Qualitative Data Analysis: An introduction

1
Qualitative Data Analysis An introduction

• Carol Grbich
• Chapter 22. Incorporating data from multiple
  sources mixed methods

2
Mixed Methods

• Key points
• The advantages of combining quantitative and
  qualitative data are that you can maximize the
  impact of both
• For mixed methods to be successful, issues of
  sampling, design, data analysis and data
  presentation need careful attention
• Two ways of mixing methods regarding design are
  concurrent and sequential, but other new mixes
  are emerging
• Does qualitative data miss out in such a mix? and
  what is the next move?

3
A brief history of Qualitative and Quantitative
approaches

• The three stages of debate relating to
  quantitative and qualitative approaches
• 1. never the twain shall meet.
• 2. rapprochement
• 3. co-operation and mixing

4
What are the major differences between
qualitative and quantitative?

• Qualitative is an inductive approach
• Questions tend to be exploratory and open ended
  and data is often in narrative form.
• Reality is seen as a shifting feast, subjectivity
  is usually viewed as important
• P power is shared with the participants who are
  the experts on the matter under investigation.
• Analysis predominantly deals with meanings,
  descriptions, values and characteristics of
  people and things.
• The outcome sought is the development of
  explanatory concepts and models
• Widespread generalisation (apart from logical
  generalisation that is from similar instance to
  similar instance) is avoided.

5
What are the major differences between
qualitative and quantitative

• Quantitative is deductive
• Reality is seen as static and measurable
• Objectivity (distance, neutrality) and linearity
  (cause effect) may be sought
• Outcomes are pre specified , hypotheses l dictate
  questions and approach.
• Researcher control of the total process is
  paramount, precision and predictability are
  important
• Statistical approaches identify numbers and
  clarify relationships between variables.
• Theory testing is the key and generalisation and
  predictability the desired outcomes.
• Survey and experimental research are the main
  design options.
• Conclusions drawn follow logically from certain
  premises - usually rule based - which are
  themselves often viewed as proven, valid or
  true.

6
Advantages of combining quantitative and
qualitative results

• clarifying and answering more questions from
  different perspectives
• enhancing the validity of your findings
• increasing the capacity to cross check one data
  set against another
• Providing the detail of individual experiences
  behind the statistics.
• Helping in the development of particular
  measures
• Tracking stages over time.

7
Philosophical integration of qualitative and
quantitative approaches

• Two current options
• Pragmatism
• seeks ways through the polarised
  quantitative qualitative debate to find
  practical solutions to the problem of differing
  ideologies and methodologies recognising culture,
  context, individual experience, the constructed
  nature of reality, uncertainty, eclecticism,
  pluralism and the need for creative innovation of
  method.
• The Transformative paradigm
• multiple realities are shaped by knowledge is
  historically and socially situated
• issues of power between researcher and researched
  need to be explicitly addressed
• the incorporation of qualitative and quantitative
  methods are appropriate.
• the transformative ethical orientation
  comprises a strong human rights agenda
  within notions of beneficence and social justice

8
Conducting mixed methods research prior questions

• Is your research question one for which mixed
  methods would be the best approach?
• If so, which design would be the best?
• A mutual research design? involving acceptance
  that the two approaches come from completely
  different paradigms , celebrating their
  differences and keeping them separate within the
  design process the separate but together
  position?.
• Mixed methods?
• at which points will mixing occur? Design?
  Analysis? Interpretation?
• What sampling approaches will you utilise from
  the probability and non-probabily suite?
• How are you going to manage data analysis?   
• quantitizing - converting qualitative data into
  quantitative data or qualitizing - converting
  quantitative data into qualitative data
• To what degree will you qualitatively analyse
  quantitative data and vice versa?
• How are you going to display your results? -
  Separately? Integrated? consolidated?

9
Mixed method design

• Various forms of labeling and terminology have
  been used for mixed method design synergy,
  integration, triangulation, concurrent, parallel,
  merging, concurrent, sequential, exploratory and
  explanatory. Concurrent or sequential are the 2
  main options
• 1. Concurrent or parallel methods
• Here you would consider using multiple reference
  points where separate data sets are collected at
  the same time with the ultimate aim of merging
  the two data sets either,
• in a visual display such as a matrix
• by transforming the data (see quantitizing and
  qualititzing data in crossover/mixed analyses
  below) or
• in the final discussion.
• Design might involve using dual sites with the
  same sampling approach but with different data
  (quantitative and qualitative) then using the
  synthesised results to build up a complex
  picture.

10
Design 2. Sequential explanatory/exploratory

• You could undertake a qualitative study to
  explore a particular issue or phenomenon and
  using an iterative approach you could create
  hypotheses from these results which you could
  test using a survey or experimental design.
• Or, you could develop a short questionnaire
  survey to elicit key issues which can then be
  explained in depth using qualitative approaches
  of interviewing and observation. Synthesis of the
  two sets of results is needed to clarify the dual
  outcomes and to utilise the increased validity
  these two approaches provide.

11
A typical sequencing design

• Stage 1 Representative survey of the population
• Stage 2 Exploratory qualitative interviews or
  focus groups to tease out the findings of the
  survey
• Stage 3 Hypotheses generated from stages 1 and 2
  are tested in various interventions which are
  then evaluated
• Stage 4 Participatory action research where the
  participants take control of the development,
  implementation and evaluation of the most
  successful of these interventions.

12
Issues to consider in attempting to combine data
sets

• You need to be familiar with both quantitative
  and qualitative approaches
• Mixing of paradigms, data collection, analysis
  and interpretation, takes time and skills to do
  well
• Combined designs are more expensive than single
  designs
• Are there benefits to converting qualitative to
  quantitative data?

13
Crossover/mixed analysis

• Suggestions
• reduce dimensionality of either data set
  (quantifying to basics)
• integrate data display (visual presentation of
  both sets as one)
• transform data (Qual to quantb(numerical codes)
  and quant to qual (themes) for analysis)
• correlate data (correlate results from
  quantitizing and qualitizing)
• consolidate data (merging multiple data sets to
  create new codes, variables etc)
• compare data (compare findings)
• integrate data (into one or two sets of data)
• use warrented assertion analysis (seeking
  meta-inferences from both sets)
• import data (using follow-up findings from
  qualitative to inform quantitative analysis and
  vice versa)

14
Presentation of dual results

• Separate data sets
• Requires a very large results section and
  requires regular summaries of data findings which
  will need to culminate in a final drawing
  together of the findings so that the reader can
  make sense of the diversity presented.
• Combined data sets
• Amalgamate the findings in such a way that a neat
  display of graphical information occurs, followed
  by a few carefully chosen qualitative quotes to
  display the homogeneity (or diversity) of the
  data gathered. Matrixes can bring together
  variables, themes and cases as can lists,
  network diagrams and graphical displays.
• Multiple data sets
• Currently the majority of data collected is still
  within the survey/interview/observation/document
  analysis framework with the documents
  traditionally being written communications
  displayed in a variety of creative ways.