6. What kind of mixed methods research is
PAC Analysis?
PAC分析はどのような種類の混合研究法か?
Takehiko Ito いとうたけひこ
Wako University, Tokyo 和光大学(東京)
0-17 (Proceedings page 69.)
www.itotakehiko.com www.pacanalysis.com
Mixed Methods International Research Association (MMIRA) Asia Regional
Conference 2017/ 3rd Japan Society for Mixed Methods Research (JSMMR)
Conference 2017
Ritsumeikan University
August 6, 2017
(国際混合研究法学会アジア地域会議/第3回日本混合研究法学会年次大会
2017年8月6日)
6
9. Purpose
? Personal Attitude Construct (PAC)
Analysis was proposed by Tetsuo Naito
(1993, 1997) as a method to seek for
the individual structure of attitude. 個
人別態度構造分析(内藤)
? The purpose of the present research is
to recharacterize PAC Analysis from the
perspective of mixed methods research
under the Methods-Strand Matrix
paradigm方法*ストランド行列表
proposed by Teddlie and Tashakkori
(2009).
9
10. What is PAC Analysis?
? Personal Attitude Construct
(PAC) Analysis個人別態度構造
分析 was proposed by Tetsuo
Naito (1993, 1997) 内藤哲雄
? as a method to seek the
individual structure of attitude
(Case study)事例研究法
? using visualization
(dendrogram:QUAN data)樹形
図による量的データの可視化
and a dialogical interview
(Narrative: QUAL data). 対話
的面接(語り:質的データ
10
11. Two unique research methods originated in Japan
often referred as qualitative research methods
(QUAL) using diagrams(QUAN/QUAL)
日本発信の図を用いた2つの質的研究法
? PAC Analysis
(個人別態度行動分析)
? Trajectory Equfinality Approach
(複線径路等至性モデル)
11
12. Characteristics of PAC Analysis
? Case study/single subject事例研究?1被験者
? QUAL > quan 質中心
? Sequential: QUAN -> QUAL 順次的: ->
? Successive: 継続的:
? Data conversion データ変換 QUAN -> QUAL
12
14. Phase ① QUAL (open-ended questions)
Procedure: By stimulus sentence (n=1),刺激文
Word association produced 連想法
Product: List of associated word/sentence連想項目
14
15. Phase ② quan (closed-ended questions)
Procedure: Paired-words rating of distance
一対比較法
Product: Distance Matrix距離行列
15
16. Phase ③ quan (multivariate analysis)多変量解析
Procedure: Cluster Analysis クラスター分析
Product: Dendrogram 樹形図
16
17. Phase ④ quan → QUAL (open-ended questions)
Procedure: Ask the meaning of each and higher clusters
Product: Dialogue narrative & cluster name (=meaning)
面接でのナラティブ生成によるクラスターの意味化と命名
17
18. Phase ⑤: Interpretation(open-ended questions)
Procedure: Cooperative dialogue to final interpretation
Product: cluster name (=meaning) &
fully explained dendrogram
高次クラスターの命名?意味化と全体の説明
18
21. Three Hypotheses: PAC Analysis can
be characterized as either
? Hypothesis (1):multiplied mixed methods
research 乗算的ミックス法? (Inoue & Ito, 2011)
or
? Hypothesis (2): explanatory sequential design
説明的順次デザイン? (Ito, et al., 2015)
or
? Hypothesis (3): (monostrand) conversion
design (単線型)変換型デザイン? (Teddlie &
Tashakkori, 2009)
21
22. (1)Characterizing PAC Analysis as
multiplied mixed methods research
(乗算的ミックス法) (Inoue & Ito,2011)
? Inoue and Ito (2011) argued that the PAC
Analysis is one of multiplied mixed
methods(乗算的ミックス法),
? while the mixed methods research is usually
referred to NOT multiplicative
(QUAL*QUAN:かけ算的)
? BUT additive combination of qualitative and
quantitative approaches
(QUAL+QUAN:足し算的). ※1+1=3
? ※ multiplied ≒ converted (QUAL?QUAN)
乗算≒データ変換 (質的?量的)
22
23. (2)Characterizing PAC Analysis
as Explanatory sequential design (2014-2015)
From Creswell and Fetters’ Workshop at MMIRA Conference in Boston College,
2014 to MMIRA in Ritsumeikan U, 2015
23
28. (3) Characterizing PAC Analysis
as monostrand or multustrand conversion
design 単線型or複線型変換型混合デザイン
? Seven sequential phases of PAC analysis procedure were
examined whether they have single strand or plural
strands in order to decide whether PAC Analysis is
monostrand conversion quasi-mixed design or multistrand
conversion mixed design. 単線型か複線型か
? A strand 小縄:研究のひとまとまり includes three stages:
1 conceptualization stage, 概念化段階 (=[問題?]目的)
2 experiential stage, データ収集/分析段階(=結果)
3 and inferential stage. 推論段階(=考察[?結論])
28
29. Data conversion データ変換
Data conversion occurs
when collected QUAN data are converted into narratives
or when QUAL data are converted into numbers.
