Subscribe To Our Newsletter

Get tips and tools to tell your data story better.

No, thanks

 In Data Analysis Concepts Simplified, Experts, How To

Last week, I talked about how social sector organizations can use the “Play the Winner” alternative to randomized controlled trials to get around some of the ethical challenges RCTs present. This week, I want to talk about another good alternative to RCTs. Like “Play the Winner”, this method suits trials where the subjects are people, not things. So let’s jump right in. What are PRPTs and why should you care?

First things first. PRPT stands for Partially Randomized Preference Trials (also sometimes called Partially Randomized Patient Preference Trials). PRPTs include some of the best elements of RCTs, but also allow participants some say in what happens to them during the study.

It probably sounds confusing. Why would you let participants control your trial? How would that even work? As with RCTs or pretty much all of the alternatives, there are pros and cons.

 

Why Choose PRPTs?

If “Play the Winner” is the logical choice for sidestepping ethical concerns about conducting trials on people, PRPTs are the answer to practical issues. Namely, issues that arise because your trial participants are not as invested in your results as you are. They have free will and are going to do what they perceive to best for themselves — not for your data.

Often, participants in randomised trials will have a preference about which group in your trial they wind up in. (Obviously, this is most common with medical trials; patients have an idea of which treatment would help them more, etc., but this can happen with any trial where you are offering some program or treatment that would benefit some participants.)

This presents some problems.

  • Participants may refuse randomisation entirely. Their absence will affect your results.
  • Participants may not receive the benefit they prefer and not comply with the program. Their lack of compliance can affect your results.
  • Participants may receive their preferred benefit and comply better than the average. This can also affect your results.

So how do we get around these problems? Using PRPTs.

 

How Do PRPTs Work?

Partially randomized preference trials work by dividing participants into groups based on both their willingness to be randomized and their preferred program option. They are sorted into three groups:

  • No strong preferences and a willingness to be randomised
  • A preference for one program option and a willingness to be randomised
  • A preference for one program and unwillingness to be randomised

 

Brewin and Bradley preference design.

Groups are further broken down based on the number of program options. In a trial with two options, we end up with four groups. There are participants who:

  • Receive option A randomly
  • Choose option A
  • Receive option B randomly
  • Choose option B

This system is not without problems. When we analyze data that includes the non-randomised groups, we must treat the data as observational. We can’t make reliable comparisons between the randomised and non-randomised groups because there may be confounding variables present we can’t control for.

Using this design, we can combine some of the best elements of RCTs and feasibility studies (where participants have a say in which option they receive). PRPTs are not a perfect solution, but they can help nonprofits study the impact of their programs in certain circumstances.

 

Need Help Measuring Your Impact?

The team at Datassist is here to help nonprofits and social sector organizations measure the impact of the work they do. If you want help conducting an RCT, PRPT or any other trial to gauge the success of your programs, we can help. Drop us a line to discuss your needs.

Recommended Posts

Start typing and press Enter to search

Could the “Play the Winner” method solve some of the ethical problems with RCTs?