Mixed Methodology

1
Mixed Methodology

• Choosing an appropriate research design
• Dr. Victor Lofgreen
• Walden University
• Atlanta Residency, Nov 06

2
Logical Positivism

• Ontology (Nature of reality) There is a single
  reality.
• Epistemology (relationship of the knower to the
  known) The knower and the known are independent
• Axiology (role of values in inquiry) Inquiry is
  value free
• Generalizations Time and context free
  generalizations are possible.
• Causal Linkages There are real causes that are
  temporally precedent or simultaneous with
  effects.
• Deductive Logical Emphasis on arguing from the
  general to the specific, or a particular emphasis
  on a priori hypotheses testing (or theory.)

3
Research Paradigms

• Logical Positivism
• Constructivism

4
Conflict in Paradigms

• Two approaches
• Positivist/empiricist
• Constructivist/phenomenological

5
Paradigm to Methods

• Positivist paradigm
• Quantitative Methods
• Constructivist paradigm
• Qualitative Methods

6
Constructivist

• Ontology (Nature of reality. There are multiple,
  constructed realities.
• Epistemology (relationship of the knower to the
  known) The knower and the known are inseparable.
• Axiology (role of values in inquiry) Inquiry is
  value-bound
• Generalizations Time and context free
  generalizations are not possible.
• Causal Linkages It is impossible to distinguish
  causes from effects
• Inductive Logic Emphasis on arguing from the
  particular to the general, there is emphasis on
  grounded Theory.

7
Post Positivist Position

• Value-ladenness of inquiry Research is
  influenced by the values of investigators.
• Theory-ladenness of the facts Research is
  influenced by the theory or hypothesis or
  framework that the researcher uses.
• Nature of reality Our understanding of reality
  is constructed.

8
The Evolution of Methodological Approaches

• Period 1 The Monomethod or Purist Era
• The purely Quantitative Orientation
• Single Data Source (QUAN)
• Within one paradigm/Model, multiple data sources
• Sequential (QUAN/QUAN)
• Parallel Simultaneous (QUANQUAN)
• The Purely Qualitative Orientation
• Single Source (QUAL)
• Within one paradigm/Model, Multiple Data Sources
• Sequential (QUAL/QUAL)
• Parallel/Simultaneous (QUALQUAL)

9
ControversyOntology and Causality

• Naïve Realism Objective External Reality
• Critical Realism Objective Reality known
  approximately or probabilistically.
• Transcendental Realism Social phenomena exist
  in an objective world. There are some stable
  lawful relationships.
• Ontological Relativism There are multiple social
  realities that are parts of human intellect and
  that may change as their constructors change.

10
Causal Relationships from Ontological
Distinctions

• Post positivists believe in the proportional view
  of the truth
• Pragmatists believe there may be causal
  relationships but we may never be able to pin
  them down.
• Constructivists believe that all entities are
  simultaneously shaping each other

11
Pragmatism and the Choice of Strategy

• Pragmatists consider the research question to be
  more important than the either the method or the
  world view that is supposed to underlie the
  method.

12
Research Cycle
Inductive Reasoning
Deductive Reasoning
13
Paradigm Comparison
14
Mixed Model Designs

• Combined the qualitative and quantitative
  approaches in different phases of the research
  process.

15
Five Mixed Method Designs

• Sequential Studies, (Two Phase)
• Parallel /Simultaneous
• Equivalent Status Designs
• Dominant Less Dominant Studies
• Multilevel Designs (Levels of Aggregation)

16
MAXMINCON Principle

• Maximize the experimental variance to allow
  enough difference between groups to allow the
  effect to occur.
• Minimize the error variance provides power for
  detecting the difference between groups. Take
  out the noise to better detect the signal. Error
  variance comes from random fluctuations, in
  reactions, behaviors, an/or measurements.
• Control of extraneous variables remove all
  competing variables

17
Triangulation Techniques

• Data Triangulation
• Investigator Triangulation
• Theory Triangulation
• Methodological Triangulation

18
Taxonomy of Data Collection
19
Period 2 The Emergence of Mixed Models

• Equivalent Status Designs (Across both
  Paradigms/Models)
• Sequential (i.e. two-phase sequential studies)
• (QUAL/QUAN)
• (QUAN/QUAL)

20
Period 2 The Emergence of Mixed Models

• Parallel / Simultaneous
• QUALQUAN
• QUANQUAL

21
Period 2 The Emergence of Mixed Models

• Dominant Less Dominant (Across both
  paradigms/models)
• Sequential
• QUAL/quan
• QUAN/qual
• Parallel /Simultaneous
• QUAL/quan
• QUAN/qual
• Designs with Multi level Approaches

22
Period 3 The Emergence of Mixed Methods

• Multiple Applications
• The Type of Inquiry QUAL or QUAN
• Data Collection/Operations QUAL or QUAN
• Analysis of Inference QUAL or QUAN

23
Period 3 The Emergence of Mixed Methods

• Single Application within a Stage of Study
• Type of Inquiry QUAL or QUAN
• Data Collection/Operations QUAL or QUAN
• Analysis of Inference QUAL or QUAN

24
Prototypes

• Quan___________________Qual
• Experiment Case Study

25
Classification of Methods

• Type of investigation
• Type of Data Collection
• Type of analysis or inference

