Kerttuli Visuri

1
Analysis and Interpretation of Qualitative
Data 26.11.2001, VeTO, SEMS Sarcous

• Kerttuli Visuri Jarno Vähäniitty

2
Topics of this presentation

• Some words on qualitative research
• Data analysis phases and terminology
• Preparing for analysis
• Analysis
• Techniques for analysing qualitative data
• Differences of single (within-case) and
  cross-case analyses
• Tool support for data analysis
• Interpretation
• drawing and verifying conlusions
• confirming the findings
• Summary

3
Qualitative research basic terminology and
concepts

• The aim of the analysis is to understand the
  research phenomenon from the viewpoint of the
  research subject
• Prerequisites
• Knowledge of the existing literature/research of
  the selected research field
• Awareness of the theoretical framework, to which
  the current research is going to be clung to
• Progression of a qualitative reseach process
• The research problem may change during the
  research process
• Typical data
• Derived from interviews
• Written documents and specifications, magazines,
  agreements, video tapes, etc.

4
Phases of Qualitative Research and Our Focus

• Data collection
• Data analysis and interpretation
• Documenting and reporting

5
Steps in Analysis and Interpretation

• Design the data analysis and interpretation phase
• Data reduction
• Data display
• Explore and describe
• Explain and predict
• Interpret and draw conlusions
• Verify the findings

6
Designing the Analysis Phase

• Issues to bear in mind while designing the data
  analysis phase
• What type of data do you have?
• Qualitative, quantitative or both?
• How to link qualitative data to quantitative?
• Management issues
• staffing and scheduling?
• other study participants verifying the findings
  who and when?
• data mangement data storage, data analysis
  techniques?
• possibilities for computer use in order to
  facilitate data analysis?
• Reserve time and resources for data analysis
• It is the BIGGEST TASK in qualitative research
  projects!

7
Analysis Data Reduction

• Selecting, focusing, simplifying, abstracting and
  transforming the data that appear in written-up
  field notes or transcriptions
• Goal Organise the data in such a way that
  final conclusions can be drawn

8
Displaying Reduced Data

• Data Display organised, compressed assembly of
  information that permits conclusion drawing and
  action
• You know what you display
• Two major approaches for displaying reduced
  data
• matrices
• networks
• Displays may sort to data according to
• chronological sequence (flow) of events,
  happenings and processes
• role-ordered positions of the participating
  personnel
• conceptual dependences (variables and their
  interaction)
• Different display types suited to different
  analysis problems
• Also, linked to various tactics for drawing and
  confirming conclusions

9
Analysis Exploring and describing

• What, where and when?
• Making complicated things understandable by
  showing how their parts fit together according to
  some rules
• Plausible reasons for why things are happening as
  they are
• Objectives
• Compress and display the data in order to permit
  drawing conclusions and
• Guard against the overload and potential for bias
  that appear when analysing unreduced data

10
Data Displays for Exploring Describing Purposes

• Partially ordered displays
• Uncover and describe what is happening in a
  setting, no matter how how messy or surprising
• Example Context chart
• Shows relationships between the roles and groups
  that make up the context
• Summarises first understandings and locates
  questions for next-step data collection
• Time-ordered displays
• For understanding flow and sequence of events and
  processes
• Example Event listing
• Arranges a series of events by time periods and
  sorts them into categories
• For understanding extended processes
• Role-ordered displays
• Sort people according to their position-related
  experiences
• Conceptually ordered displays
• Emphasise well-defined variables and their
  interaction

11
Analysis Explaining and Predicting

• Why and how?
• Aim to allow the researchers to see the
  underlying mechanisms of influences
• Two suggested approaches
• variable-oriented (conceptual approach)
• process oriented (storylike approach)

12
Data Displays for Explanation Predicting
Purposes

• Explanatory effects matrix
• First step towards answering why things happened
  the way they did
• Looks at outcomes or results of a process
• Case dynamics matrix
• Displays a set of forces and traces the outcomes
• A way of seeing what leads to what
• Causal networks
• Display of the most important variables and their
  relationships
• Pulling together independent and dependent
  variables and their relationships into a coherent
  picture
• Straight predictions
• Inferences that the researcher makes about the
  probable evolution of case events or outcomes for
  the future
• Ultimate test of explanatory power

13
Within-case and cross-case analysis differences
and similarities (1/2)

• Within-case analysis
• one in-depth analysis per one case may include
  various viewpoints
• Cross-case analysis
• looking at several cases one after another in
  order the gain a bigger picture of the research
  phenomenon
• The aim of cross-case analyses is to derive good
  explanations and better theories by looking at
  multiple cases instead of only one
• Increases generalisability through deepened
  understanding of the research phenomenon
• Summarizing the themes is not enough -gt the
  generalization has to be done across the variable
  and process factors
• firstly, individually in each case in order to
  gain an in-depth analysis of each case
• are the variables/processes similar in each case?
• if not, how do they differ from each other in
  each of the cases?
• Generalisation possible based on a careful
  analysis of each case

14
Within-case and cross-case analysis differences
and similarities (2/2)

• Some suggestions for how to do generalizations
• avoid aggregating or smoothing
• keep the local case configuration (basic
  conditions) intact
• join the variable- and process-oriented
  approaches
• cases can often be sorted into explanatory groups
  or families sharing common scenarions
• However
• Deviating cases are at least as important as
  those that fit nicely
• Dont try to fit the case in by force but strive
  to understand why a certain case deviates from
  the common stream
• These findings can support your theory, too
• Some suggested techniques for exploring and
  describing the cross-case data
• partially ordered matrices
• conceptually ordered matrices
• case-ordered presentations
• time-ordered matrices/presentations

15
Tool Support for Analysis

• Preparing data for analysis
• Data annotation / memoing
• Data coding / classification
• Analysis
• Data linking
• Search and retrieval
• Data display
• Graphics editing
• Conceptual / theory development
• Example
• find all data referring to requirements
  management

16
Conclusion Drawing and Verification

• People make quickly sense of the most chaotic
  events
• We keep our world consistent and predictable by
  organising and interpreting it
• But, are the meanings found right, valid or
  repeatable?
• Qualitative analyses can be evocative,
  illuminating, masterful and wrong
• Coming up tactics for
• Generating meaning
• Testing and confirming meanings
• Also, look at Hubermann Miles for a series of
  questions for the researcher to ask himself when
  assessing the quality of a study

17
Tactics for Generating Meaning

• Whats going on?
• Noting patterns and themes
• Seeing plausibility (or, lack of it)
• Clustering
• Making metaphors
• Counting
• Sharpening the understanding
• Making contrasts and comparisons
• Differentiation
• Partitioning variables

• Abstracting
• Subsuming particulars into the general
• Factoring
• Noting relations between variables
• Finding intervening variables
• Establishing understanding
• Building a logical chain of evidence
• Making conceptual / theoretical coherence

18
Tactics for Testing and Confirming Meanings Found

• Assessing quality of the data
• Checking for
• Representativeness
• Researcher effects
• Triangulating (across data sources and /or
  methods)
• Weighting the evidence
• Saying what the found pattern is not like
• Checking the meaning of outliers
• Using extreme cases
• Following up surprises
• Looking for negative evidence

• Testing our explanations and theories
• Making if-then tests
• Ruling out spurious relations
• Replicating a finding
• Checking out rival explanations
• Getting feedback!

19
Summary Concurrent Flows of Activity in
Qualitative Data Analysis
20
Let the discussion begin!