Running Your Thesis Questionnaire Online: From Design to SPSS Export

Most thesis questionnaires are run online today, and most of their problems start there too: duplicate responses, data that has to be coded by hand, respondents who were never part of the population, and a methods chapter that doesn't know how to describe the procedure. This guide walks the whole path step by step, from before you build to the SPSS file, with the exact settings to switch on in Porsino's online questionnaire builder. Most of the advice doesn't depend on the tool and works in any system.

Before you build: have four things ready

1. The final, approved questionnaire. Every change after publishing costs you; finalise the text with your supervisor first.
2. Sample size and attrition. The number you must reach and the share of responses you will probably exclude (how to work it out).
3. Sampling method and channels. Do you have a defined list (an organisation's staff, a faculty's students), or will you share a link in groups? This decision shapes your settings and your methods chapter.
4. Approvals. Ethics approval before data collection where your institution requires it (always in medical and sensitive human-subjects research); permission from the organisation if your population is its staff; and a licence for commercial standard instruments (which instruments need permission).

Step 1: Choose your starting point

• A standard questionnaire. If your instrument is one of the common ones, start from the standard questionnaires with automatic scoring; items, scale and reverse-scored items are ready. Just make sure the version matches the validation paper you cite in your methods chapter.
• A researcher-made questionnaire. Start a blank form and give each dimension a matrix question (six rows at most; split longer dimensions into two matrices). Ask demographics as single-choice ranges (age: under 30, 30–39, …), not exact numbers.

On phones the matrix automatically becomes one card per row, so respondents never have to scroll a wide grid sideways; that matters, because a large share of responses come from phones.

Step 2: Do the coding now, not after data collection

• Give every question a short Latin key (q1, q2, … or job1, job2, …). In the SPSS export the variable name is built from this key, and the question text becomes the variable label.
• Option codes: keep the scale running the same direction throughout (1 = strongly disagree to 5 = strongly agree). The default columns of a five-point matrix already carry codes 1 to 5; just relabel them "strongly disagree" to "strongly agree" and the codes stay.
• Ask negatively worded items on the same scale as the rest and reverse them in the analysis; flipping the scale mid-questionnaire confuses respondents.
• "No opinion": if it is the midpoint of the scale, it is one of the columns, coded 3. The separate "no opinion" option you can switch on in a matrix's settings is something else: an option outside the scale, exported to SPSS as code 99 and declared missing. Don't confuse the two.
• "Other" gets code 97 in SPSS, and the typed text goes into a separate variable with the suffix _other.

Step 3: The first page: informed consent in a few lines

Examiners read this page. A template where you only fill in the brackets:

Dear respondent, this questionnaire is part of a master's thesis in [field] at [university] on [topic]. It takes about [7] minutes. Participation is voluntary and you may stop at any time. Responses are anonymous and reported only in aggregate. [Ethics approval: …] Questions: [university email].

Below this text add a required confirmation field ("Having read the above, I agree to take part"). If you have an eligibility criterion (say, "upper-secondary teachers only"), ask a screening question here and use conditional logic to send anyone who answers "no" to the end of the form; that keeps out-of-population data from coming in at all.

Step 4: Settings that save your data

Setting	Why	Note
Anonymous mode	IP address and device details aren't stored, and the form says so	If you promised anonymity, turn it on and don't ask for names or phone numbers
One response per person (browser-based)	Stops casual repeats	Can be bypassed with a second phone or a private window
One response per code	For a defined population: student number or personnel code	Conflicts with full anonymity; choose deliberately
Required items	No missing data inside dimensions	Keep sensitive demographic questions optional
Time window and response cap	Collection stays within the period you report	Set the cap to your sample size plus attrition
Link password	When the questionnaire is for one class or organisation only	Share the password separately from the link

Step 5: Pilot with about 30 people

Publish the questionnaire and send it to about 30 people from the population who won't be in the main sample. Add an optional open question at the end: "Which question was unclear or repetitive?" Then check three things: completion time, alpha for each dimension (paste the export into the Cronbach's alpha calculator or take it straight into SPSS) and the answers to that open question.

