Date post: | 21-Dec-2015 |
Category: |
Documents |
Upload: | maria-copeland |
View: | 219 times |
Download: | 0 times |
Department of Environmental EngineeringDemocritus University of Thrace
Angeliki, N. Menegaki1, Søren Bøye Olsen2, and Konstantinos P. Tsagarakis3
3 Democritus University of Thrace, Department of Environmental Engineering, Business Economics and Environmental Technology Lab, www.beteco.org
Towards a common standard – A reporting checklist for web-based stated preference
valuation surveys
1 a) Organismos Georgikon Asfaliseon, Regional Branch of Eastern Macedonia & Thrace, 69100, Komotini, Greece
b) Hellenic Open University, Parodos Aristotelous 18, 26335, Patras, Greece
2 Department of Food and Resource Economics, Faculty of Science, University of Copenhagen, Rolighedsvej 25, 1958 Frederiksberg C, Denmark
• Why we need valuation?
Προσεγγίσεις και μέθοδοι αξιολόγησης όταν δεν υπάρχουν αγορές
Non market valuation methods
Δαπάνες προς αποφυγή
Averting expenditures
Ωφελιμιστική τιμολόγηση
Hedonic pricing
Μέθοδος κόστους ταξιδιού
Travel cost method
Πιθανολογική αξιολόγηση
Contingent valuation
Πειραματική αξιολόγηση
Choice experiments
Αγορές υποκατάστασης
Surrogate markets
Υποθετικές αγορές
Hypothetical markets
Μεταφορά ωφελειών
Benefit transfer
Revealed preference techniques
Stated preference techniques
We considered 182 web based valuation surveys
an update of the …
Figure 1. Numbers of collected surveys (2001-205).
Figure 2. Valuation methods perused in web surveys (2001-2015).
Figure 3. Allocation of incentives perused in web surveys (2001-2015).
Figure 5. Web surveys topics of the collectedweb surveys (2001-2015)
Important Findings
• Finding 1: The response rate is higher, the more recent the year of publication of the study.
• Finding 2: There is difference in response rates among continents.
• Finding 3: There is difference in response rates among the different research companies that undertake sampling and survey administration.
• Finding 4: There is no difference in the response rate provided by surveys that use the cash incentive and the prize incentive.
Important Findings
• Finding 5: There is difference in the response rate between studies providing cash and prize incentives and studies providing no incentives.
• Finding 6: There is no difference in the response rate between CV and CE methods.
• Finding 7: There is no difference in the response rate produced by single mode and mixed mode studies.
• Finding 8: The response rate is independent of the topic of the survey.
sStimuli & remuneration
How does the information menu appear?Has respondent effort been measured? (How manyrespondents chose to access supplementarymaterial? How long did they spend reading that?)Was there a scrolling design (1 page) or a screen byscreen design (more pages)?Were there text entry boxes or multiple choicequestions? Were there long entry boxes or shortentry boxes?Were there drop-down menus?Were respondents allowed to drop the survey andresume later?Was there a real-time presentation of the surveyresults, which allowed respondents to adapt theiranswer in order to reach the required target?How is innovativeness of the study proved?How are ethics in the study guaranteed? (e.g.Approval by ethics committee)How was confidentiality guaranteed?How was variance of scope sensitivity examined?Has the survey taken into consideration Dillman'sprinciples?How were respondents prevented from answeringmultiple times?In case of mixed mode survey, did respondents havethe option of choosing between the two modes?Has the survey explicitly explained its usefulness torespondents?Did the two mode surveys take place concurrently orsequentially?Were online focus groups and chat rooms available?Did the survey allow respondents the opportunity towrite their opinion about it? (e.g. Did they find theelicitation format easy to understand?)Was mixed mode simultaneous or consecutive?What type of paradata (if any) were observed?
Description of the sampling methodDetails of panelists' recruitment by theresearch companyHow were computer non-owners orcomputer illiterate people handled?(e.g. was there a hotline available? Didthe survey provide its own PCs?)Did the survey collect information aboutrespondents who decided not toparticipate in the survey?If there was a convenience sample:Elaborate on its representativenessIf sample was random: Elaborate on itsrandomness.Describe piloting process
How many times was a respondentinvited for participation?Questionnaire positioning: Was it sentby Email or communicated by phone?Was it embedded in a web page? Wasit communicated with a link? Was anindividual password sent?Was a screening phone call used?Were key officials contacted to sendEmails to their employees?For how long was the invitation keptalive?After how much time were initialinvitations followed by Emailreminders? Were there automaticreminders?Was the invitation reminder sent byEmail or another form ofcommunication?
Did colored photographs have an influence onrespondents?What animations, graphics, sounds, fonts wereused?Was any introductory video played?What incentive was given to participants (gift, cash,prize etc.)?If no gift was available, is that clearly stated?Was completion time measured?What was speed of return for a completedquestionnaire?Was a minimum completion time imposed?
Questionnaire design & webadministration
Sampling frame & selfselection
Invitation
Figure 8. Checklist for web-surveys
Figure 8. Web add-ins in the process of a typical stated
preference survey
Debrief
A non-response sample can be better analyzed and correlated to its demographics.Trace response contradictions to debrief questions compared to answers in main questions (also with cross-tabulations).Collection of paradata is possible (e.g. completion time, number of clicks, number of changes made, use of help functionsetc).Text areas can be added to those who wish to fill them.Web panel data allow for comparisons of results over time, because the individual information is continually updated (e.g.Knowledge Networks updates profiling data every 2-4 months).