26
Type of Investigation

• Confirmatory
• Exploratory

27
Type of Data Collection

• Qualitative
• Quantitive
• Dimension or Stage of Research

28
Type of Analysis or Inference

• Qualitative
• Statistical

29
Confirmatory Investigation
Quantitative Quantitative Qualitative Qualitative
Statistical Analysis and Inference Qualitative Analysis Inference Statistical Analysis and Inference Qualitative Analysis Inference
Pure Quan Mixed Type V (Rare) Mixed Type I Mixed Type II
30
Pure Quantitative

• Data are Quantitative
• Analysis is Quantitative
• Based on a priori theory or hypothesis

31
Type 1 Confirmatory

• Collect Qualitative Data
• Data are quantified
• Data are subjected to statistical analysis

32
Type II Confirmatory

• Begins with a priori theory or hypothesis
• Qualitative data Interviews / Observations
• Data are analyzed in qualitative form

33
Type V Confirmatory

• Data are Quantitative
• Data are reclassified into qualitative form
• Data are analyzed to generate profiles and
  categories.
• The results are then used for further research

34
Exploratory Research
Quantitative Quantitative Qualitative Qualitative
Statistical Analysis and Inference Qualitative Analysis and Inference Statistical Analysis and Inference Qualitative Analysis and Inference
Mixed Type III Mixed Type VI (Rare) Mixed Type IV Pure Qual
35
Type III Exploratory

• Data are quantitative
• No a priori theory of hypothesis
• Data are statistically analyzed
• Traditional quantitative exploratory study

36
Type IV Exploratory

• Data are Qualitative
• Sentence Completion
• Story telling
• Data are converted to Quantitive form
• Data are subjected to statistical analysis
• Nonparametric
• Log linear modeling
• Logistic regression

37
Type VI Exploratory

• Data are Quantitative
• Data converted to Qualitative
• Profiles or Group Identities
• Data Analyzed as Qualitative
• Results are used to build models or determine
  prototypes

38
Pure Qualitative

• Data Qualitative
• Data Analysis Qualitative
• No A Priori Theory or Hypothesis

39
Multiple Application Designs

• Parallel Mixed Models
• Sequential Mixed Models

40
Mixed Model Features

• Mix both research hypothesis and research
  questions
• Mixed data collection
• Mixed Data Analysis

41
Type VII Parallel Mixed Model

• At least one stage of the research includes qual
  and quan data
• The data are collected and analyzed independently

42
Data Analysis

• Descriptive Methods
• Inferential Methods
• Univariate vs. Multivariate

43
Descriptive Measures

• Measures of Central Tendency
• Mean
• Mode
• Median

44
Descriptive Measures

• Measures of Variability
• Average deviation
• Variance
• Standard Deviation
• Interquartile Range

45
Descriptive Measures

• Measure of Relative Standing
• Percentile Rank

46
Quantitative Data Analysis

• Data Analysis Matrix

Type of Data Relationship between variables Differences between Groups
Interval / Ordinal Pearson r correlation Multiple Regression Canonical Correlation Regression Analysis Factor Analysis T-test for Ind. Samples ANOVA / ANCOVA MANOVA /MANCOVA Discriminate Analysis
Ordinal / Nominal Rho Chi-Square Phi Cramers V Logistic Regression Sign Test Wilcoxon matched pairs
47
Type VIII Sequential Mixed Model

• Data are collected in phases
• Each phase emphasizes one type of data
• Data are analyzed and results support the next
  phase
• Final results include variety of results

48
Inferential MethodsTests difference between
group means

• Compare a group mean with a population mean Z
  score
• Compare the means of two samples
• Independent observation T Test
• Non-independent T Test for Non-independent
  measures

49
Inferential Methods Cont.

• Compare the means of two or more samples
• Compare more than one variable (factorial
  analysis)
• ANOVA Analysis of Variance

50
Inferential Cont

• Comparing means of two or more samples while
  controlling for an extraneous variable
• ANCOVA Analysis of covariance

51
Inferential methods Cont.

• Correlation Coefficients not 0
• T-Test for significance of Pearsons r
• F-Test for significance of multiple correlation
• T-test or F-test for significance of slope in
  multiple regression analysis

52
Measures of Association

• Pearsons R correlation
• Chi Square test of Independence

53
Multivariate Methods

• Multiple Regression
• Several Independent Variables compared to a
  single dependent variable
• Canonical Correlation
• Several independent variables compared to several
  dependent variables.

54
Multivariate Analysis Cont

• Discriminate Function Analysis
• To find a set of variables that differentiate two
  or more groups
• Factor Analysis
• Explanatory to find underlying constructs of a
  set of variables
• Confirmatory find the predicted construct of a
  set of variables

55
Qualitative Data Analysis

• Qualitative Typology Matrix

Type of Theme More Simple Schemes More Complex Schemes
A priori Simple valence analysis Effects matrices (Miles Huberman, 1994)
Emerging Latent content analysis Constant comparative Analysis (Glaser Strauss, 967 Lincoln Gub, 1985) Developmental research sequence (Spradley, 1979,1980)
56
Mixed Data Analysis Strategies

• Data Transformation
• conversion from one type to another
• Typology development
• Analysis of one type yields a typology that is
  use by the other method.
• Extreme-case analysis
• Identify extreme cases for data collection from
  another method
• Data consolidation/merging
• Joint review to create or consolidate variables
  or data sets for further analysis

57
The End

• Victor Lofgreen, PhD
• Walden University
• Atlanta Residency
• Nov. 2006
• Reference
• Tashakkori, A Teddlie, C, 1998, Mixed
  Methodology Combining Qualitative and
  Quantitative Approaches, Thousand Oaks, CA, Sage
  Publications