L12: Single Case Studies

Question 0

Single case studies: uses and benefits

Answer 0

• idiographic approach (unique amongst individuals)
• qualitative and quantitative approaches
• extreme case sampling (rare cases)
• sources of ideas for future testing
• research in therapy: communicating clinical observations
• falsification (is one case enough?)
• potentially valuable info can be lost in averaging across groups
• genuine sources of variation mistaken with ‘random error’
• mistakes of aggregation
• each person is a unique case requiring separate explanation

Question 1

Why do we test groups?

Answer 1

• we want a reliable sample to represent the general case
• minimize the influence of normal variation/extreme cases
• external validity & generalization
• statistical analysis and power

Question 2

Potential problems with single cases

Answer 2

• anecdotal; non-public observation; reliability, replicability
• selection bias
• subjectivity/observer bias
• internal validity? How ambiguous or unambiguous are claims being made?
• external validity? Generalisability?

Question 3

Single cases in neuropsychology

Answer 3

• Identifying neural substrate of a particular brain function through case studies
• selective cog deficits produced by brain damage may help us to understand the functional architecture of the normal cognitive system
• single cases particularly suited, can indicate uniquely abnormal cognition
• brain damage may disrupt cognitive processes in a variety of different ways
• therefore performance differences within a group of brain damaged subjects cannot be dismissed as ‘noise’
• averaging over group is inappropriate

Question 4

What do single case brain damage studies tell us about the normal mind?

Answer 4

• well defined theories place constraints on how processing can break down
• a theory can be falsified (or require modification) by single case counter-example
• the way in which a system breaks down can tell you about the way it works normally
• application: cognitive neural repair

Question 5

Dissociations

Answer 5

• used to assess neural substrate of functions
• performance on task 1 impaired but performance on task 2 intact I.e. handled by different cognitive processes
• possible problem: impairment is apparent only because the task might be tapping more difficult elements of the same functions

Question 6

Double dissociations

Answer 6

• S: A, performance on task 1 fine, but performance on task 2 impaired
• S: B, performance on task 1 impaired, but performance on task 2 fine.
• stronger evidence that there are neural/cognitive processes involved in task 1 that are not involved in task 2 & vice versa
  • I.e. two processes function independently

Question 7

Use standardized tests & compare case performance to (large-N) normative data to judge performance deficits in single cases

Answer 7

• often not possible
• interesting phenomena may not be covered by standardized tests
• new constructs in neuropsych
• Norms can be difficult to collect
• convert to Z score: refer to area under normal curve, statistics of control sample treated as pop’n parameter, ok when sample large but typically not

Question 8

Ignore norms & use intra-individual comparisons to judge performance deficits in single cases

Answer 8

• can be misleading since no normal anchor point
• single cases appear to have deficits initially but not when compared to normal OR deficits are considered normal until compared to normal

Question 9

Compare the case’s performance to that of a (moderate-N) matched control sample to judge performance deficits in single cases

Answer 9

• construct ‘designer’ test & access control sample
• what counts as deficits & normal performance
• problems of statistical analysis (low-N)

Question 10

Problems with z score

Answer 10

• Inflated type 1 error rate when z is used with small control samples
• z assumes large sample sizes
• solution: modified t-test which pretends that population parameters are in play