Hello,
Based on other studies and after careful deliberation, I decided to use a specific semi-random sampling scheme in my study. Six times a day, participants will be prompted within a 30-minute time frame. Additionally, each participant gets 30 minutes to complete the questions. Yet, while programming the trigger logic, I came across the distribution function. Although I’ve read a lot on fixed, random, and semi-random sampling schemes, I haven’t read about the specific distribution of the prompts (uniform vs. normal) in the literature.
I can imagine several pros and cons for both uniform and normally distributed prompts, but I wondered if you could provide me with more information on these options. I would love to hear about (simulation) studies that used different forms or even compared the two.
Another small question:
I wondered if it is possible for participants to either choose not applicable OR fill in a slider as answer options to one single question.
Thank you very much!
Hi @h.j.m.nieterau
Welcome to our forum, and thanks for your detailed question.
The two distribution functions simply specify how Ethica should pick a time within the define time-window.
For example, assume you program your triggering logic to prompt a survey once a day, at a random time between 9 to 10 am. Now if you choose the “Uniform Distribution Function”, the chance of selecting every minute between 9 to 10 am will be equal (follows a uniform distribution), but if you choose “Normal”, the chance of choosing 9:30 will be more than choosing 9 or 10 (follows normal distribution).
Your second question was a bit unclear. You mean you want to have a Single Answer question, in which one option is “N/A”? Well, that can be done very easily.
Or you want to have a Single Answer Question where one option is a Slider Question? That, we don’t support and I cannot think of a use case for it!
Thanks,
Mohammad
Dear Mohammad,
Thank you for the quick response.
Also thanks for the clarification on the Distribution Function. Do you think a normal distribution would improve predictability for participants? And thus, maybe improve compliance rates but negatively affect ecological validity?
Here is an example for the second question: “In the last hour, thinking of my child made me happy.” Answer options on the slider: “not happy - extremely happy” However, if the participant did not think about its child, it’s hard to fill in. When there is an option N/A; “I didn’t think about my child”, then it would save us one extra question: (1)“In the last hour, I thought of my child” answer options “yes/no” and if they fill in yes: (2) “these thoughts were… not happy - extremely happy”.
Thank you.
Best,
Merlin
Well, I cannot give an educated opinion here. I have not run any analysis nor have seen any myself. Sorry.
Here is an example for the second question: “In the last hour, thinking of my child made me happy.” Answer options on the slider: “not happy - extremely happy” However, if the participant did not think about its child, it’s hard to fill in. When there is an option N/A; “I didn’t think about my child”, then it would save us one extra question: (1)“In the last hour, I thought of my child” answer options “yes/no” and if they fill in yes: (2) “these thoughts were… not happy - extremely happy”.
Unfortunately there is no other way than having two questions (as you said towards the end of your message). And I would argue this is a preferred method as well. This about it from two aspects:
- From the analysis, have two variables for “did_think_about_child” with “yes/no” options, and “happiness_from_thinking_about_child” with a number value is better than having just one variable containing both of these.
- From the user experience perspective, combining these two questions into one leads to a more complex user interface IMO.
Hope it helps,
Mohammad
Dear Mohammad,
Yes, it sure did. Thanks!
Merlin