Human-subjects work is the area where students most often start too early and have to discard everything. The sequence is not negotiable.
Approval comes first. Your research plan, instrument, and consent forms are reviewed before you recruit a single participant. Data collected before approval generally cannot be used, and no appeal fixes it retroactively.
Consent and assent are different. For participants under 18, a parent or guardian gives informed consent and the participant gives assent. You need both, and both are documented.
"Just a survey" is still human-subjects research. So is a taste test, a reaction-time task, an interview, and observing people in a setting where they expect privacy.
Minimise what you collect. If you do not need names, birthdates, or contact details, do not collect them. Every identifier you gather is one you have to protect and justify.
Vulnerable populations raise the bar. Participants under 13, or in a clinical or institutional setting, or your own classmates over whom you have influence — all of these need extra care and usually extra review.
Your fair's rules and your SRC or IRB are the authority here. Nothing on this page is an approval, and no forum answer is either.