After revising, make a copy of the form for the main run so pilot responses don't mix with your real data; or, if you keep the same form, export only from the date the main run started.

Step 6: Distribution: a short link, a QR code and a message people read

Every form has a permanent short link, and on the sharing page you can download its QR code for a classroom, a staff room or a poster. In messaging groups, a short, human message works better than "please fill in this questionnaire":

Hi, I'm [name], a master's student in [field] at [university]. For my thesis on [topic] I need the views of [upper-secondary teachers]. The questionnaire is anonymous and takes about 7 minutes: [link]. If you know anyone in [this group], please pass it on. Thank you.

• Only post the link in groups whose members belong to your population.
• One reminder after three or four days is enough; more than that annoys people.
• If you have a defined list and want random sampling, send one-time invitations instead of a public link; you see who hasn't responded and reminders go only to them.

Step 7: Monitor daily

• Responses against the target. If you aren't at half the target after a week, you need a new channel, not more patience.
• Drop-off point. The "drop-off point per question" report shows where respondents quit; if everyone quits at one matrix, shorten it.
• Suspicious responses. The "suspicious responses" report flags speeders and straight-lined matrices.
• Group balance. If you need quotas (say, equal numbers from each faculty), cap the options of that question so a full group closes.

Step 8: Export to Excel or SPSS

In the Responses tab, pick Excel, CSV or SPSS (.sav) from the Export menu. The SPSS file comes with:

• variable names from the question keys and each question's text as its label;
• a code and label for every option (value labels) and the measurement level;
• multiple-choice questions split into 0/1 variables, one per option;
• missing values: unanswered left empty, and the matrix's "no opinion" declared missing;
• dates as SPSS dates plus a Persian-calendar text twin.

If Persian labels look garbled in SPSS, switch to Unicode under Edit → Options → Language before opening the file. Continue the analysis from here with analyzing a Likert questionnaire in Excel and SPSS.

Methods-chapter text about the online procedure

Data were collected between [date] and [date] with an electronic questionnaire built on the Porsino platform. The link was distributed to [the population] through [channels], and informed consent was obtained before participants answered. To prevent duplicate responses, [each student number could submit only once]. Of the [412] responses received, [27] were excluded because [they were completed in under 90 seconds or gave identical answers to every item, reverse-worded ones included], and the analysis used [385] responses.

Mistakes that only show up at the end

• Changing item wording mid-collection; half the sample saw one instrument and half another.
• Pilot and main-run responses mixed in one file.
• "5 = strongly disagree" in one section and "5 = strongly agree" in another.
• Promising anonymity on page one and asking for a phone number on the last page.
• Sharing the link in public groups, then writing "random sampling" in the methods chapter.
• Not looking at the data until the last day; most problems show up in the first two.

Frequently asked questions

Will my university accept an online questionnaire?

In practice yes, if your methods chapter describes the procedure, the collection period, duplicate control and the rule for excluding responses. Agree it with your supervisor first, and if your faculty requires ethics approval, get that before you start.

Does building an online questionnaire cost anything?

Building and publishing in Porsino is free. The free account has a monthly response cap, shown on the pricing page; if your sample is larger, the extra responses aren't lost and unlock when you take a response licence.

Can I change a question after collection has started?

Technically yes: every response is stored with the form version it saw, so earlier data isn't damaged. Methodologically, apart from fixing a typo, don't change item wording mid-way.

How do I know who hasn't responded?

Only with personal invitations (a one-time link per person). With a public link and anonymous mode it isn't possible, and it shouldn't be.

Can I add paper responses to the same data?

Yes; enter the paper responses yourself through the same online form so you have one dataset, and state in your methods chapter how many responses came in by each route.

---

Ready? Build your thesis questionnaire online or start from one of the ready-made standard questionnaires; run the pilot this week and take the remaining steps from this list.