Dear Mats,

I read the articles above, but did not find answer to your question, even the research books still did not give clear answer.

My opinion that there are 3 types of response rate:

1. ratio of the final number of complete participants in all rounds to the total number of approached participants (in your case 24/136)

2. ratio for the participants in the first phase to total of those who sent you back they accepted or denied participation (in your case 36/ you give the #)

3. ratio for those who completed all phases to those who started (24/36).

In addition to the above, the issue remains debateable and deserve hearing from scholars to reach consensus on this issue.

best luck 

Hej Mats, 

Forskarna till artiklarna är här på RG så du kan ju kontakta dem direkt. De borde veta:

Article Research guidelines for the Delphi Survey Technique

Article Research guidelines for the Delphi survey technique

Dear Mats,

I read the articles above, but did not find answer to your question, even the research books still did not give clear answer.

My opinion that there are 3 types of response rate:

1. ratio of the final number of complete participants in all rounds to the total number of approached participants (in your case 24/136)

2. ratio for the participants in the first phase to total of those who sent you back they accepted or denied participation (in your case 36/ you give the #)

3. ratio for those who completed all phases to those who started (24/36).

In addition to the above, the issue remains debateable and deserve hearing from scholars to reach consensus on this issue.

best luck 

I had a similar challenge-- lots of inquiries went out to obtain participants, but then some dropped off with each phase of the study.  I wrote this up just as it occurred (and probably a little lengthy, but it was clear).  This was not a published study, but an internal report for our agency. 

The initial recruitment email was distributed to (let's say for ease of example) 100 people. There were 25 potential respondents who expressed an interest in participating, yielding an overall response rate of 25%. One hundred percent of these 25 participants completed phase one and 24 of 25 participants (96%) completed phase two. However, only 21 of the 25 respondents (84% of the initial sample) were able to participate in all three rounds of the Delphi study.

According to Keeney et al 2011 The Deplhi-technique in Nursing and Helath research, the response rate is described as in its connection to every round. However, they do not mention how you take the initial numbers of asked participants into account. 

All your efforts and contributes have given me ways of thinking. I will continue with the thinking. I have read studie that don not report how many participants who were asked of participation. However, I think its intresting to report that so many as 96 out ow 132 did not respond or declined the invitation. The participants were ambulance care managers. We have earlier had problems including them in studies. Therefor it is interesting to highlight that issue as in the discussion.

Thanks for you efforts!

Hej Mats, 

Här är en annan artikel om Delfi-metoden. Den metoden har jag endast träffat på hos svenska forskare. När du berättar exakt hur du gjorde så säger du ju att du först frågade 136 via email, sedan att du gick vidare med 36 och slutligen landade du med 24. 

Conference Paper The Delphi Survey Method to achieve national consensus: The ...

I would argue that the response rate is 24/36. There is a preliminary stage of a Delphi study in which you approach participants with an invitation to participate. This (a) helps qualify them as 'expert' and (b) ascertains their interest in being part of the research. It also reduces participant mortality during the rounds of the study. The Delphi study only commences when the first round propositions are disseminated to those who have agreed to take part. Please see a recent paper on the use of Delphi method in public relations research.

Article Delphi 2.0: A reappraisal of Delphi method for public relati...

Dear Tom, 

he should nevertheless first tell that he approached 136 persons via email out of which 36 agreed to answer. After that out of 36 he ended with 24. In this kind of research you shall tell everything you did. 

I agree with Beatrice -- rather than stress about specific response rates, a more transparent and complete way of reporting is to present a flowchart  with all the numbers you just explained.  That way people can judge for themselves who fell out of your sample at what step.

One of the unaddressed issues is why this question (proportion of participants) is important.  If the issue is statistical reliability.....having 100/136 complete is not only preferable from a participation standpoint but statistically provides more stability in the statistical values generated.  If the question is about projectibility or representativeness...we need to know if there is anything different about the respondents who did not agree to participate in the survey vs. those who did.  If, in this case, only ambulance care managers from large organizations agreed to participate because they had the time to devote to a questionnaire where managers from smaller firms were too busy to participate...the results could be non-representative.

I agree with Sylvia and Beatrice that simple reporting is the best option. So, start with the overall response rate (36/136) and then the continuation rate (24/36).

FYI I think the original response rate is fairly reasonable for an email survey, especially something that requires as much effort to participate as a Delphi. In general, web-based approaches tend to produce low response rates.

Also, I have attached a review article on Delphi.

Even though this a bit late response, this is a very interesting discussion! I would prompt using a flowchart detailing the recruitment process and participation rates including number of experts initially approached. I would tend to consider the participation rate as the one between each rounds.