“Student engagement” is one of the most frequently measured constructs in education theses, and one of the most loosely defined. A supervisor who opens a draft proposal on the topic will usually find the word used for at least three different things: time on task, a feeling of belonging, and a score from a national survey. The data and the instrument a student chooses decide which of those the thesis actually studies. This guide is for the supervisor, programme director or graduate school that has to steer a cohort of engagement theses: which survey sources exist, what each one measures, what each one cannot tell a thesis, and how to set a faculty standard so that proposals cite them correctly.
Start with the construct, not the survey
A frequently referenced conceptual review of the topic is Fredricks, Blumenfeld and Paris, “School Engagement: Potential of the Concept, State of the Evidence”, Review of Educational Research, 74(1), 59–109 (2004). It treats engagement as a multidimensional idea with behavioural, emotional and cognitive strands, and it is aimed at school-age learners, so a higher education thesis that borrows it should say what it is adapting. The first supervision question is therefore simple: which strand is the student measuring, and in which population?
A short test a supervisor can apply at proposal stage:
- Does the research question use the word “engagement” for something the chosen instrument actually asks about?
- Is the population the same as the one the instrument was built for (for example, first-year and senior undergraduates rather than a taught master’s cohort)?
- Is the thesis analysing the student’s own data, or quoting published institutional results?
The answers decide whether the student needs their own questionnaire, a licence to use an existing one, or only the published national tables.
NSSE: the instrument the UK survey was derived from
The National Survey of Student Engagement (NSSE) is run from the Center for Postsecondary Research in the Indiana University School of Education. According to its own site, it was conceived in 1998, piloted in 1999 with funding from The Pew Charitable Trusts, and first administered fully in 2000. The instrument, The College Student Report, surveys first-year and senior students at bachelor’s degree-granting institutions. NSSE reports that nearly 1,700 four-year colleges and universities have taken part, and that 252,336 students responded in 2022.
What matters for a thesis is how the survey is organised. NSSE reports ten Engagement Indicators grouped into four themes:
| Theme | Engagement Indicators |
|---|---|
| Academic Challenge | Higher-Order Learning; Reflective & Integrative Learning; Learning Strategies; Quantitative Reasoning |
| Learning with Peers | Collaborative Learning; Discussions with Diverse Others |
| Experiences with Faculty | Student-Faculty Interaction; Effective Teaching Practices |
| Campus Environment | Quality of Interactions; Supportive Environment |
NSSE also reports high-impact practices separately. For psychometric evidence it points to its Psychometric Portfolio, which reports Cronbach’s alpha for each indicator, and it states that construct validity for the indicators was supported by exploratory and confirmatory factor analysis. A student who cites NSSE scores should be asked to cite that evidence, not simply say the survey is “validated”.
Two limits belong in every supervision note. First, the population is undergraduate first-year and senior students, so NSSE results say little about a postgraduate or part-time cohort. Second, NSSE measures the experiences and practices students report, not learning outcomes, so a thesis claiming that higher engagement scores “cause” better grades needs its own design and its own outcome data.
UKES: the UK adaptation and what its documentation shows
The UK Engagement Survey (UKES) was developed by the Higher Education Academy, now part of Advance HE. The report on its second trial year, written by Dr Alex Buckley, states that UKES was derived mainly from NSSE: the 2014 questionnaire had 50 items, 39 drawn from NSSE and 11 unique to UKES. In that year 32 institutions took part and 25,500 responses were collected, up from nine institutions and 8,500 responses in the first trial year in 2013. The report also states that UKES was due to run as a full survey in 2015. A supervisor should check Advance HE’s current pages before telling a student that UKES data is available for a given year, because participation and publication have changed over time.
The same report is useful as a methods exemplar. It groups items into scales such as higher-order learning, collaborative learning, engagement with research and formulating and exploring questions, and it reports Cronbach’s alpha for each, including a scale at 0.80 and another at 0.77, and it notes one case where removing an item would raise the alpha. That is the style of reporting a supervisor can ask students to reproduce in their own methods chapter. It also shows why published scale reliabilities should not be copied across: reliability belongs to a sample, and a thesis with its own sample must compute its own.
The wider point for governance is that engagement surveys differ by country, by year and by version. A faculty that expects theses to compare “NSSE results” with “UKES results” should require the student to state the version, the year, the population and the item wording for each.
What published survey data can and cannot support
Published engagement data is institution-level or sector-level. It is excellent for context chapters and for justifying a research gap. It is a poor substitute for primary data in a thesis whose question is about a specific programme, because the published scores average over students the thesis does not study. A supervisor can use this decision table:
| Thesis question | Published survey tables | Own questionnaire or secondary microdata |
|---|---|---|
| How does engagement in our sector compare over time? | Suitable, with the version and year stated | Not needed |
| Does a new teaching method change engagement in one course? | Context only | Required, with a defined comparison |
| Which student groups report lower engagement? | Only if the survey publishes the breakdown | Usually required |
| Is engagement associated with progression? | Not suitable | Requires linked institutional records and ethics approval |
The last row is where proposals most often go wrong. Linking survey responses to academic records raises consent and data-handling questions that need ethics review before any data is collected, and the faculty should say so in its proposal checklist rather than leave students to discover it.
