Together with some colleagues, I am designing a DCE in the area of cardiology. It concerns patient preference for different types of wearable technology for supported self-monitoring of arrhythmia in adults with congenital heart disease. I haven't used Ngene in 18 months or so, so am a little rusty, so thought I'd post the design here in case anyone has any comments or suggestions.
A bit about the DCE. Based on a review of literature, we have six attributes: four with four levels and two with three levels:
Type of wearable
Accuracy
Charging
Data review
Cost
Support
In our current design, each respondent will complete ten choices tasks, with the design blocked into six blocks. There is an opt out option to account for the fact that some participants may not want to consider a wearable. All attributes dummy coded and optimised for MNL. Target sample size is 300, which we are confident of achieving. Please find below our current syntax:
My two queries were:Design
;alts = ChoiceA, ChoiceB, OptOut
;rows = 60
;block = 6
;eff = (mnl, d)
;con
;model:
U(ChoiceA) =
b1_WearableType.dummy[0|0|0] * WearableType[1,2,3,0] ? 0 = Wrist, 1 = Ring, 2 = Chest Strap, 3= Built-in to clothing
b2_Accuracy.dummy[0.0001|0.0002] * Accuracy[1,2,0] ? 0 = Low (<85%), 1 = Moderate (85-94%), 2 = High (95%+)
b3_Charging.dummy[0.0001|0.0002|0.0003] * Charging[1,2,3,0] ? 0 = Daily Charging, 1 = Charge Device Every 3 Days, 2 = Charge Device Every 7 Days, 3 = Self-Charging
b4_DataReview.dummy[0.0001|0.0002|0.0003] * DataReview[1,2,3,0] ? 0 = Alert only, monthly data review, 1 = Alert and explanation, weekly data review, 2 = Alert with summary, daily data review, 3 = Personalised insights
b5_Cost.dummy[0.0001|0.0002|0.0003] * Cost[1,2,3,0] ? 0 = £50, 1 = £100, 2 = £250, 3 = £500
b6_Support.dummy[0|0] * Support[1,2,0] ? 0 = Start Up Manual Only, 1 = 24 Hour Online Chatbot Support, 2 = Human Support During Business Hours
/
U(ChoiceB) =
b1_WearableType.dummy * WearableType
b2_Accuracy.dummy * Accuracy
b3_Charging.dummy * Charging
b4_DataReview.dummy * DataReview
b5_Cost.dummy * Cost
b6_Support.dummy * Support
/
U(OptOut) = b0
$
Have we specified the opt out correctly? I think we have in this instance by including the ASC
Does the blocking look okay given the target sample size? 50 respondents per block/choice task is comfortably above the rule of thumb estimates from the literature, but would be interested in anyone had any thoughts regarding this.
Thanks,
Tom