Randomized group allocation & Multiple avatars

[Patric]

Hello everyone,

I’m currently trying to prepare everything for my bachelor-thesis at the ZHAW. My supervisor is Sandra Hauser-Ulrich and we wondered IF and HOW the following scenarios are possible to implement in MobileCoach:

1. Our test subjects should be randomally assigned to three provided groups (control group / exp. group 1 / exp. group 2) as soon as they interact with the mobile coach intervention for the first time. Is this type of randomized group allocation possible?

2. While the control group won’t be equipped with an avatar, the exp. group 1 and 2 will be. I’d like to create and implement two different types of avatar figures for exp. group 1 and 2. Is it possible to use the same intervention (text interactions) but: a) No avatar in the experimental group, b) avatar figure X in exp. group 1 and c) avatar figure Y in exp. group 2?

I’m really grateful for any help provided by any of the members in this forum.

Thanks in advance!

Kind Regards
Patric

[Tobi]

Hi Patric,

Thanks a lot for your questions.

Ad 1: Yes, you can do so. For example, …

1. Add two variables $randomzationGroups and $group in the Variables tab as shown here. It is important that you use a comma separated value list for $randomzationGroups where each element of the list represents one of your conditions (as you see, this approach is scalable and works also for two conditions or more 3+ conditions )

2. Then, in the very first Micro Dialog, add a Decision Point and…

3. … add a rule that looks like this here:

4. Finally, after the new Decision Point is processed, you can use the $group variable to, for example, jump into different Micro Dialogs depending on your intervention design.

Try it out

Best,
Tobi

[Tobi]

Hi Patrick,

Regarding your 2nd question:

@Patric

While the control group won’t be equipped with an avatar, the exp. group 1 and 2 will be. I’d like to create and implement two different types of avatar figures for exp. group 1 and 2. Is it possible to use the same intervention (text interactions) but: a) No avatar in the experimental group, b) avatar figure X in exp. group 1 and c) avatar figure Y in exp. group 2?

This requires a ReactNative expert and adaptations in the MobileCoach ReactNative app. It is definitely doable with limited effort but does not work “out-of-the-box”.

I hope this helps,
Tobi

[hauu]

Hi Tobi
Many thanks for your timely response to our questions . It seems that we cannot test the use of different avatars the way we planned to. Have you got any suggestions how we could test 3 different avatars in MC? Or could we ask ZHAW / Petra to fix this ?
The current intervention Patrick is working on hast two different avatars (harry and sally as you know). Can we exchange these avatars by the ones Patrick will design or would you have to do this? And if so, could you do so?
I am looking forward to your reply!
Best, Sandi

[Tobi]

To do this, you need a ReactNative “expert” that understands the structure of the MobileCoach client app. In case you stick to 2 avatars, the only things to be done are (1) exchange the files for the avatars (you need to deliver the new files), (2) adding the new names in the code, (3) compiling the new app, and finally, (4) make the new app available as beta test version. I recommend you ask your technical MobileCoach experts on this at your institution first, if support is possible. Unfortunately, we at CDHI do not have any additional resources currently but are working on this issue.

Best, Tobi

[hauu]

Hi MobileCoach Team
We have managed to store different avatars with a random allocation. Is it possible, and if so how, to direct participants to different dialogs according to their avatar allocation? For example, if one gets allocated to avatar A, he/she should start with another dialog than the one who gets allocated to avatar B. Basically, we would like to use different dialogs for different avatars…Can this be done in USER INTENTION?
Many thanks and best, Sandi

[hauu]

Dear MobileCoach Team, Dear Tobi
One general question about avatars: We would like to know what avatar someone selected (or was allocated in case of random allocation). Is it possible to retrieve this information when extracting the data? Many thanks and best wishes, Sandi

[Anna]

Hello Mobile Coach Team and Tobi,

While I have successfully set up the randomisation for a group of 5 conditions, I ran it myself to test out for 50 times to see how equally distributed it could be, however I noticed it fell heavily into one condition (19) but not enough for the others - so is there any suggestions or ways to make sure the equal distribution to take place?
Ideally 30 in each condition. Many thanks and best wishes, Anna