A questionnaire is a measuring instrument, and it can be built badly. Most of the damage is structural: no consent statement, demographics asked before eligibility, questions that answer no objective, and open questions so demanding that respondents abandon the form.
Enter your topic and this tool lays out the full structure — consent, screening, background, main items, open response and closing — with guidance at each step. The bracketed parts are yours to write from your own objectives.
Consent comes first. Departments increasingly require it and data collected without it can be rejected outright. Screening follows, so ineligible respondents leave before investing effort — nothing inflates dropout like discovering at question 30 that someone should never have begun.
Background questions come next, then main items grouped by objective, then open questions. Open questions sit last because they are the most demanding and most often skipped; losing them at the end costs less than losing the entire form midway.
Map each question to an objective before distributing. A question mapping to nothing produces data you will not analyse and measurably lowers completion. Demographics are the worst offenders: age, gender, education, caste and income get collected by reflex and then never appear in the analysis.
Run the check in reverse too. Take each objective and ask which questions answer it. An objective with no questions behind it cannot be met, and finding that now is far cheaper than finding it after fieldwork in a district you cannot easily revisit.
Closed factual questions give clean data. Multiple choice needs options that are exhaustive and mutually exclusive — add "Other, please specify" wherever you cannot be certain you have covered every case. Likert blocks measure attitudes and should share one scale, stated once at the top rather than repeated on every row.
Ranking questions give rich data but demand effort; keep the list to five or six. Open questions capture what you failed to anticipate, which is their point, but two is the practical limit for a student survey.
If your respondents are more comfortable in Nepali, administer the questionnaire in Nepali. Translate, then back-translate to English to confirm the meaning held, and describe that process in your methodology — it is direct evidence of instrument validity. A question that shifts meaning in translation quietly changes what you measured.
Pilot with five to ten people from your target population, time it, and ask them to say what each question means to them. Any hesitation, re-reading or divergent interpretation is a defect you can still fix for nothing. Report that you piloted and what you changed.
Your consent statement should say who you are, what the research is for, how long it takes, that participation is voluntary, that responses are anonymous or confidential, and how data will be stored. Give a contact for questions.
Anonymous and confidential differ: anonymous means you cannot identify respondents at all; confidential means you can but will not disclose. Do not promise anonymity while collecting phone numbers. Where you are surveying inside a school, hospital or office, obtain written permission from the institution as well as consent from individuals — many Nepali departments require that letter in your appendix.
Under ten minutes for most student surveys. Completion falls away quickly beyond that, especially online.
After screening and before the main items — and only the ones you will actually analyse.
Whichever your respondents are more comfortable with. If Nepali, translate and back-translate, and document the process in your methodology.
If you are collecting data inside an institution — a campus, hospital, office or school — yes, usually alongside individual consent. Most departments expect it in the appendix.
Yes. Five to ten respondents from your target population will expose ambiguous wording and give a realistic completion time, both of which belong in your methodology.