? Quantitizing data: the process of converting QUAL data into
numbers that can be statistically analyzed.
Eg. associated words連想語 →distance matrix距離行列
→dendrogram 樹形図
? Qualitizing data: the process whereby QUAN data are
transformed into narrative data that can be analyzed
qualitatively.
Eg. dendrogram 樹形図→interview 面接→narratives 本人の語り
29
33. ※Text mining as an example of
monostrand conversion design
単線型変換型混合デザインとしてのテキストマイニング
? Quantitizing data: the process of converting
QUAL data(=narrative) into numbers that can
be statistically analyzed (=word count).
? Word frequency analysis単語頻度分析
? Word association analysis係り受け分析
? Word network analysis単語ネットワーク分析
? Positive/ Negative expression analysis 評判分析
33
35. Model 1: QUAN to QUAL monostrand
conversion case study method
単線型変換型混合事例研究法
(1) QUAN data collection: 量的データ生成
Phase 1: Word association語連想
Phase 2: Pair words distance rating距離の評定
Phase 3: Cluster analysis (Data 1)クラスター分析による樹形図
(2) Conversion from QUAN to QUAL: 質的データへ変換
Phase 4: Cluster interpretation クラスター解釈
Phase 5: Structural interpretation 全体構造の解釈
(Phase 6: Category table カテゴリー図の生成(Data 2))
(3) Inference 推論
Phase 7: Discussion/Conclusion考察?結論
35
36. Model 2: QUAN/QUAL multi(bi-)strand
conversion case study method
複線型変換型混合事例研究法
(1) QUAN strand: 量的データ生成
Phase 1: Word association語連想
Phase 2: Pair words distance rating距離の評定
Phase 3: Cluster analysis (Data 1)クラスター分析による樹形図
(2) QUAL strand: 質的データ生成
Phase 4: Cluster interpretation クラスター解釈
Phase 5: Structural interpretation 全体構造の解釈
Phase 6: Category table カテゴリー図の生成(Data 2)
(3) Meta-Inference (Integration) メタ推論
Phase 7: Discussion/Conclusion全体的考察?結論
36
37. Model 3:QUAL/QUAN/QUAL sequential
multi(tri-)strand conversion case study
複々線型変換型混合事例研究法
(1) QUAL strand: 質的データ生成
Phase 1: Word association(Data 1)語連想
(2) QUAN strand: 量的データ生成
Phase 2: Pair words distance rating距離の評定
Phase 3: Cluster analysis (Data 2)クラスター分析による樹形図
(3) QUAL strand: 質的データ生成
Phase 4: Cluster interpretation クラスター解釈
Phase 5: Structural interpretation 全体構造の解釈
Phase 6: Category table カテゴリー図の生成(Data 3)
(4) Meta-Inference (Integration) メタ推論
Phase 7: Discussion/Conclusion全体的考察?結論 37
38. Tentative Conclusion(1)
PAC Analysis can be characterized as either :
? (1) QUAN to QUAL monostrand conversion case study method
単線型変換型混合事例研究法
or
? (2) QUAN/QUAL multi(bi-)strand conversion case study method
複線型変換型混合事例研究法
or
? (3) QUAL/QUAN/QUAL multi(tri-)strand conversion case study
複々線型変換型混合事例研究法
? ※Depending on interpretation of unit of strand
38
39. Tentative Conclusion(2)
PAC Analysis can be characterized as both:
(1) QUAN to QUAL
monostrand conversion
case study method 単線
型変換型混合事例研究
法, if we skip Phase 6
(2) QUAN/QUAL multi(bi-
)strand conversion case
study method 複線型変
換型混合事例研究法 with
Phase 6 (Joint Display)
39