level, civil status, number of children, smoking (yes or no), general health (5 levels), fertility problems (yes or no), actually pregnant (yes or no), desire for a child (yes or no), and a ranking of 10 items (eg, have good health, have children, be financially comfortable).
The third component was the WTP question about receiving ovulation failure treatment, along with another question about the degree of certainty of the respondent’s answer. For DC and DCOE, 7 price levels were randomly assigned to different versions of the questionnaire (Can $200, Can $500, Can $1000, Can $1500, Can $2000, Can $3000, and Can $5000). For the DCOE, if the answer was
Overall, 2 comparison criteria were used to judge the accuracy of the estimates. The obtained estimated WTP and standard deviations were compared between the 3 subsamples. The efficiency of the estimates was measured with the ratio of confidence interval on mean WTP. The expected efficiency associated with a followup WTP question is based on the fact that the confidence intervals should be narrower and closer to the mean WT
To calculate the mean WT
The total sample consisted of 680 women; of these women, 215 responded to the DC, 255 responded to the DCOE, and 210 responded to the MBDC (
Flowchart of respondents randomly distributed among the 3 elicited methods (dichotomous choice, dichotomous choice followed by an openended question, or multiplebounded discrete choice).
Descriptive statistics.
Variable  Dichotomous choice (n=199)  Dichotomous choice followed by an openended question (n=230)  Multiplebounded discrete choice (n=181)  
Age (years), mean (range)  32 (1845)  30 (1845)^{a}  30 (1845)^{a}  
Annual income (Can $), mean (range)  42,622 (7500130,000)  37,615 (7500130,000)^{b}  34,475 (7452130,000)^{a,c}  



Secondary school  31 (15.6)  41 (17.8)  42 (23.2)^{b,}^{d}  

College  78 (39.2)  93 (40.4)  63 (34.8)  

University  90 (45.2)  96 (41.7)  76 (42.0)  
Very good health, n (%)  58 (29.2)  43 (18.7)^{a}  36 (19.9)^{a}  
Current smoker, n (%)  39 (19.6)  45 (19.6)  51 (28.2)^{a,e}  
Stressful job, n (%)  174 (87.4)  198 (86.1)^{a}  145 (80.1)^{c}  
Ovulation failure, n (%)  7 (3.5)  14 (6.1)^{a}  5 (2.8)^{e}  
Infertility problem, n (%)  26 (13.1)  25 (10.9)^{a}  22 (12.2)^{e}  
Having a child is important, n (%)  125 (62.8)  147 (63.9)  108 (59.7)  
Employee, n (%)  158 (79.4)  183 (79.6)  130 (71.8)^{b,e} 
^{a}
^{b}
^{c}
^{d}
^{e}
As we expected, the higher the offered bid price, the lower the proportion of
Cumulative decreasing frequencies of positive answers yes. DC: dichotomous choice; DCOE: dichotomous choice followed by an openended question; MBDC: multiplebounded discrete choice.
The results of the probit analysis are presented in
Estimation results with probit and bivariate analysis.
Variables  Dichotomous choice  Dichotomous choice followed by an openended question  Multiplebounded discrete choice  
Probit (R^{2}=0.2375; N=199)  Probit (R^{2}=0.0863; N=230)  Bivariate (R^{2}=0.0633; N=230)  Probit (R^{2}=0.2807; N=1245)  
Coefficient  Coefficient  Coefficient  Coefficient  
Bid  − 
− 
− 
− 
—^{d}  —  − 
− 

Age (years)  −0.0264  −1.54  −0.0171  −1.15  −11.50  −0.52  − 
− 

Annual income  0.00000310  0.69  0.00000509  1.00  0.00749  0.97 



University 









Very good health 


−0.0571  −0.24  −110.2  −0.31  0.0157  0.15  
Infertility problem  −0.165  −0.43  −0.113  −1.07  −170.3  −1.08  −0.202  −1.36  
Ovulation failure  0.728  1.03  0.0395  0.29  −34.94  −0.17  0.384  1.39  
Current smoker  0.401  1.49  −0.388  −1.62  − 
− 



Stressful job 


−0.0770  −0.30  83.55  0.21  −0.0546  −0.49  
Having a child is important 


0.312  1.63  209.1  0.73 



Employee  0.421  1.63  0.242  0.99  207.8  0.58 



Constant  0.149  0.25  0.280  0.48 





Sigma  —  —  —  — 


—  — 
^{a}Italics indicate that the term is statistically significant.
^{b}
^{c}
^{d}Not applicable.
^{e}
The coefficients of the explanatory variables of the bivariate model directly illustrate women’s WTP. For DCOE, when the offer (bid) was higher, the WTP was significantly lower (
The probit model results are presented in
Estimation results using the model by Welsh and Poe.
Variables  Definitely yes (N=181)^{a}  Probably yes (N=181)^{a}  Do not know (N=181)^{a}  
Estimate  Estimate  Estimate  



