Methodology Chapter Sample: Mixed-Methods Classroom Study

Note: this is a single chapter presented as reference material; the full dissertation would include the remaining chapters.

Model Answer

Research Design

This chapter sets out the methodology used to investigate the effect of a structured peer-tutoring intervention on primary pupils’ self-reported numeracy confidence. The study adopted a mixed-methods, explanatory sequential design (Creswell and Plano Clark, 2018), in which a quantitative phase, a pre- and post-intervention numeracy confidence questionnaire administered to two intact classes, was followed by a qualitative phase, semi-structured interviews with a purposive subsample of pupils, undertaken to help explain and contextualise the quantitative results.

This design was selected in preference to a purely quantitative approach because numeracy confidence is a subjective, self-perceived construct that questionnaire data alone cannot fully explain; a change, or absence of change, in mean confidence scores indicates whether an effect occurred but not why. Consistent with a pragmatist research paradigm (Creswell and Plano Clark, 2018), the study prioritises the practical research question, whether and how the intervention affected pupils’ numeracy confidence, over allegiance to a single methodological tradition, using quantitative methods to establish whether a measurable change occurred and qualitative methods to explore pupils’ own accounts of that change.

A true experimental design with random allocation to intervention and control conditions was not feasible within the constraints of a single-placement undergraduate dissertation conducted within normal school timetabling, since pupils could not ethically or practically be reassigned between existing class groups. The study therefore adopted a quasi-experimental, non-equivalent comparison-group design (Cohen, Manion and Morrison, 2018), comparing an intervention class receiving the peer-tutoring programme with a comparison class continuing standard numeracy teaching, while acknowledging in the limitations discussion below that pre-existing differences between the two classes cannot be fully ruled out as an alternative explanation for any observed difference.

Research Setting and Sample

The study was conducted in a two-form-entry primary school in the East Midlands, referred to throughout by the pseudonym Elm Bank Primary School to preserve the anonymity of the school and its pupils in line with the ethical principles discussed later in this chapter. The school was selected through convenience sampling, reflecting the researcher’s existing placement relationship with the school, a pragmatic and commonly used sampling approach in small-scale undergraduate education research (Cohen, Manion and Morrison, 2018), while introducing a limitation, discussed further below, that the single-school setting restricts the generalisability of the findings beyond broadly similar UK primary contexts.

Participants were drawn from two intact Year 4 classes, comprising 56 pupils in total following the exclusion of pupils whose parents or carers did not provide consent, as set out in Table 1. One class was assigned, following discussion with the class teachers regarding existing timetabling and curriculum-coverage constraints, to receive the six-week peer-tutoring intervention, while the second class continued with standard whole-class numeracy teaching and served as the comparison group. A purposive subsample of eight pupils from the intervention class, selected to reflect a spread of pre-intervention confidence scores, was subsequently invited to take part in the semi-structured interviews comprising the qualitative phase.

As Table 1 shows, the two classes were broadly comparable in size and gender balance, though no formal baseline test of statistical equivalence beyond the pre-intervention confidence questionnaire itself, discussed in the following section, was conducted, a methodological limitation consistent with the non-equivalent comparison-group design acknowledged above.

Data Collection Instruments

Two data collection instruments were used. The quantitative instrument was a ten-item numeracy confidence questionnaire adapted from an established self-efficacy scale format (Bandura, 2006), using child-friendly language and a five-point smiley-face Likert scale appropriate for Year 4 pupils, covering constructs including confidence attempting new problem types, willingness to answer questions in class, and perceived difficulty of numeracy homework. The questionnaire was administered identically to both classes immediately before the intervention period began and again immediately after its completion, generating paired pre- and post-intervention scores for each consenting pupil.

The qualitative instrument was a semi-structured interview schedule comprising six open-ended questions, developed following Braun and Clarke’s (2021) guidance on designing interview schedules to support reflexive thematic analysis, focusing on pupils’ own accounts of what, if anything, had changed in how they felt about numeracy, and what aspects of the peer-tutoring sessions they attributed any change to. The semi-structured format allowed sufficient flexibility to follow up individual pupils’ responses with clarifying prompts while maintaining consistency of core topic coverage across interviews, balancing depth of insight against comparability across the eight interviews conducted.

