PhD Research Methodology Guide: Design & Analysis

1. The Research Onion: Establishing Your Philosophical Paradigm

Doctoral methodology begins with epistemological and ontological clarity. Saunders' Research Onion framework requires researchers to systematically peel through six methodological layers before collecting a single data point.

Positivism / Post-Positivism

Assumes an objective reality measurable via independent observation. Governs deductive quantitative hypothesis testing, econometric modeling, and experimental trials.

Interpretivism / Constructivism

Assumes social reality is constructed through subjective human meaning. Governs inductive qualitative exploration, phenomenology, and grounded theory.

2. Core Methodological Designs Compared

Dimension	Quantitative	Qualitative	Mixed-Methods (MMR)
Epistemological Stance	Positivist / Deductive	Interpretivist / Inductive	Pragmatist / Abductive
Core Purpose	Hypothesis testing, causal inference	Contextual meaning, theory building	Generalizability + deep context
Data Instruments	Structured Likert surveys, datasets	Semi-structured interviews, focus groups	Sequential survey + depth interviews
Analytical Tools	SPSS, AMOS, SmartPLS, R, STATA	NVivo, ATLAS.ti, MAXQDA	SPSS + NVivo (Joint Display Matrix)
Evaluation Criteria	Reliability (\(\alpha > 0.70\)), AVE (\(> 0.50\)), HTMT	Credibility, Transferability, Dependability	Triangulation, Meta-Inferences

3. Sampling Protocol & Statistical Power Calculations

Arbitrary sample sizes (e.g., "we surveyed 100 managers") lead to immediate thesis revisions. University examiners require mathematically defensible sample size determination.

A. G*Power A-Priori Sample Calculation:

For multiple linear regression or Structural Equation Modeling (PLS-SEM), sample size must be calculated using \(G*\text{Power}\) based on three parameters:

Effect Size (\(f^2\)): Small (\(f^2 = 0.02\)), Medium (\(f^2 = 0.15\)), or Large (\(f^2 = 0.35\)) per Cohen (1988).
Significance Level (\(\alpha\)): Standardized at \(\alpha = 0.05\) (95% confidence level).
Statistical Power (\(1 - \beta\)): Minimum institutional standard is \(0.80\) (80% power), with \(0.90\) preferred for high-stakes modeling.

B. Finite Population Formulas:

When sampling from a known finite population (\(N\)), use Yamane's formula or Cochran's formula with finite population correction:

n = N / [ 1 + N * (e)^2 ] (where e = margin of error, typically 0.05)

Methodology Design Blueprint Format: .DOCX (Word Document)

Download PhD Research Methodology Blueprint & G*Power Sampling Calculator Guide

Get the official institutional framework (.docx) detailing Saunders Onion paradigms, sample size formulas, survey scale design, and APA 7th validity tables.

Instant download • No spam

4. Psychometric Scale Validation & Measurement Model Metrics

Before testing structural paths (\(\beta\)), your measurement model must satisfy four rigorous validity thresholds:

1. Indicator Reliability: Standardized factor loadings (\(\lambda\)) for each observed item should exceed 0.708, indicating the construct explains >50% of the indicator's variance.

2. Internal Consistency: Composite Reliability (\(CR\)) and Cronbach's Alpha (\(\alpha\)) must range between 0.70 and 0.95.

3. Convergent Validity (AVE): Average Variance Extracted (\(AVE\)) must exceed 0.50, proving the latent construct captures more than half of its indicator variance.

4. Discriminant Validity (HTMT): Heterotrait-Monotrait ratio of correlations (\(HTMT\)) must be strictly < 0.85 (conservative) or < 0.90 (liberal threshold) per Henseler et al. (2015).

5. Controlling for Common Method Bias (CMB)

When cross-sectional data is collected from a single respondent source at a single point in time, examiners will scrutinize CMB:

Procedural Remedies: Anonymity assurance, clear item wording, and separating predictor and criterion variables across survey sections.
Harman's Single-Factor Test: Unrotated exploratory factor analysis where the first factor must account for < 50% of the total variance.
Full Collinearity Assessment (Kock, 2015): In PLS-SEM, variance inflation factors (\(VIF\)) for all latent constructs must be ≤ 3.3.

Quick Navigation

1. Philosophical Paradigms
2. Research Designs Compared
3. Sampling & G*Power
4. Scale Validation Metrics
5. Common Method Bias

Need Methodology Consultation?

Consult with our empirical methodologists for sampling validation, questionnaire scale audits, or software modeling.

Request Methodology Review Chat on WhatsApp

PhD Research Methodology Guide: Design, Sampling & Validation