The instrumentation section of a methods chapter is usually short. The appendix behind it is not, and that is where most theses lose marks. Examiners turn to the appendix to see the actual instrument, and they find either a validated scale reproduced without permission, or a page of questions with no response format, no prompts and no recording structure. This guide gives you the three data collection instruments a thesis actually needs in blank, copy-able form: a questionnaire shell, a semi-structured interview guide, and an observation schedule. Each comes with worked example rows, the wording to put above it in the appendix, and the sentences the methods chapter has to carry about it.
This is a template pack, not a design course. If you are still deciding on question wording, Likert points, sampling and pilot size, work through our guide to designing a survey and questionnaire for academic research first, then come back here for the document you will hand in. If a validated scale already exists for your construct, use it: our catalogue of validated instruments for an education dissertation and the parallel instruments guide for social work dissertations show what is available and who owns it. Writing your own instrument is the fallback, not the default.
A data collection instrument is the physical tool you use to record data: a questionnaire, an interview guide, an observation schedule, a test, or a document extraction form. A thesis appendix should contain the complete instrument exactly as participants or observers saw it, and the methods chapter should state, for each instrument, what it measures, how many items it has, its response format, where it came from, how it was piloted, and what reliability or agreement evidence you have for it.
Which Instrument Your Design Needs
The instrument follows the question, not the other way round. Creswell and Creswell, in Research Design (6th edition, SAGE), group data collection for a thesis into four families: questionnaires and scales, interviews, observations, and documents or artefacts. Mixed-methods designs use two or more and must supply a separate instrument for each.
| If your question asks | Instrument | Appendix contains |
|---|---|---|
| How much, how many, how strongly related | Questionnaire or validated scale | Full item list with response options, or a sample item plus citation if licensed |
| How people understand, experience or account for something | Semi-structured interview guide | Opening script, main questions, probes, closing |
| What actually happens in a setting | Observation schedule | Coding categories, definitions, recording grid, field-note conventions |
| What records or policies show | Document extraction form | Field list, inclusion rules, source register |
Before you write a single item, make sure each variable is already defined at the level of an indicator. If it is not, the instrument will drift. Our step-by-step on operationalization of variables turns a construct into indicators you can write items for, and research variables explained with thesis examples covers which variables need measuring at all.
Template 1: Questionnaire Shell
Copy this structure and replace the bracketed text. The section headings are the ones examiners look for; keep them.
Appendix A. [Study title] Participant Questionnaire
Section 0. Information and consent. [One paragraph: purpose, who may take part, voluntary, anonymous or confidential, estimated minutes, ethics approval reference, researcher contact.] Tick to confirm you have read the above and agree to take part. ☐
Section 1. Screening. [Eligibility item, e.g. “Are you currently enrolled on a [programme] at [institution type]?” Yes / No. Route “No” to the end.]
Section 2. [Construct 1 name]. Instruction: “For each statement, choose the option that best describes you over the past [time window].” Response format: 1 Strongly disagree, 2 Disagree, 3 Neither agree nor disagree, 4 Agree, 5 Strongly agree.
2.1 [Item] · 2.2 [Item] · 2.3 [Item, reverse-scored] · 2.4 [Item] · 2.5 [Item]
Section 3. [Construct 2 name]. [Same pattern. Keep one response format across the whole instrument.]
Section 4. Behaviour or outcome items. [Frequency or count items with a defined window, e.g. “In the past 30 days, how many times did you…” with non-overlapping bands.]
Section 5. Open comment (optional). “Is there anything else about [topic] you would like to add?”
Section 6. About you. [Demographics last: age band, gender with self-describe and prefer-not-to-say options, year of study, discipline, any variable you will use as a covariate and nothing more.]
Worked example rows for a Section 2 on academic self-efficacy, five-point agreement scale, item 2.3 reverse-scored:
| Item | Wording | Indicator it serves |
|---|---|---|
| 2.1 | I can plan a long piece of academic writing without help. | Planning confidence |
| 2.2 | When a draft goes badly, I keep working on it. | Persistence |
| 2.3 | I avoid starting written assignments until the deadline is close. (reverse) | Avoidance, reversed |
| 2.4 | I can find the sources I need for an assignment on my own. | Source-search confidence |
Three rules make the difference between a shell and a usable instrument. Every item must map to exactly one indicator, or you cannot interpret the subscale. Every frequency item must name a time window, or respondents each invent their own. And one response format runs the whole way through, because mixing a five-point and a seven-point scale in one questionnaire makes the scores non-comparable.