Main (valuation) questions
Experiment between scrollable and screen by screen designs.Upon completion of a WTP question, the web design can immediately calculate how much this amounts is in the personalbudget and remind the respondent of his budget constraint in order to reduce overstatement and make the exercise morerealistic.Offer clever filters of don't know responses.Offers discrete environment.Can offer a hotline for guidance and technical support.Experiment on issues such as questionnaire length, framing, order, scope, open vs closed questions etc.Can estimate respondent fatigue and adapt questionnaire to prevent satisficing.Avoid multiple answers when only one is required.Reduces information and questionnaire space with drop-down menus. before the end of the survey.
Piloting
Online focus groups and chat rooms are possible.Pre-test surveys with multiple browsers and screen settings.
Warm-up questions
Can screen respondent panel further with questions that aim to find out whether one is in the right frame of mind to answer,or record his sentiments.Keep respondents motivated with progress indicator features on the screen.
Information session
Better visualization (photos, videos), colours, sounds, fonts.The respondent can click on the information and have it read aloud and clearly.The information can expand with various pop-up tools (depending on how much information the person needs based on thelevel of information he already has at hand).The information can be heard from a voice instead of being given a separate instructions manual. Also the instructions can begiven step by step when needed.
Sampling frame
It is performed with web panels (mainly recruited from research companies etc).Personalized invitations and reminders.
Figure 8. Web add-ins in the process of a typical stated
preference survey
Debrief
A non-response sample can be better analyzed and correlated to its demographics.Trace response contradictions to debrief questions compared to answers in main questions (also with cross-tabulations).Collection of paradata is possible (e.g. completion time, number of clicks, number of changes made, use of help functionsetc).Text areas can be added to those who wish to fill them.Web panel data allow for comparisons of results over time, because the individual information is continually updated (e.g.Knowledge Networks updates profiling data every 2-4 months).
Main (valuation) questions
Experiment between scrollable and screen by screen designs.Upon completion of a WTP question, the web design can immediately calculate how much this amounts is in the personalbudget and remind the respondent of his budget constraint in order to reduce overstatement and make the exercise morerealistic.Offer clever filters of don't know responses.Offers discrete environment.Can offer a hotline for guidance and technical support.Experiment on issues such as questionnaire length, framing, order, scope, open vs closed questions etc.Can estimate respondent fatigue and adapt questionnaire to prevent satisficing.Avoid multiple answers when only one is required.Reduces information and questionnaire space with drop-down menus. before the end of the survey.
Piloting
Online focus groups and chat rooms are possible.Pre-test surveys with multiple browsers and screen settings.
Warm-up questions
Can screen respondent panel further with questions that aim to find out whether one is in the right frame of mind to answer,or record his sentiments.Keep respondents motivated with progress indicator features on the screen.
Information session
Better visualization (photos, videos), colours, sounds, fonts.The respondent can click on the information and have it read aloud and clearly.The information can expand with various pop-up tools (depending on how much information the person needs based on thelevel of information he already has at hand).The information can be heard from a voice instead of being given a separate instructions manual. Also the instructions can begiven step by step when needed.
Sampling frame
It is performed with web panels (mainly recruited from research companies etc).Personalized invitations and reminders.
Figure 8. Web add-ins in the process of a typical stated
preference survey
Debrief
A non-response sample can be better analyzed and correlated to its demographics.Trace response contradictions to debrief questions compared to answers in main questions (also with cross-tabulations).Collection of paradata is possible (e.g. completion time, number of clicks, number of changes made, use of help functionsetc).Text areas can be added to those who wish to fill them.Web panel data allow for comparisons of results over time, because the individual information is continually updated (e.g.Knowledge Networks updates profiling data every 2-4 months).
Main (valuation) questions
Experiment between scrollable and screen by screen designs.Upon completion of a WTP question, the web design can immediately calculate how much this amounts is in the personalbudget and remind the respondent of his budget constraint in order to reduce overstatement and make the exercise morerealistic.Offer clever filters of don't know responses.Offers discrete environment.Can offer a hotline for guidance and technical support.Experiment on issues such as questionnaire length, framing, order, scope, open vs closed questions etc.Can estimate respondent fatigue and adapt questionnaire to prevent satisficing.Avoid multiple answers when only one is required.Reduces information and questionnaire space with drop-down menus. before the end of the survey.
Piloting
Online focus groups and chat rooms are possible.Pre-test surveys with multiple browsers and screen settings.
Warm-up questions
Can screen respondent panel further with questions that aim to find out whether one is in the right frame of mind to answer,or record his sentiments.Keep respondents motivated with progress indicator features on the screen.
Information session
Better visualization (photos, videos), colours, sounds, fonts.The respondent can click on the information and have it read aloud and clearly.The information can expand with various pop-up tools (depending on how much information the person needs based on thelevel of information he already has at hand).The information can be heard from a voice instead of being given a separate instructions manual. Also the instructions can begiven step by step when needed.
Sampling frame
It is performed with web panels (mainly recruited from research companies etc).Personalized invitations and reminders.
Characteristics Face-to-face (personal)
Telephone Mail CAPI Web
Cost 3 2 1Response rate 1 2 3Design richness 2 2 1Representativeness 1 2 3Suitability for sensitive topics 3 2 1Prone to warm glow effect 3 2 1Self-selection 1 2 3Data handling possibilities (e.g. input, transformation etc)
3 2 1
Table 7. Web survey comparison to other modes of survey with respect to eight characteristics
Full Technology
Face to faceinterview
Telephoneinterview
CATI
CAPI
WEBMail survey
Zero technology
No humaninteraction
Full humaninteraction
Figure 9. Survey modes box
Acknowledgement
The authors acknowledge the insights from the participation to the WEBDATANET (COST Action IS1004) network (http://webdatanet.cbs.dk/).