Both instruments were piloted before use: the questionnaire was piloted with a small group of pupils from a parallel Year 4 class not involved in the main study, resulting in the simplification of two items whose wording pupils found confusing during piloting, and the interview schedule was piloted with two pupils to check that question phrasing was age-appropriate and that the planned interview length, approximately ten minutes, was realistic for maintaining pupil engagement and concentration.

Procedure

Following ethical approval, described in the following section, information sheets and consent forms were distributed to parents and carers of pupils in both Year 4 classes, and assent was separately sought from pupils themselves using an age-appropriate, illustrated assent form, consistent with guidance on research involving children (BERA, 2018). The pre-intervention questionnaire was administered to both classes in the same week, read aloud item-by-item by the researcher to minimise the effect of variation in individual pupils’ reading ability on their responses.

The intervention class then received the six-week structured peer-tutoring programme, in which pupils worked in fixed same-ability-adjacent pairs for three fifteen-minute sessions per week, using a structured worksheet-based protocol in which the designated tutor pupil worked through problems alongside the tutee following a script provided by the class teacher, with pairs rotating the tutor and tutee role weekly. The comparison class continued with the school’s standard whole-class numeracy teaching over the same six-week period, with no equivalent paired-working component introduced.

The post-intervention questionnaire was administered to both classes in the week immediately following the intervention period, using the same read-aloud administration procedure as the pre-intervention questionnaire to maintain consistency. The eight semi-structured interviews with the purposive subsample were conducted individually, in a quiet room made available by the school, within the following two weeks, audio-recorded with parental and pupil consent, and transcribed verbatim by the researcher shortly after each interview.

Data Analysis

Quantitative questionnaire data were analysed using a Wilcoxon signed-rank test to compare each pupil’s pre- and post-intervention confidence scores within the intervention group, and a Mann-Whitney U test to compare the change in scores between the intervention and comparison groups; non-parametric tests were selected because five-point Likert-scale data of this kind are ordinal rather than interval in nature, and the relatively small sample size increases the risk that parametric assumptions, including normality, would not be reliably met (Cohen, Manion and Morrison, 2018).

Qualitative interview data were analysed using Braun and Clarke’s (2021) reflexive thematic analysis approach, involving familiarisation with the transcribed data, generation of initial codes, development of candidate themes, and iterative review and refinement of those themes against the coded dataset and the research question. Coding was conducted manually rather than using specialist software, reflecting both the modest scale of the dataset, eight interview transcripts, and the researcher’s own developing familiarity with computer-assisted qualitative analysis tools at undergraduate level.

Consistent with the explanatory sequential design outlined above, the qualitative themes were interpreted specifically in relation to the quantitative results, examining where pupils’ own accounts helped to explain the pattern of confidence-score change identified in the quantitative phase, an integration approach commonly termed ‘connecting’ the two strands of data (Creswell and Plano Clark, 2018), and reported together in the findings chapter that follows this methodology chapter.

Validity, Reliability and Positionality

Several strategies were used to strengthen the validity and reliability of the study’s findings. Data source triangulation, combining the quantitative questionnaire with qualitative interview data, was the primary strategy, allowing convergence, or divergence, between the two data strands to be examined directly rather than relying on a single method and data source, consistent with recommended practice for explanatory sequential mixed-methods designs (Creswell and Plano Clark, 2018). Instrument reliability was supported through the piloting process described above, and through the questionnaire’s adaptation from an established self-efficacy scale format rather than the use of an entirely untested, newly developed measure (Bandura, 2006).

