8.3 Research Designs and Methodologies for Data Collection and Analysis
Key Takeaways
- TOS competencies B.2.3, B.2.4 and B.2.5 require candidates to distinguish the flow of the research process and to examine and utilise common research designs and methodologies.
- Quantitative designs range from descriptive and correlational through quasi-experimental to experimental, differing in control over assignment.
- Qualitative designs include phenomenology, case study, ethnography, grounded theory and narrative inquiry, each with a distinct product.
- Participatory action research and single-system designs are the two designs most distinctive to social work.
- Validity, reliability and trustworthiness are the quality criteria, and triangulation strengthens both traditions.
8.3 Research Designs and Methodologies for Data Collection and Analysis
Blueprint anchor. TOS competencies B.2.3 ("distinguish the distinctive flow in each step of the research process"), B.2.4 ("examine some common research designs, research methodologies for data collection and analysis") and B.2.5 ("utilize common research designs, research methodologies") together account for three items and supply the design vocabulary used throughout the research block.
1. Quantitative Designs
| Design | Structure | What it can claim | Social work example |
|---|---|---|---|
| Descriptive | Measure and summarise, no comparison | Prevalence, distribution, profile | Barangay profile of persons with disabilities by type and service use |
| Comparative | Compare existing groups | Group differences, not causes | Service satisfaction of upland versus lowland clients |
| Correlational | Measure association between variables | Strength and direction of relationship | Association between session attendance and outcome scores |
| Ex post facto | Examine an outcome after a naturally occurring cause | Suggestive causal patterns only | Comparing outcomes of children who were and were not placed in foster care |
| Quasi-experimental | Intervention with a comparison group but no random assignment | Stronger causal inference; selection bias remains | Programme barangays versus matched non-programme barangays |
| Experimental | Random assignment to intervention and control | Strongest causal inference | Randomised allocation to two counselling protocols |
| Longitudinal | Repeated measurement over time | Change and sequence | Tracking grantee enterprises at 6, 12 and 24 months |
| Cross-sectional | Single point in time | Snapshot; no sequence | One-round household survey |
[!IMPORTANT] Ethical limit on experimentation. Random assignment that withholds a legally mandated service from a control group is not permissible. Social work evaluation therefore leans heavily on quasi-experimental designs, waiting-list comparisons and single-system designs. An item offering "randomly deny assistance to half the applicants" is offering an ethics violation dressed as rigour.
2. Qualitative Designs
| Tradition | Central question | Typical data | Product |
|---|---|---|---|
| Phenomenology | What is the lived experience of this phenomenon? | In-depth interviews | Description of the essence of the experience |
| Case study | What is happening in this bounded case? | Multiple sources on one case | Rich account of a case in context |
| Ethnography | What is the culture of this group? | Prolonged participant observation | Cultural description |
| Grounded theory | What theory emerges from these data? | Interviews with constant comparison | A theory grounded in the data |
| Narrative inquiry | How do people story their lives? | Life histories | Analysed narrative accounts |
| Participatory action research | What will we change together? | Community-generated data | Action plus knowledge |
Sampling in qualitative work is purposive, and adequacy is judged by saturation — the point at which new cases stop yielding new categories — rather than by a pre-computed sample size.
3. Designs Distinctive to Social Work
Participatory action research (PAR). The community defines the problem, participates in gathering and analysing data, and owns the resulting action. Knowledge production and social change are the same activity. This is the design that anchors the community organisation method.
Single-system design (SSD). Evaluates change in one client, family, group or community over time.
| Notation | Structure | Use |
|---|---|---|
| AB | Baseline (A) then intervention (B) | The common practice design |
| ABA | Baseline, intervention, withdrawal | Rarely ethical when the intervention helps |
| ABAB | Baseline, intervention, withdrawal, reinstatement | Strong internal evidence; ethically constrained |
| ABC | Baseline then two successive interventions | When the first intervention is changed |
The strength of SSD is that it can be run inside ordinary practice with a repeated measure — frequency of a target behaviour, a standardised scale, a self-anchored rating — recorded at each contact.
4. Mixed Methods
| Type | Sequence | Purpose |
|---|---|---|
| Explanatory sequential | Quantitative → qualitative | Numbers first, then explain why |
| Exploratory sequential | Qualitative → quantitative | Build an instrument from local meaning, then measure |
| Convergent parallel | Both at once, then compare | Corroborate through triangulation |
| Embedded | One nested inside the other | Process evaluation inside an outcome study |
5. Data Collection Methods
| Method | Strength | Weakness | Fit |
|---|---|---|---|
| Survey questionnaire | Reaches many; standardised | Shallow; literacy dependent | Prevalence and profile |
| Structured interview | Comparable across respondents | Interviewer time | Household surveys with low literacy |
| In-depth interview | Depth and meaning | Not generalisable | Experience and process |
| Focus group discussion | Group meanings; efficient | Dominant voices; unsuitable for sensitive disclosure | Community perceptions |
| Observation | Behaviour as it occurs | Reactivity; observer effect | Group dynamics; facility practice |
| Document and record review | Non-reactive; historical | Records may be incomplete or biased | Case-record studies; policy analysis |
| Participatory tools | Community-owned; visual, low-literacy friendly | Facilitation-dependent | Community profiling and PAR |
6. Quality Criteria
Quantitative.
- Reliability — consistency of measurement across time, items and raters.
- Validity — measuring what is intended: content, criterion and construct validity.
- Internal validity — confidence that the intervention, not something else, produced the change.
- External validity — generalisability beyond the study setting.
Qualitative (trustworthiness).
- Credibility — findings ring true; supported by prolonged engagement and member checking.
- Transferability — thick description lets readers judge applicability elsewhere.
- Dependability — the process is documented and auditable.
- Confirmability — findings traceable to data rather than researcher preference.
Triangulation — of data sources, methods, investigators or theories — strengthens both traditions.
7. Worked Practice Application
A provincial office must evaluate a two-year community-based drug rehabilitation programme and answer three questions: whether relapse rates changed, what participants experienced, and what the barangays should do next.
Design decision — a convergent mixed-methods evaluation with a participatory component.
| Question | Design element | Justification |
|---|---|---|
| Did relapse rates change? | Quasi-experimental comparison of programme barangays with matched non-programme barangays, plus routine record review | Random assignment would mean withholding a mandated service; matching is the ethical alternative |
| What did participants experience? | Phenomenological in-depth interviews to saturation | Meaning and mechanism cannot be read from rates |
| What should barangays do next? | Participatory action research cycle with barangay anti-drug councils and family members | Recommendations that communities did not help write are rarely implemented |
Quality safeguards. Triangulate record data with interview accounts; document the matching procedure for dependability; member-check the interview themes with participants; and report the limitation that non-random assignment leaves selection differences between barangays unresolved.
The examinable competence is precisely this: matching each question to the design capable of answering it, and refusing designs that are rigorous on paper but unethical in a service setting.
An evaluator proposes to randomly deny crisis assistance to half of eligible applicants in order to create a control group. What is the correct professional response?
A worker records a client's target behaviour for three weeks before beginning an intervention, then continues recording during the intervention. This is an example of:
In a qualitative study, adequacy of the sample is judged primarily by: