Research Methodology – Detailed Process & Notes

Research Process: Overview

  • Research process consists of interconnected, overlapping activities (labeled I–VII in the chart, expanded to 11 actionable steps).
  • Steps rarely follow a rigid linear order; later requirements must be anticipated early to avoid obstacles that can jeopardize completion.
  • Useful procedural guideline (11 steps):
    1. Formulating the research problem
    2. Extensive literature survey
    3. Developing the hypothesis
    4. Preparing the research design
    5. Determining the sample design
    6. Collecting the data
    7. Execution of the project
    8. Analysis of data
    9. Hypothesis testing
    10. Generalisations and interpretation
    11. Preparation of the report / presentation of results

Step 1 – Formulating the Research Problem

  • Two broad types of problems:
    • Questions about states of nature ("what is?")
    • Questions about relationships between variables ("why/how do variables interact?").
  • Activities:
    • Select and narrow a general topic into a specific, operational problem statement.
    • Resolve ambiguities; assess feasibility of potential solutions.
    • Ensure unambiguous definitions to discriminate relevant from irrelevant data.
  • Essential sub-steps:
    • Thorough understanding (discussion with colleagues, experts, supervisors, administrative agencies).
    • Rephrasing into analytical/operational terms.
  • Literature acquaintance (conceptual + empirical) supplies available data context.
  • Quote (Prof. W. A. Neiswanger): Statement of objectives determines data, relevant characteristics, relationships to explore, analytical techniques, and report form.
  • Sequential narrowing often used—each iteration more specific, analytical, realistic.
  • Ethical/practical implication: mis-definition introduces bias, wasted resources, and invalid conclusions.

Step 2 – Extensive Literature Survey

  • Write a brief summary/synopsis; mandatory for Ph.D. proposals.
  • Use abstracting/indexing journals, bibliographies, academic journals, conference proceedings, government reports, books.
  • Follow citation chains; a good library is invaluable.
  • Purpose: identify conceptual frameworks, data sources, methodological precedents, research gaps.

Step 3 – Development of Working Hypotheses

  • Working hypothesis: tentative assumption formulated to derive and test logical/empirical consequences.
  • Roles:
    • Focal point for research; guides data requirements and analysis techniques.
    • Delimits scope; sharpens thinking.
  • Desirable features: specific, limited, testable, precisely worded.
  • Four common routes to generate hypotheses:
    a. Expert/colleague discussions on origins & objectives.
    b. Examination of existing data/records for trends & peculiarities.
    c. Review of similar or related studies.
    d. Exploratory field investigation (limited interviews).
  • Arise from a-priori reasoning, data review, expert counsel.
  • Exception: Exploratory/formulative studies may proceed without hypotheses.

Step 4 – Preparing the Research Design

  • Research design = conceptual structure for efficient evidence collection (max info, minimal , time, effort).
  • Purpose categories shape design:
    • Exploration (flexible design).
    • Description (bias minimization, reliability maximization).
    • Diagnosis.
    • Experimentation.
  • Broad classes:
    • Experimental designs: informal (before–after w/ or w/o control; after-only w/ control) & formal (completely randomized, randomized block, Latin square, simple/complex factorial).
    • Non-experimental designs.
  • Preparation considerations:
    1. Information sources & collection methods.
    2. Skill availability (researcher & staff).
    3. Organization & rationale of chosen methods.
    4. Time constraints.
    5. Financial constraints.

Step 5 – Determining the Sample Design

  • Universe/Population = all items of interest.
  • Census: complete enumeration; theoretically bias-free but costly, time-consuming, still vulnerable to accumulating bias.
  • Sampling chosen when census impractical (e.g., blood testing).
  • Sample design = pre-specified plan for selecting sample from population.
  • Two umbrella categories:
    • Probability samples (known P(i)foreachelement).<br/>•Non−probabilitysamples(unknownfor each element). <br /> • Non-probability samples (unknownP(i)).
  • Detailed designs:
    1. Deliberate/Purposive (includes Convenience & Judgement sampling) – interviewer discretion; useful in qualitative hypothesis formation; risk of bias when population heterogeneous.
    2. Simple Random Sampling – equal chance P(i)=\frac{n}{N};selectionvialotteryorrandom−numbertables.Forinfinitepopulations:identicalandindependentselectionprobabilities.</li><li>SystematicSampling–selectevery; selection via lottery or random-number tables. For infinite populations: identical and independent selection probabilities. </li> <li>Systematic Sampling – select everyk^{th} element after random start; efficient when ordered list exists.
    3. Stratified Sampling – divide into non-overlapping strata, then simple random sample within each; ensures representation, reduces variance.
    4. Quota Sampling – cost-saving variant of stratification; interviewer fills proportional quotas; non-probability, judgement-based.
    5. Cluster Sampling – sample entire groups (clusters) rather than individuals; e.g., 100 clusters of 150 credit-card holders; cheaper fieldwork, higher sampling error.
    6. Area Sampling – geographical clustering when no population list; interview everyone in selected areas.
    7. Multi-Stage Sampling – hierarchical cluster approach (states → districts → towns → families); if random at each stage = multi-stage random sampling.
    8. Sequential Sampling – sample size decided adaptively via statistical rules (common in quality-control acceptance sampling).
  • Mixed sampling: combining several methods within one study.
  • Practical rule: prefer random sampling for bias control and error estimation; purposive designs acceptable for small, well-known universes or cost reasons.