Build Your Methods Chapter Around Your Instrument
Tesify drafts the instrumentation section from the instrument you actually plan to use, keeps every source attached to the sentence that cites it, and leaves 100% of the words yours. More than 9,000 students have written their chapters with it.
Template 2: Semi-Structured Interview Guide
A semi-structured guide is not a list of questions. It is a script with three layers: what you say, what you ask, and what you say next if the answer is thin. The third layer is the one students omit and the one that decides whether the transcripts are analysable.
Appendix B. Semi-Structured Interview Guide
Before recording. [Confirm consent form signed; confirm permission to record; state that participation can stop at any time and that quotations will be pseudonymised; ask if there are questions.]
Opening (2 to 3 minutes). “To start, could you tell me a little about your role and how long you have been doing it?” Purpose: settle the participant, establish context, produce a speech sample.
Q1 [Topic area 1]. “Can you walk me through what happens when [process]?”
Probes: “What happens just before that?” · “Who else is involved at that point?” · “Can you give me a recent example?”
Q2 [Topic area 2]. “You mentioned [X]. How do you decide [decision]?”
Probes: “What makes that easier or harder?” · “Has that changed?” · “Tell me more about that.”
Q3 [Topic area 3, the sensitive one, placed late]. “Some people have said [neutral framing of a difficulty]. How does that compare with your experience?”
Probes: “What would need to be different?” · silence, four seconds.
Q4 [Contrast or negative case]. “Has there been a time when it worked differently?”
Closing. “Is there anything important about [topic] that I have not asked about?” [Thank; explain what happens next; offer to share a summary.]
After recording. [Write a contact summary within 24 hours: setting, non-verbal notes, anything that shapes interpretation, and your own reactions.]
Four conventions keep a guide defensible. Keep main questions to between four and eight, because a 45-minute interview cannot carry more. Write every main question as an open invitation to describe, never as a yes-or-no. Order topics from easy to sensitive, so that trust exists before the hardest question arrives. And attach probes to each question in writing, so that a second interviewer, or you on a tired day, prompts consistently.
Note what the guide is not. It is not a schedule you read verbatim; the point of semi-structured interviewing is that you follow what the participant raises. Say exactly that in the methods chapter, and say that the guide was used flexibly with the order varied to follow the participant.
Template 3: Observation Schedule
An observation schedule is the instrument students most often improvise and most often get sent back on, because an observation with no predefined categories produces field notes rather than data. A schedule has four parts: a header, defined categories, a recording grid, and a field-note column for what the categories cannot hold.
Appendix C. Observation Schedule
Header. Observer ID · Date · Setting · Session start and end time · Number of people present · Observer role (non-participant / participant) · Consent status of those present.
Recording method. [Choose and state one: time sampling, e.g. code the dominant category in each two-minute interval; or event sampling, e.g. record every occurrence of the target behaviour with its time.]
Category definitions. [One line per code, each stating what counts and what does not. A category that cannot be explained to a second observer in one sentence is not yet a category.]
Worked example for a classroom study using two-minute time sampling, showing the definitions and the grid together:
| Code | Counts as | Does not count as |
|---|---|---|
| T-EXP | Teacher explaining to the whole class | Teacher giving procedural instructions |
| T-QST | Teacher asking a question that invites reasoning | Checking recall of a fact |
| S-GRP | Students working in groups on the task | Students seated in groups but working alone |
| S-IND | Students working individually on the task | Waiting for materials |
| TRN | Transition: moving, distributing, settling | Any on-task activity |
The grid itself is deliberately dull, and that is the point:
| Interval | Clock | Code | Field note (optional) |
|---|---|---|---|
| 1 | 09:00–09:02 | TRN | Late arrivals still entering |
| 2 | 09:02–09:04 | T-EXP | |
| 3 | 09:04–09:06 | T-QST | Three students answer unprompted |
| 4 | 09:06–09:08 | S-GRP |
Two requirements turn this from a worksheet into an instrument. The categories must be mutually exclusive and jointly exhaustive within an interval, so that every interval receives exactly one code and no interval is left blank. And a second observer must double-code a share of the sessions, conventionally around twenty per cent, so that you can report an agreement statistic. Percentage agreement alone is not enough for a committee that knows the field; report Cohen’s kappa for two observers, or an intraclass correlation for continuous ratings.
