Skip to content
PaperFren

Research method

Self-Report Survey

51 studies

Ask PaperFrenWatch this

A self-report survey asks students or instructors to rate their own experiences, attitudes, practices or satisfaction, usually on closed items. The product is a set of counts, scale scores and sometimes profiles derived from those answers. It is not a trace of behaviour, not an interview, and not an experiment: people tell you what they notice about themselves, at one (or a few) moments, under whatever selection and social-desirability conditions the form creates.

Education researchers reach for surveys when they need comparable ratings from many people — emergency-remote challenges, lecture-mode choice, student voice, online self-efficacy, instructor assessment habits. They answer 'how did this sample describe itself?' Their main limitation is that a high rating is not observed engagement, not a grade, and not proof that changing the rated factor would change learning.

Studies

51

Findings

6

6 supporting · 0 challenging · 0 qualifying citations

Open tensions

2

Currently

What we know

  1. Among 138 university students in Finland and India, 132 faced multiple emergency-remote challenges (about four on average); 52% cited schedule management and home distractions. Engagement was moderate: students gradually adjusted but did not fully accept emergency remote learning as a long-term normal. That is a catalogue of reported difficulties, not a test of an instructional intervention.
  2. A technical university distributed 43,979 survey forms and analysed 17,641 responses across 381 courses (13,585 complete and linked to administrative grades). Self-chosen lecture mode — in person, live online, or recorded — had virtually no direct association with exam grades except among students without a consistent strategy. Split on-site/online attendance scored slightly worse; 70.8% of students who watched recordings for at least half of lectures would recommend that approach. Students selected their own mode, so the survey cannot causalise attendance.
  3. In two U.S. urban high schools already known for student-voice work, a survey of 1,152 students sat alongside interviews, focus groups and observations. Teachers known for voice more often sought feedback and input than true collaborative decision-making on curriculum and rules. The survey describes pedagogical foundations in those schools; it does not prove achievement gains from student-voice practices.
  4. Of 2,522 UOC communications students, 564 completed a social-media learning-ecology survey. PCA plus clustering distinguished academically guided resource users from informally oriented digital learners; one cluster was about 19% of the sample with high academically guided scores. Cross-sectional self-reports cannot prove that matching instruction to those profiles improves learning.
  5. Across 22 Saudi universities in May 2020, 1,226 students rated emergency remote e-learning. Online-learning self-efficacy domains (learning, time management, technology) related to satisfaction; the satisfaction median was 69 (IQR 21), and 599 respondents (49%) sat below it. The association is correlational: raising self-efficacy was not experimentally shown to raise satisfaction.

Largest unresolved question

What a survey can claim depends on whether ratings are linked to anything outside the form. The lecture-mode study ties 13,585 complete surveys to exam grades and finds almost no mode–grade link. The emergency-remote, self-efficacy and Jesuit blended papers report challenges, satisfaction and success-factor ratings without those outcome links. Instructor CRE assessment models are faculty accounts of practice, not student performance. Treating all six as one 'engagement survey finding' mixes different questions.

Procedure

How it works

The main stages of the method. Specific protocols and analysis choices vary between studies.

  1. Write the items and the scale

    The researcher chooses closed questions and a response format — Likert anchors, checklists, or frequency counts. Wording, missing options, and whether 'not applicable' exists decide what later scores can mean; a poorly anchored item cannot be rescued by a large N.

  2. Recruit the sample and administer the form

    People who sit the survey are whoever was invited and willing, at one or a few moments. Selection, timing (for example during an emergency-remote term), and social-desirability pressure are part of the measurement, not background colour.

  3. Score scales and, sometimes, cluster

    Items are summed or averaged into scale scores; some papers then cluster respondents into profiles of reported practice. Those profiles describe how this sample's answers group. They are not an observed trace of study behaviour.

  4. Report counts, means, or profiles

    The product is how the sample described itself: engagement ratings, satisfaction, self-efficacy, instructor habits. A high rating is a judgement on a form, not proof that the person was behaving as an engaged learner or that the rated factor caused a grade.

  5. Do not treat the rating as the behaviour

    Surveys answer how this sample described itself under the form's conditions. They do not observe engagement, do not measure learning gains, and do not show that matching instruction to a self-report profile would change outcomes.

Common misconceptions

  • If students report high engagement or satisfaction, they were behaving as engaged learners.

    Self-report is a judgement on a form. Emergency-remote students reported moderate engagement while 132 of 138 faced multiple challenges; Saudi satisfaction sat around a median of 69 with about half below it. Neither dataset observed study behaviour or proved that the rating caused achievement.

  • A survey that clusters learners into profiles tells you how to assign instruction.

    The UOC communications clusters describe resource-use patterns in one online programme. The authors do not show that matching teaching to those profiles raises learning outcomes.

  • If instructors say they know many assessment approaches, students learned more under those models.

    The CRE paper maps faculty familiarity and three intended grading models. It does not measure student learning gains under each model.

Related