Research Design, Methodology, and Sampling
❦
What the Methodology Chapter Must Contain
| Section | Content |
|---|---|
| Research philosophy and approach | Paradigm; deductive, inductive, or abductive logic |
| Research design and strategy | Survey, case study, mixed methods, and so on; time horizon |
| Population and sampling | Target population, sampling frame, technique, sample size |
| Data collection | Instrument, procedure, pilot test, period |
| Measures | Operationalization, scales, sources, reliability and validity |
| Data analysis plan | Techniques and software, matched to each objective |
| Ethical considerations | Clearance, consent, confidentiality |
| Limitations of the design | Honest statement of constraints |
Write the methodology in the past tense in the final dissertation and the future tense in the proposal, give reasons for each choice, and cite methodological sources (Saunders et al., 2019).
Population and Sampling
- Population: the entire group of interest (for example, all employees of licensed commercial banks in Jaffna District).
- Sampling frame: the list from which you sample. In Sri Lanka, possible frames include lists of companies listed on the Colombo Stock Exchange, registers of companies and business associations, chambers of commerce, divisional secretariat lists of enterprises, and organizational staff lists.
- Sampling units and the unit of analysis (individual, firm, or branch).
| Probability sampling | Description |
|---|---|
| Simple random | Every unit has an equal chance |
| Systematic | Every k-th unit |
| Stratified | Random samples within subgroups (for example by firm size) |
| Cluster | Random selection of groups (for example districts), then units |
| Non-probability sampling | Description |
|---|---|
| Convenience | Easily accessible units |
| Purposive (judgment) | Chosen for their relevance; typical in qualitative research |
| Quota | Fixed numbers from subgroups |
| Snowball | Participants recommend others (useful for hidden populations) |
Probability sampling supports statistical generalization; non-probability sampling is common in student work but limits generalization, so say so.
Sample Size
| Approach | Formula or rule | Example |
|---|---|---|
| Krejcie and Morgan (1970) table | Tabulated sample sizes for a given population | About 278 for N = 1 000; about 370 for N = 10 000; about 384 for very large populations |
| Cochran (1977) | n₀ = z² p(1 − p) / e²; with p = 0.5, z = 1.96, e = 0.05, n₀ = 384.16; finite population correction: n = n₀ / [1 + (n₀ − 1) / N] | For N = 2 000, n ≈ 322 |
| Yamane (1967) | n = N / (1 + N e²) | N = 2 000, e = 0.05: n = 2 000 / 6 = 333 |
| Regression rule (Green, 1991) | N ≥ 50 + 8m (to test the overall model); N ≥ 104 + m (to test individual predictors), where m is the number of predictors | Five predictors: at least 90 and 109 |
| Power analysis | Use G*Power with the expected effect size, α = 0.05, and power of 0.80 (Cohen, 1988) | |
| SEM | Often 200 or more cases, or a ratio of cases to parameters (Kline, 2016) | |
| PLS-SEM | The "inverse square root method" and minimum R² methods (Kock & Hadaya, 2018); avoid the discredited "10 times" rule where possible | |
| Interviews | No fixed number; see information power (Malterud et al., 2016) and saturation (Guest et al., 2006) | About 12 for a homogeneous sample, more if diverse |
Plan to over-sample to allow for non-response and invalid questionnaires (for example, distribute 20 to 30 percent more than the target).
Reliability and Validity: Concepts
- Reliability: consistency (internal consistency by Cronbach's alpha; test-retest).
- Validity: measuring what is intended: content (experts' judgment), construct (convergent and discriminant), criterion (relationship with an external criterion), and for the whole study, internal validity (causal inference), external validity (generalization), and statistical conclusion validity (Shadish et al., 2002).
- In qualitative research, trustworthiness criteria replace these (Chapter 12).
Pilot Testing
Test the instrument on a small group (about 10 to 30 respondents) from a similar population; revise unclear items; check reliability; time the questionnaire; and test your data-entry and analysis procedures.
CHAPTER 9