Step 6 – Collecting the Data

  • Data often inadequate; new collection needed.
  • Primary data via Experiment or Survey.
  • Survey collection modes:
    1. Observation – direct investigator observation of current behavior; limited scope; expensive for large samples.
    2. Personal Interview – structured questioning; quality depends on interviewer skill.
    3. Telephone Interview – rapid, useful in industrialized settings under tight deadlines.
    4. Mail Questionnaire – largest use in economic/business surveys; requires pilot testing for weaknesses.
    5. Schedules with Enumerators – trained field workers fill schedules; quality hinges on enumerator competence; field checks recommended.
  • Selection factors: investigation nature, objectives, scope, budget, time, desired accuracy, researcher experience ("common sense chief requisite; experience chief teacher" – Dr A. L. Bowley).

Step 7 – Execution of the Project

  • Proper execution ensures adequate, dependable data.
  • For questionnaire surveys: pre-coding answers for machine processing.
  • Interview studies: select & train interviewers via manuals; perform field checks for sincerity & efficiency.
  • Maintain statistical control; watch for unanticipated factors.
  • Address non-response: list non-respondents, draw sub-sample, apply expert follow-ups.

Step 8 – Analysis of Data

  • Sequential operations:
    • Editing (quality improvement) → Coding (assign symbols) → Tabulation (tables, often computer-assisted).
  • Data reduction: condense unwieldy data into manageable categories/groups.
  • Post-tabulation analysis: compute percentages, coefficients, other statistics using defined formulae.
  • Test relationships/differences via significance testing:
    • Example: two weekly-wage samples from factories; test whether means differ beyond chance.
    • Example: analysis of variance (ANOVA) for seed varieties yielding different results.

Step 9 – Hypothesis Testing

  • Determine whether data support or contradict hypotheses.
  • Common statistical tests: \chi^2(Chi−square),(Chi-square),ttest,test,F test, etc.
  • Outcomes: accept or reject hypothesis.
  • If no prior hypothesis, convert generalisations into hypotheses for future studies.

Step 10 – Generalisations & Interpretation

  • Repeatedly upheld hypotheses → generalisations → theory building.
  • If research began without hypotheses, interpret findings within theoretical frameworks.
  • Interpretation often raises new questions, sparking further research (cyclical advancement of knowledge).

Step 11 – Preparation of the Report / Thesis

  • Must be written meticulously; three main sections:
    1. Preliminary Pages – Title & date, acknowledgements, foreword, table of contents, lists of tables/graphs.
    2. Main Text –
      a. Introduction (objectives, methodology, scope, limitations).
      b. Summary of Findings & Recommendations (non-technical).
      c. Main Report (logically sequenced sections).
      d. Conclusion (clear, precise summing-up).
    3. End Matter – Appendices (technical data), Bibliography (books, journals, reports), Index (especially for published work).
  • Writing style: concise, objective, simple language; avoid vague terms ("it seems", "there may be").
  • Use charts/illustrations only when they enhance clarity.
  • State calculated confidence limits \text{CI} = \hat{\theta} \pm z{\alpha/2}\,\sigma{\hat{\theta}}$$ and note operational constraints experienced.

Interconnections & Practical/Philosophical Implications

  • Each step influences subsequent steps; early neglect of later needs can derail a study.
  • Sampling ethics: fairness, representation, transparency in probability assignment.
  • Data-collection ethics: informed consent, respondent confidentiality, honest reporting.
  • Practicality vs. rigor: balance cost/time limits against statistical robustness.
  • Continuous improvement: interpretation feeds back into literature and problem formulation for future cycles.