Concettina,

The literature is ambiguous about size needed for a Delphi study

sample . Representation is assessed by the quality of the expert

panel rather than its numbers. A consensus panel usually consists of 15 to 30 participants from the same discipline.

Literature supports that having more than 30 participants does not appear to improve the quality of results.

Validity of a Delphi study can be jeopardized if the number of participants decreases significantly between rounds. An ideal panel size needs to be large enough to account for possible attrition between different rounds.

Thanks for that Coleen Toronto . I started reading and found what you mentioned about ambiguity and numbers. Do you happen to have a reference for the numbers of 15 to 30 participants from the same discipline? Many TIA

Concettina,

I prepared a toolkit and gave a presentation on Delphi use for researchers for Internews about 8 years ago. I can send it privately to anyone ([email protected]) if interested, along with the ppt presentation if interested. I believe the 15–30 numbers comes from the RAND corporation, which invented the Delphi method back in the '50s. They have a paper on their website. https://www.rand.org/pubs/papers/P3558.html and some other materials.

It looks like I said 25–30 when I wrote my report, I had cites in it but they removed them when they reformatted the report for non academics (the report included time and cost estimates for paying for a study). I have the original Delphi papers somewhere but the 15–30 number sounds about right. Many studies go as high as 50, and I would argue the more the better with the Delphi.

let me know if you are interested.

Good Luck.

MLK

Concettina Trimboli here is the citation:

Keeney, S., Hasson, F. & McKenna, H. (2011). The Delphi technique in nursing and

health research. Oxford, United Kingdom: Wiley-Blackwell.

Brilliant, thanks Coleen Toronto

Thank you Michael L. Kent I have sent you an email

The number of expert with Delphi method depends on the qualifications of the experts, who can provide the information throughout the procedure.

Dea Dr. Coleen,

Can I ask you please which paper supports that having more than 30 participants does not appear to improve the quality of results?

Thank you so much for your valuable comments

Hi Concettina,

Hope you find this helpful.

DOI: 10.1186/1471-2288-5-37

All the best,

Ehsan Hamze

There is a good way of not loosing delphi participants between the rounds and it is to use a real-time delphi technique that is roundless. There is a new real-time delphi platform that could be useful in your future delphis.

https://4cf.pl/en/project/halnyx-delphi-platform/

Hi Riel, I don't believe there is a specific number. It all depends of what type of participants are you looking for and what you are going to do with the results. I have seen many studies with around 30 "experts", I personally usually run Delphi with between 50 to 100 but it takes some hard time to mobilize participants.

Of course there some general public national Delphi like the ones run in Japan with some thousand of participants (experts and non-experts).

There are important question is what kind of Delphi are you going to run (real-time , 2-3 rounds , etc). I have sometimes run an alternative version that I do find efficient that combine an one round Delphi, and then a second round that is run in a participatory workshop.

I hope that helped

Epaminondas Christofilopoulos thank you, that is really useful information. Just a couple of questions:

1) any tips with mobilizing participants ?

2) With the 'second round that is run in a participatory workshop', are you still trying to get consensus? if so, do you use some form of a voting system?

Many TIA,

Concettina

Concettina Trimboli

1) Personal touch - approach them one by one or give a small gift like an amazon voucher

2) For big groups there is another voting , for smaller groups, up to 10, reaching consensus - it is not that difficult

In some studies you should not expect serious results without actually remunerating the participants. Than you can expect serious committment. We recommend this to our clients and they increasingly decide to do so.

Of course there will be many studies where you won't need this, actually.

Thanks to everyone for suggestions - Very good ideas to potentially take forward!

I would also like to know. In the area of ​​transport for example, I saw many articles with around 50 to 100 participants. I used an article as the basis for my master's thesis.