Research Methodology

Research methodology is the overall, disclosed framework of justified choices that connects a research question to its findings: the research design chosen (exploratory, descriptive, or causal), the data sources and sampling approach used, the analysis technique applied, and the checks used to establish validity and reliability. It is one level of abstraction above a single research method. A "method" (a survey, a structured interview, a desk-research pull from a government database) is one data-collection technique; a "methodology" is the reasoned architecture that explains why that technique was chosen for that question, how the resulting data was sampled and analyzed, and how confident the reader should be in the output. Every market estimate, ranking, or forecast on this site should be traceable to an explicit methodology of this kind, not just a number.

This page covers the cross-disciplinary framework for designing, justifying, and disclosing a research methodology, and the standard taxonomy of research designs (exploratory, descriptive, causal) used across market research, social science, and applied business research. It does not re-cover the specific primary-vs-secondary and qualitative-vs-quantitative data-collection techniques already covered in depth on the Market Research page; this page sits one level above those, providing the structure that ties individual techniques together into an evaluable whole. It also does not cover the formula-based sizing and forecasting methods (Market Size, TAM/SAM/SOM, CAGR, Market Forecast) themselves, only the methodology-disclosure standard those methods should be presented under. It does not cover jurisdiction-specific research-ethics or data-protection law (e.g., GDPR consent requirements for panel research); it references only the professional-standards baseline (the ICC/ESOMAR Code) that most market-research bodies observe globally.

Use this framework whenever a market estimate, ranking, forecast, or claim is being published for a reader who cannot see the underlying raw data directly -- which in practice means every industry, country, city, market-intersection, and emerging-market page on this site, each of which carries a methodology field per the data contract. Apply it before commissioning or citing primary research (to pick the right design), when auditing whether a third-party report's methodology section actually discloses its sample size, source dates, and confidence level, and when writing the methodology disclosure for your own study or page. It is not needed for purely definitional or reference content that makes no empirical claim requiring evidence, such as a glossary entry.

1. Define the research question or objective precisely, and name the decision it is meant to inform -- a vague objective produces a methodology that cannot be evaluated against it.
2. Choose the research design to match the question: exploratory when the problem is not yet well defined (literature review, informal interviews, case studies) to generate hypotheses and scope the topic; descriptive when the goal is to measure and describe a defined population or market accurately (structured surveys, secondary-data analysis); causal when the goal is to establish whether and why one variable drives another (controlled experiments, statistically controlled comparisons).
3. Identify and justify the data sources: primary (collected first-hand for this specific question) versus secondary (repurposed from a source collected for another purpose) -- and disclose each source's date, geographic coverage, and known gaps.
4. Define the sampling frame and method -- probability (random, systematic, stratified) versus non-probability (convenience, purposive, snowball) -- along with the achieved sample size and, where applicable, the margin of error or confidence interval; disclose who was excluded and why.
5. Specify the analysis technique applied to the data (descriptive statistics, regression, thematic coding, etc.) and the checks used to establish validity (does it measure what it claims to measure) and reliability (would it reproduce consistently if repeated).
6. Triangulate: cross-check the finding against at least one independent source, method, or dataset before publishing it as a headline figure, rather than relying on a single input.
7. Disclose the resulting confidence level in plain language -- e.g. "high confidence, triangulated across three official sources" versus "estimate, single source, wide range" -- and set a review or refresh date.
8. Publish the methodology next to the finding itself, not buried in an appendix or omitted -- a reader should be able to see steps 1-7 without hunting for them.

Not yet available.

Illustrative example only -- a hypothetical study, not a sourced estimate for any real company or market.

Question: "What is the current market size and growth rate of AI-powered contract-review software, to decide whether to enter as a vendor?" This is decision-oriented and testable, so it passes step 1.

Design: primarily descriptive (measuring current size), with an exploratory component first to establish the category's boundaries, since it is new and not yet consistently defined across sources. A causal design is not needed -- the question is what the market's size is, not why it is growing.

Sources: (a) secondary -- disclosed revenue figures from three public legal-tech companies' investor filings; (b) secondary -- a paid industry report estimating the wider legal-tech category; (c) primary -- 12 structured interviews with in-house counsel at mid-size companies (200-2,000 employees, two industries), conducted over three weeks, to estimate category adoption.

Sampling: the 12 interviews are a non-probability, purposive sample targeted at a specific company profile; this is disclosed as a limitation rather than treated as a random, representative sample of all buyers.

Analysis: two independent estimates are built and compared -- a top-down figure (legal-tech market size x category adoption rate from the interviews) and a bottom-up figure (sum of disclosed vendor revenues, grossed up for estimated private-competitor share).

Triangulation: the top-down estimate ($470M annualized) and the bottom-up estimate ($410M annualized) land within about 15% of each other, which is treated as convergence. The published figure is the range and its midpoint, not a single false-precision number.

Confidence and review: labeled "moderate confidence -- convergent estimate from two independent methods; primary sample small (n=12) and non-random" with a 12-month review date set given how fast the category is moving.

Disclosure: the resulting page's methodology field states all of the above in a few sentences, and each contributing figure is footnoted with its actual source and date -- so a reader can judge the estimate's reliability without re-doing the research themselves.

Not yet available.

• ESOMAR, "Data, Research and Insights Explained" -- the ICC/ESOMAR International Code, the global professional-standards benchmark for market-research disclosure and ethics.
• Pew Research Center, "Our Methods" -- a real-world example of a research organization publishing full methodological transparency alongside its findings.
• Golafshani, N. (2003), "Understanding Reliability and Validity in Qualitative Research," The Qualitative Report -- the standard reference on validity/reliability across qualitative and quantitative paradigms.
• Western Sydney University, "Types of Research Design" (Customer Insights coursebook) -- the exploratory/descriptive/causal research-design taxonomy, after Cooper & Schindler.