Research methodology decision tree for any study design

# Research Methodology Decision Tree **Research Question:** "How does remote work frequency affect employee burnout and productivity in tech companies?" **Discipline:** Organizational Psychology --- ## Decision Point 1: What is the nature of your research question? **Ask yourself:** Am I exploring, describing, or testing a causal relationship? | Option | When Appropriate | Your Fit | |--------|-----------------|----------| | Exploratory | New/understudied phenomenon, generating theory | ❌ Remote work burnout is well-studied enough for hypotheses | | Descriptive | Documenting prevalence, characteristics, patterns | ⚠️ Partially — you want to know current burnout levels | | Explanatory/Causal | Testing whether X causes or predicts Y | ✅ **Best fit** — you want to know if remote work *affects* burnout/productivity | **Decision:** Explanatory design — you have a clear IV (remote work frequency) and DVs (burnout, productivity). --- ## Decision Point 2: Can you manipulate the independent variable? **Ask yourself:** Can I randomly assign participants to different remote work schedules? | Option | Trade-offs | |--------|-----------| | True experiment (RCT) | **Gold standard for causality.** But: requires organizational buy-in to change employees' schedules, expensive, ethical concerns about forcing work arrangements, likely 6+ months | | Quasi-experiment | **Strong causality without full randomization.** Can leverage natural variation (e.g., companies with different policies). Threats: selection bias, confounds | | Non-experimental (observational) | **Most feasible.** Survey existing workers about their current arrangements. Threats: self-selection, reverse causality, unmeasured confounds | **Decision:** Quasi-experimental design is your sweet spot. True experiments are impractical (you cannot randomly assign employees to hybrid schedules without employer participation). But pure observational loses too much causal leverage. **Recommended approach:** Cross-sectional survey with natural groups (fully remote, hybrid 2-3 days, hybrid 4 days, fully in-office) + propensity score matching to reduce selection bias. --- ## Decision Point 3: Cross-sectional or longitudinal? **Ask yourself:** Do I need to track changes over time, or is a snapshot sufficient? | Option | Pros | Cons | Time/Cost | |--------|------|------|-----------| | Cross-sectional | Fast, cheaper, larger sample possible | Cannot establish temporal precedence, snapshot bias | 2-3 months, $2-5K | | Longitudinal (panel) | Tracks changes, stronger causal inference | Attrition, practice effects, expensive, slow | 12-18 months, $15-30K | | Accelerated longitudinal | Combines cohorts at different stages | Complex analysis, requires cohort equivalence | 6-9 months, $8-15K | **Decision:** For a dissertation or funded study, a **two-wave panel design** (baseline + 6-month follow-up) offers the best trade-off. For an initial study or thesis, cross-sectional with statistical controls is defensible. --- ## Decision Point 4: Quantitative, qualitative, or mixed? **Ask yourself:** Do I need generalizable patterns, deep understanding, or both? | Option | When to Choose | Your Fit | |--------|---------------|----------| | Quantitative only | Testing specific hypotheses with validated instruments | ✅ Burnout (MBI) and productivity have established scales | | Qualitative only | Understanding lived experience, generating theory | ❌ You have specific variables to measure | | Mixed methods — explanatory sequential | Quant first, then qual to explain unexpected findings | ✅✅ **Best fit** — survey for patterns, then interviews to understand mechanisms | **Decision:** Explanatory sequential mixed methods. Phase 1: quantitative survey (n=400+). Phase 2: semi-structured interviews (n=15-20) with purposively sampled participants from interesting subgroups. --- ## Decision Point 5: Sampling strategy? **Recommendation:** Stratified purposive sampling across company sizes and industries within tech. Recruit through professional networks (LinkedIn, Blind app), company HR partnerships, and tech community Slack groups. Target n=500 for survey (anticipating 80% completion = 400 usable responses). --- ## Recommended Final Design **Mixed-methods explanatory sequential quasi-experiment** - Phase 1: Online survey of 500 tech workers across 4 natural remote-work groups - Instruments: Maslach Burnout Inventory (MBI-GS), Individual Work Performance Questionnaire (IWPQ), custom remote work frequency measure, demographic and control variables - Analysis: MANCOVA with propensity score matching, SEM for path analysis - Phase 2: 20 semi-structured interviews (5 per group), thematic analysis - Timeline: 8-10 months total - Estimated cost: $5-8K (incentives, software, transcription) --- ## Draft Methods Section > This study employed an explanatory sequential mixed-methods design (Creswell & Plano Clark, 2018) to investigate the relationship between remote work frequency and employee burnout and productivity in the technology sector. In the quantitative phase, a cross-sectional online survey was administered to approximately 500 technology professionals recruited through stratified purposive sampling across four natural groups based on remote work frequency: fully in-office (0 days remote/week), low hybrid (1-2 days), high hybrid (3-4 days), and fully remote (5 days). Burnout was measured using the Maslach Burnout Inventory — General Survey (MBI-GS; Schaufeli et al., 1996), and individual work performance was assessed using the Individual Work Performance Questionnaire (IWPQ; Koopmans et al., 2014). Propensity score matching was used to reduce selection bias across groups, controlling for age, tenure, job level, caregiving responsibilities, and pre-existing mental health conditions. MANCOVA and structural equation modeling (SEM) were conducted to test hypothesized relationships. In the qualitative phase, 20 participants were purposively selected from the quantitative sample to represent diverse experiences within each group. Semi-structured interviews lasting 45-60 minutes explored mechanisms underlying observed patterns. Interviews were transcribed verbatim and analyzed using reflexive thematic analysis (Braun & Clarke, 2019).