Sampling decisions apply to observations exactly as they do to people: how many sessions, in which settings, at which times of day. Our guide to defining population and sample with thesis examples covers the reasoning, and it applies to observation occasions as much as to participants.
What the Methods Chapter Must Say About Each Instrument
Give the committee a table before you give them prose. One row per instrument, with these columns, answers most of what a methodologist would otherwise write in the margin.
| Column | What goes in it |
|---|---|
| Construct or focus | The variable or phenomenon this instrument captures |
| Source | Published scale with citation, adapted from a named source, or researcher-developed |
| Structure | Number of items, subscales, or codes and recording method |
| Response format | Scale points and anchors, or interval length |
| Piloting | Who, how many, what changed as a result |
| Quality evidence | Reliability in your own sample per subscale, or inter-rater agreement |
| Permission | Free for research with citation, written permission obtained, or licensed |
| Appendix | Which appendix holds the full instrument, or a sample item if reproduction is not permitted |
Follow the table with one short paragraph per instrument explaining why this instrument and not its nearest rival. That justification is what examiners look for and what students most often leave out.
Five Mistakes Examiners Flag
- A licensed scale reproduced in full in the appendix. Commercially licensed instruments are purchased per administration and may not be reprinted. Include a sample item and a citation to where the instrument can be obtained, and say in the methods chapter that permission terms prevent reproduction.
- An interview guide with no probes. Without written probes, prompting varies between interviews and you cannot claim the guide was applied consistently. Probes take one line each and they belong in the appendix.
- Observation categories that overlap. If a moment could reasonably be coded two ways, two observers will code it two ways, and the agreement statistic will collapse. Write the “does not count as” column before you go into the field, not after.
- Pilot testing reported as a fact rather than a change. “The instrument was piloted with five participants” says nothing. State what the pilot revealed and what you altered because of it, or the pilot reads as decoration.
- Reporting only the published reliability. A published alpha describes the developer’s sample, not yours. Report internal consistency per subscale in your own data, and discuss any subscale that falls short rather than quietly dropping it.
Frequently Asked Questions
What is the difference between a research instrument and a research method?
The method is the overall approach, such as a survey, a case study or an ethnography. The instrument is the physical tool that records the data within that method: the questionnaire, the interview guide, the observation schedule. A methods chapter describes both, and the appendix contains the instrument.
Do I have to put the full instrument in the appendix?
Yes for anything you wrote or adapted yourself, and for scales the developer has released for free research use. No for licensed commercial instruments, where reproduction breaches the licence. In that case include a sample item, the number of items and subscales, and the citation to the manual.
How many questions should a semi-structured interview guide have?
Between four and eight main questions for an interview of 45 to 60 minutes, each with two or three written probes. More main questions than that produces rushed answers, because participants elaborate and the elaboration is the data.
What is the difference between time sampling and event sampling in an observation schedule?
Time sampling divides the session into fixed intervals and records what is happening in each one, which suits continuous states such as whether students are working in groups. Event sampling records every occurrence of a defined behaviour with its time, which suits discrete events such as a particular type of question. Choose one, state it in the schedule header, and do not mix them within a session.
Can I write my own scale instead of using a validated one?
You can, but the burden of evidence moves to you: you must show content validity through expert review, pilot the items, and report the factor structure and reliability from your own data. Where a validated scale exists for your construct, adapting it is almost always the stronger choice, and it makes your results comparable to prior studies.
How many observations need to be double-coded?
Around twenty per cent of sessions is the common convention, coded independently by a second trained observer. Report the number double-coded, the training the observers received, and an agreement statistic such as Cohen’s kappa rather than percentage agreement alone.
Next Step
Pick the template that matches your design, fill the brackets, and then read it as though you were the second observer or a participant seeing it cold. Anything you would have to explain out loud is something the instrument is missing. When the instrument is settled, write the instrumentation table first and the prose second. If you want a workspace that keeps the instrument, the variables and the methods chapter in view together, start your thesis in Tesify.
Write your thesis with AI
Structure, draft, cite, and format your thesis faster with Tesify’s AI writing tools, automatic bibliography, and plagiarism checker. Free to start, no credit card required.
Leave a Reply