Research design Study Guide

📖 Core Concepts Research design – the overall plan that links questions, data‑collection, and analysis to produce valid answers. Epistemology – beliefs about how knowledge is generated; shapes the choice of design. Ontology – beliefs about what reality is like; influences what can be measured. Fixed design – fully specified before data collection; usually quantitative and theory‑driven. Flexible design – can be altered during data collection; often qualitative. Confirmatory research – tests pre‑specified (a priori) hypotheses; aims to keep the type I error rate ($\alpha$) low. Exploratory research – generates hypotheses after looking at the data; prioritizes detecting real effects (low type II error, $\beta$). State problem – asks “what is the condition now?” – needs a single measurement. Process problem – asks “how does it change over time?” – needs repeated measurements. Experimental design – manipulates an independent variable, randomizes participants, controls confounds. Non‑experimental (relational) design – measures variables without manipulation; uses correlation or group comparisons. Case study – deep, contextual examination of one case. Grounded theory – systematic coding and constant comparison to build theory directly from data. --- 📌 Must Remember Valid answers ↔ strong research design; unreliable answers ↔ weak design. Fixed = pre‑defined variables & analysis → quantitative. Flexible = can adjust grouping/measurements → qualitative. Confirmatory = a priori hypotheses → control $\alpha$; Exploratory = post‑hoc hypotheses → lower $\beta$. HARKing = “hypothesizing after results are known” – a major questionable practice. State problem → single‑time point; Process problem → longitudinal/repeated‑measure design. Random assignment + manipulation = hallmark of experimental designs. Correlation ≠ causation – never infer causality from relational designs. Power analysis (pre‑study) determines sample size for desired $\alpha$ / $\beta$ levels. --- 🔄 Key Processes Design Development Clarify theory/model → formulate research question(s). Choose epistemological & ontological stance → influences design type. Select study type (descriptive, correlational, experimental, review, meta‑analytic). Define variables, hypotheses, measurement tools, and analysis plan. Fixed Design Workflow Operationalize all variables before data collection. Conduct power analysis → decide sample size. Randomly assign participants (if experimental). Collect data → apply pre‑specified statistical test. Flexible Design Workflow Begin with broad research question. Collect data (e.g., interviews, observations). Perform iterative coding & constant comparison. Refine focus, grouping, or measurement tools as insights emerge. Confirmatory vs Exploratory Confirmatory: Write hypotheses → design study → collect data → test with pre‑registered analysis. Exploratory: Collect data → explore patterns → generate hypotheses → (optional) follow‑up confirmatory study. --- 🔍 Key Comparisons Fixed vs Flexible Fixed: variables known → quantitative; design locked in early. Flexible: variables may evolve → qualitative; design adaptable. Confirmatory vs Exploratory Confirmatory: a priori hypotheses, low $\alpha$, high inferential power. Exploratory: post‑hoc hypotheses, lower $\beta$, higher discovery potential. Experimental vs Relational (Non‑experimental) Experimental: manipulation + randomization → can infer causality. Relational: measurement only → only associations, no causal claim. State vs Process Problems State: single snapshot → cross‑sectional design. Process: change over time → longitudinal/repeated‑measure design. --- ⚠️ Common Misunderstandings “Correlation proves causation.” → Correlation shows co‑movement, not directionality. “All research can be fixed.” → Qualitative phenomena often require flexible designs. “Exploratory findings are automatically publishable as confirmatory.” → Leads to HARKing; must label appropriately. “A single measurement is enough for any question.” – Wrong for process problems; need repeated measures. “Random assignment guarantees no bias.” – It reduces, not eliminates, all sources of bias (e.g., selection, measurement). --- 🧠 Mental Models / Intuition “Blueprint vs. Scaffold” – Fixed designs are a blueprint (exact plan); flexible designs are a scaffold you can adjust as the building rises. “Detective vs. Explorer” – Confirmatory research is a detective testing a suspect (hypothesis); exploratory research is an explorer mapping unknown terrain. “Snapshot vs. Movie” – State problems are a photograph; process problems are a movie requiring multiple frames. --- 🚩 Exceptions & Edge Cases Quasi‑experiments – experimental‑like but lack full randomization; treat as hybrid (fixed but with weaker causal claims). Mixed‑methods – may combine fixed quantitative components with flexible qualitative components. Grounded theory can be used within a broader fixed framework if theory development is the primary goal. --- 📍 When to Use Which Use Fixed design when variables are well‑defined, you have a strong theory, and you need causal inference (e.g., lab experiment). Use Flexible design when the phenomenon is poorly understood, variables are emergent, or you need rich contextual insight (e.g., case study, grounded theory). Choose Confirmatory if you have pre‑existing theory and need high‑confidence inference. Choose Exploratory for new domains, pilot work, or when generating hypotheses for later testing. Select State design for prevalence or cross‑sectional surveys. Select Process design for growth, development, or change studies (longitudinal, repeated measures). --- 👀 Patterns to Recognize Presence of random assignment → experimental, fixed, causal inference. Multiple time points → process problem, longitudinal design. Coding, constant comparison → grounded theory, flexible design. Single case with deep detail → case study. “Pre‑registered” or “a priori hypotheses” → confirmatory research. “Post‑hoc” or “data‑driven” → exploratory research. --- 🗂️ Exam Traps Distractor: “Correlation implies causation.” – Remember the directionality limitation. Distractor: “All qualitative studies are flexible.” – Some qualitative work can follow a fixed protocol (e.g., structured interviews). Distractor: “A study with multiple measurements is always longitudinal.” – Cross‑sectional designs can also collect multiple variables at one time point. Distractor: “If a study uses random assignment, it must be experimental.” – Quasi‑experiments may use partial randomization but still lack full control. Distractor: “Exploratory research never reports p‑values.” – Exploratory studies may still report statistics but with a lower significance threshold. ---