GuidedTrack How to reduce dropout in studies that collect data from each participant at multiple times seek well informed consent and commitment to the study prior to enrollment and randomization (1) GuidedTrack Docs

We recommend asking people to commit to completing your multiple-part survey prior to being randomized (if you are using *experiment or *randomize keywords in your survey/study). Prior to committing to your study, participants will need to understand it, so please refer to this section if you'd like some tips on how to check for their comprehension.

See this program (excerpts of which are provided below) for an example of how to check that participants understand the follow-up study and are willing to provide their email. For tips on sending emails to respondents, please see the section of the GT manual on emailing users.

Suppose that you're conducting a study investigating the effects on sleep quality following the implementation of certain sleep hygiene practices. You want to recruit people who will be likely to complete the intervention for the required number of days and who are unlikely to be lost to follow-up. There are several things you can do to increase the probability that the participants who are recruited to the follow-up stage will also complete the intervention as required. For example, you can:

  1. Explain what is involved in the follow-up study so that participants can make an informed decision. (This hopefully goes without saying.)
  2. Explain that their payment for the current study will not be affected by whether they opt into the follow-up study or not. (example)
  3. Ask participants to promise that they will comply with the experiment requirements, prior to starting the experiment. (example)
  4. Seek consent to email participants. (example)
*question: Are you interested in participating in the second part of the study?
	*tip: You'll still be paid for this survey you're currently completing regardless of your answer.
	Yes, I'd like to participate!
		*program: Sleep Study Rejection
	No, I'd like to stop here, actually
		>> rejectionReason = "Opted out"

Example of asking participants to promise to comply with study requirements🔗

--promise page
>> timesPromiseFailed = 0

*label: promisePage
*header: Your Compensation for Fully Completing all Parts of this Study

In addition to being paid {initialPay} for completing the first part of the study that you're doing right now, you will also be paid {payRate} upon completing the second part of this study.

*question: To make a promise to us that you will give an honest effort in doing the techniques each day for the next {studyLength} days, please type "I promise" in the box below. If you would prefer NOT to enroll in the {studyLength} day study, type "quit" instead.
	*tip: Regardless of your answer here, you will still be paid for this current HIT that you're doing right now.
	*type: text
	I promise
	Quit
		>> rejectionReason = "Opted out"
		>> candidateForPart2 = 0
	*other
		>> timesPromiseFailed = timesPromiseFailed + 1
		*if: timesPromiseFailed > 2
			>> rejectionReason = "Failed promise"
			*program: Sleep Study Rejection
			--They get 3 chances to type it correctly before getting excluded
		/Please be careful to make sure that there are no typos in your answer! ({3 - timesPromiseFailed} chance(s) remaining...)/
		*goto: promisePage

--email page
In order to provide you with the link to the daily check-in survey and reminders of how this study works over the next {studyLength} days, we ask that you provide us with a working email address.

We will not use this email address for any purpose other than communicating with you about this study, nor will we give it to any third party for any reason.

*question: Are you willing to provide an email address so that we can communicate with you about important details of this study?
	Yes, I'm willing
		*program: Get Email Details (Originally for Sleep Study v3)
		*if: doWeUseConsentForm = "yes"
			*program: Sleep Study Consent Form
	No, I'd rather not
		>> rejectionReason = "Uncooperative"
Next: