grad-mixed-methods
用聚斂式或序列式設計做混合方法研究，並真正整合質性與量化
中文翻譯
--- name: "grad-mixed-methods" description: "用聚斂式、解釋型序列或探索型序列策略，設計並進行混合方法研究，讓質性與量化的部分真正整合。當使用者需要選擇混合方法的設計、在設計、方法或詮釋層級整合質性與量化資料、以實用主義為依據說明混合的理由，或問到「該用哪種混合方法設計」「怎麼整合質性和量化的發現」「質性量化都做了就算混合方法嗎」時，就用這個 skill。" metadata: category: "WP-32 質性方法" tags: ["mixed-methods", "convergent-design", "explanatory-sequential", "exploratory-sequential", "integration", "pragmatism", "Creswell", "qual-quant"] --- # Mixed Methods Research ## Overview 混合方法研究在單一研究或一整組研究計畫中，結合質性與量化取徑，發揮兩者的優勢。它以實用主義為基礎，依哪種方法最適合研究問題來選擇方法。它的決定性特徵不是單純兩種都用，而是在設計、方法或詮釋層級真正把兩者整合起來，產出任何一種取徑單獨都得不到的洞察。 ## When to Use - 單一取徑（只用質性或只用量化）無法充分回答研究問題 - 量化結果需要質性的解釋（為什麼會出現這個效果？） - 質性發現需要量化的檢驗（這個模式能不能推廣？） - 複雜的現象需要廣度（量化）和深度（質性）兩種理解 ## When NOT to Use - 研究問題用一種取徑就能完整回答 - 研究者在質性或量化方法上缺乏能力 - 資源（時間、經費、團隊）無法充分支撐兩個部分 - 對這項特定研究，質性與量化的典範假設無法調和 ## Assumptions ``` IRON LAW: Mixed methods requires GENUINE INTEGRATION — running qual and quant in parallel without connecting findings is NOT mixed methods, it is two separate studies stapled together. Integration must occur at design, methods, or interpretation level. ``` 關鍵假設： 1. 以實用主義作為哲學基礎，對研究問題有用的，就決定方法 2. 質性與量化資料都有正當的知識主張 3. 整合是決定性特徵，不是單純合併，而是連結、融合或嵌入 4. 研究問題驅動設計的選擇，不是方法論上的門派忠誠 ## Methodology ### Step 1: Select the Mixed Methods Design | Design | Structure | Purpose | |--------|-----------|---------| | **Convergent** | QUAL + QUANT 同時進行 | 比較並融合發現，以求完整性或驗證 | | **Explanatory Sequential** | QUANT → qual | 用質性來解釋、闡述或提供量化結果的脈絡 | | **Exploratory Sequential** | QUAL → quant | 用質性發展量化要檢驗的工具、變數或類型學 | | **Embedded** | QUANT 內嵌 qual（或反過來） | 在更大的設計中，一個部分支援另一個 | 用大寫表示主導的部分，小寫表示輔助的部分。 ### Step 2: Implement Each Strand with Rigor 對每個部分各自套用完整的方法嚴謹度。質性部分遵循質性的品質標準（可信性、可轉移性）。量化部分遵循量化的標準（效度、信度）。不要為了遷就一方而犧牲另一方。 ### Step 3: Integrate the Strands 各層級的整合策略： | Level | Strategy | Example | |-------|----------|---------| | **Design** | 把一個部分嵌入另一個 | RCT 裡的質性訪談 | | **Methods** | 由一個部分建構另一個 | 質性主題變成問卷題項 | | **Interpretation** | 聯合呈現、融合或敘事編織 | 並列比較表 | ### Step 4: Draw Meta-Inferences 把兩個部分的發現綜合成後設推論，超越任何一個部分單獨能產出的。處理兩個部分之間的收斂、互補或分歧。 ## Output Format ```markdown ## Mixed Methods Analysis: [Context] ### Design - Type: [convergent / explanatory sequential / exploratory sequential / embedded] - Priority: [QUAL+QUANT / QUANT→qual / QUAL→quant] - Rationale: [why this design fits the research question] ### Quantitative Strand - Method: [survey / experiment / secondary data] - Sample: [N, sampling strategy] - Key findings: [statistical results] ### Qualitative Strand - Method: [interviews / focus groups / observations] - Sample: [N, sampling strategy] - Key findings: [themes or categories] ### Integration (Joint Display) | Quantitative Finding | Qualitative Finding | Meta-Inference | |---------------------|--------------------|--------------------| | [statistical result] | [theme/quote] | [integrated insight] | ### Convergence Assessment - Confirmed: [where qual and quant agree] - Complementary: [where one strand adds to the other] - Divergent: [where findings conflict — and how resolved] ### Meta-Inferences 1. [Integrated conclusion that neither strand alone could produce] 2. [Integrated conclusion that neither strand alone could produce] ``` ## Gotchas - 聯合呈現表是展示整合的黃金標準，如果你做不出一張，整合可能不存在 - 不要偏重某一個部分，除非設計明確要求（例如以 QUANT 為主的解釋型序列） - 混合典範需要哲學上的正當理由，實用主義很常見，但不是唯一選項（辯證多元論是另一個） - 解釋型序列需要先完成 QUANT 階段，才能設計 qual 階段，你不能一開始就把兩者都設計好 - 兩個部分的樣本數不同：量化樣本依檢定力分析，質性樣本依飽和或資訊豐富度 - 審查者常批評「準混合」的研究，也就是兩個部分根本沒有連起來，所以要把整合明確表現出來 ## References - Creswell, J. W., & Plano Clark, V. L. (2018). *Designing and Conducting Mixed Methods Research* (3rd ed.). Sage. - Fetters, M. D., Curry, L. A., & Creswell, J. W. (2013). Achieving integration in mixed methods designs. *Health Services Research*, 48(6pt2), 2134-2156. - Teddlie, C., & Tashakkori, A. (2009). *Foundations of Mixed Methods Research*. Sage.