Two threats to internal validity specific to this design are acknowledged. First, a Hawthorne-type effect is possible, in which pupils in the intervention class responded more positively on the post-intervention questionnaire partly because they were aware of receiving a distinct, novel intervention, independent of any genuine change in underlying numeracy confidence; this threat cannot be fully ruled out within a non-equivalent comparison-group design of this kind, though the inclusion of a comparison group receiving standard teaching, rather than no comparison group at all, partially mitigates it. Second, teacher effects cannot be fully separated from intervention effects, since the intervention and comparison classes were taught by different class teachers whose general teaching style and rapport with pupils may independently influence numeracy confidence irrespective of the peer-tutoring intervention itself, a limitation inherent to the use of pre-existing intact class groups rather than individually randomised allocation.

Finally, the researcher’s positionality as a placement student, rather than an independent external researcher, is acknowledged as a potential influence on the study, since pupils’ familiarity with the researcher from regular classroom contact may have affected how comfortably they responded to interview questions, and the researcher’s dual role as trainee-teacher and researcher required careful attention, discussed further in the ethical considerations below, to ensure pupils understood participation was voluntary and separate from their normal classroom relationship with the researcher.

Ethical Considerations

The study received ethical approval from the researcher’s university ethics committee prior to any contact with the school, pupils or parents, and was conducted in accordance with the British Educational Research Association’s (BERA, 2018) ethical guidelines for educational research. Four ethical considerations were of particular importance given that the research involved children in a school setting.

First, informed consent was sought at two levels, parental or carer consent and pupil assent, with the pupil assent process using age-appropriate, illustrated materials explaining the study’s purpose in language accessible to Year 4 pupils, and pupils were explicitly informed, in terms appropriate to their age, that they could withdraw from the questionnaire or interview at any point without needing to give a reason and without any consequence for their normal schoolwork or teacher relationships. Second, anonymity and confidentiality were protected throughout: the school and all pupils are referred to by pseudonym in this dissertation, questionnaire data were coded numerically rather than by name once matched across the two administration points, and audio recordings and transcripts were stored on an encrypted, password-protected university drive accessible only to the researcher and deleted following the moderation of the dissertation, in line with the university’s data protection and research governance policies.

Third, safeguarding procedures agreed with the school’s designated safeguarding lead were followed throughout, including a requirement that the researcher would immediately report any safeguarding disclosure made by a pupil during data collection to the school’s safeguarding lead rather than attempting to manage it independently, consistent with the researcher’s status as a placement student rather than a qualified, safeguarding-trained member of school staff. Fourth, the researcher considered the potential for a comparison-group design to disadvantage pupils who did not receive the intervention; this was addressed by agreeing with the class teacher that, subject to the intervention showing a positive effect, an adapted version of the peer-tutoring programme would subsequently be offered to the comparison class after the research data collection period had concluded, so that participation in the comparison group did not permanently exclude pupils from a potentially beneficial intervention.

References

• Bandura, A. (2006) ‘Guide for constructing self-efficacy scales’, in Pajares, F. and Urdan, T. (eds.) Self-Efficacy Beliefs of Adolescents. Greenwich, CT: Information Age Publishing, pp. 307–337.
• BERA (2018) Ethical Guidelines for Educational Research. 4th edn. London: British Educational Research Association.
• Braun, V. and Clarke, V. (2021) Thematic Analysis: A Practical Guide. London: Sage.
• Cohen, L., Manion, L. and Morrison, K. (2018) Research Methods in Education. 8th edn. Abingdon: Routledge.
• Creswell, J.W. and Plano Clark, V.L. (2018) Designing and Conducting Mixed Methods Research. 3rd edn. Thousand Oaks, CA: Sage.
• Denscombe, M. (2021) The Good Research Guide. 7th edn. Maidenhead: Open University Press.
• Fraenkel, J.R., Wallen, N.E. and Hyun, H.H. (2019) How to Design and Evaluate Research in Education. 10th edn. New York: McGraw-Hill.
• Gorard, S. (2013) Research Design: Creating Robust Approaches for the Social Sciences. London: Sage.
• Punch, K.F. and Oancea, A. (2014) Introduction to Research Methods in Education. 2nd edn. London: Sage.
• Robson, C. and McCartan, K. (2016) Real World Research. 4th edn. Chichester: Wiley.