[email from student]
I just compiled my team's list and these are the questions we came up with. If you need further clarification, please let me know.

1. Is flagging and removing data required to be manual for the admin, or could a predetermined tolerance level be used to filter the data automatically?


Our idea was during the creation of a survey, the admin could specify the minimum and maximum answer time per question (Ex: Min: 1 second, Max: 30 seconds), or possibly
a percentage of accepted answers (Ex: 95% = within 2σ from μ). Additionally, they could specify the accepted variance for each question.



2. Could we allow users to answer past surveys?


Our team came up with the idea for a daily survey offered to the User. Each day, the user may fill out the survey and after completing the latest one, they could revisit
past surveys in case they missed one. During the creation of a survey, the admin could specify the time range of accepting answers, as well as the time range of allowing the survey to be taken.



3. Could we allow all users to answer the survey?


We thought it might be valuable to allow users of any age to take the survey, but to only store the answers of people above a certain age. If the purpose is to get data
from people above the age of 18, we can allow people under that age to take the survey without faking their age to do so.



4. Does a user have to answer every question?


We came to the conclusion that it might be better to allow an uncertain user to withdraw from answering a certain question before confirming their final submission (Ex:
User completes 19/20, survey prompts: “You completed 19/20 questions. Are you sure you would like to submit?”). However, as you mentioned, questions with even proportions could indicate high variance, thus showcasing people’s uncertainty. Therefore, requiring
every question may prove useful as well.



5. Could we decide on a custom image selection method?


In your information, you state we must use a radio button to indicate image selection. If we find that another means of selection is more visually appealing, could we instead
use that?



6. Could we train the image generation algorithm using the answer data?


I understand you want an algorithm that can determine which image is more aesthetically pleasing, but would it be worthwhile to also train the image generation algorithm
off of the user input to produce more aesthetically pleasing images? We could use the parameters of an image produced by the image generation algorithm to train it rather than the image itself.



7. What are your overall goals for this project and how do the prior questions fit into those goals?