英文原文
--- name: "grad-mixed-methods" description: "Design and conduct mixed methods research using convergent, explanatory sequential, or exploratory sequential strategies with genuine integration of qualitative and quantitative strands. Use this skill when the user needs to choose a mixed methods design, integrate qualitative and quantitative data at design, methods, or interpretation levels, justify mixing on pragmatist grounds, or when they ask 'which mixed methods design should I use', 'how do I integrate qual and quant findings', or 'is running both qual and quant enough to be mixed methods'." metadata: category: "WP-32 質性方法" tags: ["mixed-methods", "convergent-design", "explanatory-sequential", "exploratory-sequential", "integration", "pragmatism", "Creswell", "qual-quant"] --- # Mixed Methods Research ## Overview Mixed methods research combines qualitative and quantitative approaches within a single study or program of inquiry to leverage the strengths of both. Grounded in pragmatism, it selects methods based on what works best for the research question. The defining feature is not merely using both approaches but genuinely integrating them at design, methods, or interpretation levels to produce insights neither approach could achieve alone. ## When to Use - A single approach (qual or quant alone) cannot adequately address the research question - Quantitative results need qualitative explanation (why did the effect occur?) - Qualitative findings need quantitative testing (does the pattern generalize?) - Complex phenomena require both breadth (quant) and depth (qual) of understanding ## When NOT to Use - When the research question can be fully addressed by one approach - When the researcher lacks competence in either qualitative or quantitative methods - When resources (time, funding, team) cannot support both strands adequately - When the paradigmatic assumptions of qual and quant are irreconcilable for the specific study ## Assumptions ``` IRON LAW: Mixed methods requires GENUINE INTEGRATION — running qual and quant in parallel without connecting findings is NOT mixed methods, it is two separate studies stapled together. Integration must occur at design, methods, or interpretation level. ``` Key assumptions: 1. Pragmatism as the philosophical foundation — what works for the research question determines the method 2. Both qualitative and quantitative data have legitimate claims to knowledge 3. Integration is the defining feature — not merely combining, but connecting, merging, or embedding 4. The research question drives design choice, not methodological allegiance ## Methodology ### Step 1: Select the Mixed Methods Design | Design | Structure | Purpose | |--------|-----------|---------| | **Convergent** | QUAL + QUANT simultaneously | Compare and merge findings for completeness or validation | | **Explanatory Sequential** | QUANT → qual | Use qual to explain, elaborate, or contextualize quant results | | **Exploratory Sequential** | QUAL → quant | Use qual to develop instruments, variables, or typologies