Qualitative Studies: Case Studies

1
Qualitative StudiesCase Studies
2
Introduction

• In this presentation we will examine the use of
  case studies in testing research hypotheses
• Validity
• Quality
• Analysis.

3
Case study methodology

• Case studies are used as a research tool to test
  hypotheses in an ex post facto manner ie when the
  researcher CAN NOT control ANY of the variables
  under investigation.
• Case studies are therefore usually concerned with
  investigating how or why events occurred.

4
Case study methodology

• Case studies involve examining what has occurred
  to identify reasons for the occurrence.
• The lack of control leads to a number of
  potential problems with case study based research
  which, if not controlled, can invalidate any
  conclusions drawn from the research.

5
Case study methodology

• There are three typical criticisms of the case
  study methodology
• Problems of bias
• Lack of generalisability
• They are only suitable for explanatory
  investigations.

6
Problems of bias

• Many researchers are concerned with the apparent
  lack of rigour involved in case study research
  (as compared to experimental studies) bought
  about by investigator bias.
• Relationships may appear simply because you are
  looking for them.

7
Generalisability

• Because case studies allow no control over any of
  the variables many question whether conclusions
  drawn from one specific example can be applied
  generally.

8
Exploratory studies

• Many argue that only experimental research can be
  used to establish causal relationships with case
  studies being limited in use to the explanatory
  stage of a research project.
• This view misses the most important aspect of a
  case study, its holistic and real-life nature.

9
Design of a case study

• Many of the concerns associated with the use of
  case studies can be minimised by careful design.
• Validity is the most important issue to be
  addressed in the design of a case study
  methodology.

10
Concepts and indicators

• Most research involves measuring concepts (eg
  intent).
• Unfortunately concepts are normally abstract and
  thus not directly observable.
• Thus indicators (variables) are specified which
  operationalise the concept.

11
Concepts and indicators

• Unfortunately not all concepts can be easily
  operationalised and thus the validity of the
  indicator is drawn into question.
• One way around this problem is to use multiple
  indicators each of which are a partial
  operationalisation of the concept.

12
Validity of indicators

• The next question is how to evaluate the validity
  and reliability of the indicators does it
  measure what it is supposed to measure.

13
Validity of indicators

• There are various types of validity
• Face validity
• Criterion validity
• Construct validity
• Internal validity
• External validity.

14
Face validity

• Face validity involves a subjective evaluation of
  the indicator using
• Logic
• Common sense
• Previous reported studies
• Jury/Expert opinion.

15
Criterion validity

• Criterion validity uses a well established
  indicator to compare the performance of a newly
  defined indicator.
• The new indicator behaves in a similar manner to
  the established one then validity is assumed.

16
Construct validity

• Construct validity involves examining one
  indicator and its relationships to other presumed
  indicators of the same concept.
• High correlation between the indicators implies
  they are valid measures of the concept.
• Perfect correlation implies redundancy.

17
Construct validity

• In case studies three strategies can be used to
  increase construct validity
• multiple sources of evidence
• chains of evidence
• informants comments.

18
Internal validity

• Internal validity examines the extent to which a
  research finding (eg A lead to B) is valid.
• Internal validity involves eliminating all other
  relationships between other variables (eg C) and
  B.

19
Internal validity
Independent Variable
Dependent Variable
20
Internal validity
Other possible Variables
X
Independent Variable
Dependent Variable
21
Internal validity

• There are numerous factors that can effect
  internal validity.
• History/Maturity
• a long time between observations can cause
  problems.
• Testing
• people behave differently under test conditions.

22
Internal validity

• Selection
• observed differences could be the result of
  differences within the group.
• Mortality
• systematic dropout from the study.

23
Internal validity in case studies

• The nature of case studies require inferences to
  be made during data collection.
• Is the inference correct?
• Have all possibilities been considered?
• Is the evidence convergent?
• The overall quality depends on the quality of the
  investigator.

24
Internal validity in case studies
25
External validity

• External validity is concerned with the
  applicability of the research results to other
  (non examined) populations.
• Testing
• attitudes may change as a result of the
  questioning and the sample thus be
  unrepresentative.

26
External validity

• Sample selection
• self-selecting samples could be biased - include
  details of those who refused to cooperate.
• In case studies the use of multiple case studies
  usually satisfies the requirements for external
  validity

27
Reliability

• Reliability is concerned with the reproducibility
  of measurements.
• When attitudes are examined how do you know that
  each respondents ratings are the same?
• Interview structure, telephone protocols,
  standard letters, rigour in data collection and
  pilot studies all help in improving reliability.

28
Analysing case study data

• Each case study must be reported in detail
• the visit
• interviews
• conversations
• facts
• evidence in support/rejection of hypotheses

29
Analysing case study data

• conclusions
• outstanding issues
• further investigation
• other case studies.
• Across case studies
• replication
• rigourous thinking.

30
Summary

• A good case study must be
• Significant
• Complete
• Consider all perspectives
• Display sufficient evidence
• Compelling.