Constant  − 
− 
− 
− 
− 
− 




Age (years)  −0.00412  −0.31  − 
− 
−0.0133  −1.05  

Annual income  0.00000230  0.48  0.00000126  0.27  0.00000130  0.28  

University  0.271  1.42 


0.280  1.44  

Very good health  0.0331  0.16  0.0661  0.30  −0.115  −0.54  

Infertility problem  −0.175  −0.59  −0.192  −0.62  −0.130  −0.41  

Ovulation failure  0.460  0.74  0.309  0.51  0.135  0.23  

Current smoker  0.213  1.12  0.218  1.12 




Stressful job  −0.0288  −0.12  0.177  0.74  −0.175  −0.74  

Having a child is important  0.128  0.72  0.279  1.54 




Employee  −0.156  −0.77  −0.0866  −0.42  0.0772  0.38  

Constant  0.558  1.01  0.778  1.41 


^{a}
^{b}Italics indicate that the term is statistically significant.
^{c}
^{d}
^{e}
Comparing DC with DCOE, we can see that adding 1 more question after the DC WTP question improves the accuracy of the WTP estimates. However, our results also revealed the existence of an anchoring effect in the DCOE approach, where the implicit WT
Mean willingness to pay estimated with probit, bivariate, or Krinsky and Robb methods.
Method  Average WTP^{a}  Mean difference (95% CI)  CI/WTP  



DC^{b} (N=199)  4033.26  4386.01 (1840.25 to 6226.26)  1.09  

DCOE^{c} (N=230)  1857.90  1945.1 (885.35 to 2830.45)  1.05  

MBDC^{d}—probit (N=1245)  1630.63  556.64 (1352.31 to 1908.95)  0.34  

MBDC—definitely yes (N=181)  1516.73  2796.86 (118.30 to 2915.16)  1.84  

MBDC—probably yes (N=181)  1871.22  4926.29 (−591.92 to 4334.37)  2.63  

MBDC—do not know (N=181)  2514.49  3032.39 (998.29 to 4030.68)  1.21  



DC (N=199)  4750.18  6985.57 (3911.67 to 10,897.24)  1.47  

DCOE (N=230)  1857.90  1924.93 (895.44 to 2820.37)  1.04  

MBDC—probit (N=1245)  1701.37  896.51 (1103.43 to 1999.94)  0.53 
^{a}WTP: willingness to pay.
^{b}DC: dichotomous choice.
^{c}DCOE: dichotomous choice followed by an openended question.
^{d}MBDC: multiplebounded discrete choice.
A total of 3 elicitation techniques were used to assess women’s WTP for an ovulation induction treatment in case of failed ovulation. One of the main objectives was to discover whether a significant difference exists between different WTP elicitation approaches.
The results show that the DC technique yielded higher estimated WTP than the other 2 techniques. The higher value for WTP with DC methods is consistent with the literature [
The value added by a followup question is based on the fact that the confidence intervals are closer to the estimated WTP and that the latter is, therefore, more accurate [
The estimated WTP in our study shows dissimilar results to the study by Poder et al [
Our study gave coefficients of the expected signs, although the positive coefficient of the variable
This study has a number of limitations, so the results should be interpreted with caution. As we used a convenience sample, we cannot say with certainty that our regression equations will give the same results if applied to a larger or different sample because of the lack of representativeness. Another limitation of this study is that our approach used a fixed predetermined starting price of Can $1500 in the MBDC technique. This choice may have led to an anchoring effect, as individuals focus on the first proposition (Can $1500), and thus, their answers to the second and third questions may be influenced by the first bid offered. Unlike other techniques (DC and DCOE) that use random starting prices between Can $200 and Can $5000, this anchoring effect cannot be assessed in the MBDC.
Each of the 3 elicitation techniques has its disadvantages. The DC technique yielded higher estimated WTP with little WTP information (ie, only 1 WTP question, so we only know if their maximum WTP is higher or lower to the bid proposed). However, the DC technique is more similar to the real market situation of
In their study on psychosocial services for couples in infertility treatment, Read et al [
We also compared the mean WTPs of the different techniques and found significant differences among the estimated WTPs. Adding a followup question resulted in more accurate WTPs but created anchoring biases. Results also indicated that the simplified MBDC technique provided more accurate estimates of the WTP with a smaller and, therefore, more efficient confidence interval. Consequently, for the purpose of a more efficient fiscal policy, the simplified MBDC technique provided the most appropriate WT
Estimates of willingness to pay.
assisted reproductive technology
bidding game
contingent valuation method
dichotomous choice
dichotomous choice followed by an openended question
multiplebounded discrete choice
payment card
openended
willingness to pay
The authors would like to thank Renald Lemieux and Nathalie Carrier for their support during the execution of this study and for their comments. TGP is a member of the Fonds de recherche du Québec—Santé–funded Centre de recherche de l’Institut universitaire en santé mentale de Montréal.
None declared.