tested by quant | | **Embedded** | qual within QUANT (or vice versa) | One strand supports the other within a larger design | Use uppercase to indicate the dominant strand; lowercase for the supporting strand. ### Step 2: Implement Each Strand with Rigor Apply full methodological rigor to each strand independently. Qualitative strand follows qualitative quality criteria (credibility, transferability). Quantitative strand follows quantitative criteria (validity, reliability). Do not compromise one strand for the other. ### Step 3: Integrate the Strands Integration strategies by level: | Level | Strategy | Example | |-------|----------|---------| | **Design** | Embedding one strand within the other | Qual interviews within an RCT | | **Methods** | Building one strand from the other | Qual themes become survey items | | **Interpretation** | Joint display, merging, or narrative weaving | Side-by-side comparison table | ### Step 4: Draw Meta-Inferences Synthesize findings from both strands into meta-inferences that transcend what either strand alone could produce. Address convergence, complementarity, or divergence between strands. ## Output Format ```markdown ## Mixed Methods Analysis: [Context] ### Design - Type: [convergent / explanatory sequential / exploratory sequential / embedded] - Priority: [QUAL+QUANT / QUANT→qual / QUAL→quant] - Rationale: [why this design fits the research question] ### Quantitative Strand - Method: [survey / experiment / secondary data] - Sample: [N, sampling strategy] - Key findings: [statistical results] ### Qualitative Strand - Method: [interviews / focus groups / observations] - Sample: [N, sampling strategy] - Key findings: [themes or categories] ### Integration (Joint Display) | Quantitative Finding | Qualitative Finding | Meta-Inference | |---------------------|--------------------|--------------------| | [statistical result] | [theme/quote] | [integrated insight] | ### Convergence Assessment - Confirmed: [where qual and quant agree] - Complementary: [where one strand adds to the other] - Divergent: [where findings conflict — and how resolved] ### Meta-Inferences 1. [Integrated conclusion that neither strand alone could produce] 2. [Integrated conclusion that neither strand alone could produce] ``` ## Gotchas - A joint display table is the gold standard for demonstrating integration — if you cannot produce one, integration may be absent - Do NOT privilege one strand over the other unless the design explicitly calls for it (e.g., QUANT-dominant explanatory sequential) - Mixing paradigms requires philosophical justification — pragmatism is common but not the only option (dialectical pluralism is another) - Explanatory sequential requires the QUANT phase to be complete before designing the qual phase — you cannot design both at the start - Sample sizes differ between strands: the quant sample follows power analysis, the qual sample follows saturation or information richness - Reviewers often critique "quasi-mixed" studies where the two strands never actually connect — make integration explicit ## References - Creswell, J. W., & Plano Clark, V. L. (2018). *Designing and Conducting Mixed Methods Research* (3rd ed.). Sage. - Fetters, M. D., Curry, L. A., & Creswell, J. W. (2013). Achieving integration in mixed methods designs. *Health Services Research*, 48(6pt2), 2134-2156. - Teddlie, C., & Tashakkori, A. (2009). *Foundations of Mixed Methods Research*. Sage.