A faculty standard for engagement theses
A one-page standard keeps supervisors consistent across a cohort. A workable version asks every engagement proposal to state, in this order:
- The construct strand. Behavioural, emotional, cognitive or a named composite, with the source of the definition.
- The instrument. Name, authors, year and version, plus where the licence or permission to use it comes from.
- The population match. Why the instrument suits the students being studied, and what was changed if it was adapted.
- The reliability plan. Which reliability coefficient will be reported on the thesis’s own sample, and the threshold the student will use to judge it.
- The data route. Primary collection, institutional data or published tables, with the approval each needs.
This mirrors the discipline already recommended for other validated instruments in this series. A faculty weighing which scale sources to endorse can compare the approach in the piece on governing a validated teacher self-efficacy instrument bank, and the same permission questions arise in the guide to a service-quality instrument bank for a hospitality faculty.
Common supervision problems and the fix for each
Engagement theses tend to fail in predictable ways. The following list can serve as a review checklist for a methods committee.
- The word is undefined. The fix is to require one sentence in the introduction naming the strand and the source of the definition.
- A survey is adapted without disclosure. Reworded items, dropped items or a translated version change the instrument. The student should list every change and not describe the result as the original scale.
- Reliability is quoted from the manual. Published alphas describe someone else’s sample. The thesis should report its own.
- Self-report is treated as behaviour. Engagement surveys record what students say they do. Claims about actual time use or attendance need other data.
- Causal language without design. Cross-sectional survey data supports association. A supervisor should flag “leads to” and “improves” in findings sections.
- Cohort mismatch. A distance-learning or part-time cohort rarely matches an instrument built for full-time undergraduates.
Alignment checks help here. The method used to test whether a proposal’s hypotheses and variables line up is described in the guide to an alignment and operationalization review for education proposals, and engagement is a good test case because its variables are so easily mislabelled.
Using your own institution’s data responsibly
Many students will ask for the institution’s own survey results. A graduate school should decide, in advance, which results can be shared with students, in what form, and who approves the request. Aggregated tables at programme level are usually the lowest-risk product. Record-level data needs a documented purpose, a data owner and an approval route. Writing this down once saves each supervisor from negotiating it case by case, and it keeps small cell sizes from identifying students in a thesis appendix.
The same discipline applies to how an institution cites its own numbers. The site’s guide to writing centre utilisation data shows how an undefined denominator makes a headline figure unusable; the same is true of an engagement score with no stated response rate.
Ten researchable engagement questions for a cohort list
A topic list helps students avoid copying one another. These are illustrative, and each needs the instrument and population checks above before approval.
- How do reported collaborative learning practices differ between first-year and final-year students on one programme?
- Do students who report more student-faculty interaction also report a more supportive environment?
- How does engagement with research differ between taught and research-led modules?
- What does a translated engagement item set look like after cognitive testing with international students?
- How do part-time and full-time students interpret the same engagement items?
- Which engagement items show floor or ceiling effects in a single department’s sample?
- How does the timing of survey administration alter reported engagement?
- What is the internal consistency of a short engagement scale in a postgraduate sample?
- How do students describe engagement in interviews compared with their survey answers?
- How do published sector results compare with one institution’s results, given differences in sample and year?
Where Tesify fits
Choosing and defending the instrument is the faculty’s call, and so is the data route. Once a student has an approved design and is writing the thesis, Tesify helps students structure and organise their thesis while they write 100% of the work themselves: more than 9,000 students have written over 15,000 chapters with Tesify. See how Tesify supports a thesis cohort.
Frequently asked questions
What does NSSE measure?
NSSE collects information about first-year and senior students’ participation in programmes and activities their institution provides for learning. It reports ten Engagement Indicators across four themes, and it reports high-impact practices separately.
Can a thesis student use NSSE data directly?
A student can cite published NSSE reports for context. Using the instrument itself or institutional response data needs the relevant permission and, for record-level data, an approval route set by the institution.
How is UKES related to NSSE?
According to the report on UKES’s 2014 trial year, most of the UK survey derives from NSSE. The 50-item questionnaire in that year had 39 items drawn from NSSE and 11 unique to UKES.
Is UKES still running?
The 2014 trial report said it would run as a full survey in 2015. Participation and publication change over time, so check Advance HE’s current pages before telling a student that data exists for a particular year.
Which engagement strand should a thesis choose?
The choice follows the research question. Fredricks, Blumenfeld and Paris (2004) describe behavioural, emotional and cognitive strands; a thesis should name the one it measures and cite its definition.
Should students quote published reliability figures?
No. Reliability describes a particular sample. A thesis can cite the published figures as background but should report its own coefficient from its own data.
Do engagement surveys measure learning?
No. They record the experiences and practices that students report. A claim about learning outcomes needs outcome data and a design that supports it.
Can survey responses be linked to student records?
Only with an approved purpose, a data owner and, in most institutions, ethics approval. A faculty checklist should state this at proposal stage.
What should a faculty standard for engagement theses include?
The construct strand and its source, the named instrument and version with its permission route, the population match, the reliability plan and the data route with its approvals.
How many engagement theses should one supervisor handle?
There is no sector rule. Because instrument checks are specific to each proposal, a faculty should size supervisor loads from its own cohort data, such as the figures in its postgraduate enrolment by field.