Research methodology guide. Referenced by the research-designer and statistical-analyst agents when selecting research designs and analysis methods. Use for 'research methodology', 'experiment design', or 'statistical analysis' requests. Actual data collection and IRB review processing are out of scope.
Enhances the methodology design capabilities of the research-designer and statistical-analyst agents.
| Design | Purpose | Variable Control | Causal Inference |
|---|---|---|---|
| Experiment (RCT) | Verify causation | High | Strong |
| Quasi-experiment | Causation (limited control) | Medium | Medium |
| Correlational Study | Identify relationships | Low | Weak |
| Survey | Assess current status | None | None |
| Longitudinal Study | Change over time | Varies | Medium |
| Design | Purpose | Data Collection |
|---|---|---|
| Case Study | Deep contextual understanding | Interviews, observation, documents |
| Grounded Theory | Theory generation | Interviews, field notes |
| Phenomenology | Essence of experience | In-depth interviews |
| Ethnography | Cultural/group understanding | Participant observation |
Sequential: Quantitative -> Qualitative (explain results)
Qualitative -> Quantitative (generate hypotheses)
Concurrent: Quantitative + Qualitative (triangulation)
Quantitative (mean comparison):
n = (Z_alpha/2 + Z_beta)^2 x 2*sigma^2 / d^2
Z_alpha/2 = 1.96 (significance level 0.05)
Z_beta = 0.84 (power 80%)
sigma = standard deviation (from prior research)
d = effect size (minimum meaningful difference)
Survey:
n = Z^2 x p(1-p) / e^2
Z = 1.96, p = 0.5 (maximum variance), e = margin of error
e.g.: 95% confidence, 5% margin -> n = 384 participants
| Method | Description | Suitable For |
|---|---|---|
| Simple Random | Random from population | Small, homogeneous populations |
| Stratified | Proportional by subgroup | Heterogeneous populations |
| Cluster | By natural group unit | Region-based surveys |
| Convenience | Accessible participants | Exploratory research (limited) |
Q1: Variable type?
+-- Continuous -> Q2
+-- Categorical -> Q3
Q2: Number of comparison groups?
+-- 1-2 -> t-test (independent/paired)
+-- 3+ -> ANOVA
+-- Relationship analysis -> Regression/Correlation
Q3: Number of variables?
+-- 2 -> Chi-square
+-- 3+ -> Logistic regression
| Method | Purpose | Requirements | Reported Items |
|---|---|---|---|
| t-test | Compare 2 group means | Normality, equal variance | t, df, p, Cohen's d |
| ANOVA | Compare 3+ group means | Normality, equal variance | F, df, p, partial eta-squared |
| Chi-square | Categorical variable relationship | Expected frequency >= 5 | chi-square, df, p, Cramer's V |
| Correlation | Relationship between 2 variables | Linearity | r, p |
| Regression | Prediction/explanation | Linearity, independence | beta, R-squared, p, VIF |
| Metric | Small | Medium | Large |
|---|---|---|---|
| Cohen's d | 0.2 | 0.5 | 0.8 |
| r (correlation) | 0.1 | 0.3 | 0.5 |
| partial eta-squared (ANOVA) | 0.01 | 0.06 | 0.14 |
| R-squared (regression) | 0.02 | 0.13 | 0.26 |
| Item | Verification |
|---|---|
| Informed Consent | Research purpose, procedure, risks, voluntariness |
| Personal Data | De-identification, retention period, destruction |
| Vulnerable Populations | Special protections for minors, patients |
| Conflicts of Interest | Funding source, interest disclosure |
| Item | Criteria |
|---|---|
| Research Question | PICO or SPIDER structure |
| Design Fit | Design matches the research question |
| Sampling | Calculation basis + sampling method |
| Analysis Method | Suited to data characteristics |
| Effect Size | Statistical significance + practical meaning |
| Ethics | IRB approval or exemption justification |