Research Methods in
Clinical Neuropsychology
Flashcards — Study Guide
122 cards covering Weeks 1–7, Scenarios, Exam Twists & Diagnostics
University of Groningen — Dr. A. Fürmaier
,Week 1
Card 1
What are the 3 observational study designs?
1. Cross-sectional — snapshot at one point in time
2. Case-control — starts from outcome, looks back (outcome → exposure)
3. Cohort — follows a group over time (exposure → outcome)
Card 2
Describe the 3 main observational study designs: cross-sectional, case-control, and cohort.
Cross-sectional: measures variables at a single point in time — a snapshot. Example: measuring ADHD
symptoms and sleep quality in 300 students at one moment. Strengths: fast, cheap, good for hypothesis
generation. Limitation: cannot establish causality (no temporal sequence).
Case-control: starts from the outcome, looks back at causes (outcome → exposure). E.g. patients with
dementia vs. healthy controls — did they have head injuries? Risk of recall bias.
Cohort: follows a group from exposure to outcome over time (exposure → outcome). E.g. follow 1000 rugby
players for 20 years — who develops dementia? Stronger temporal inference: you measure exposure first
and outcome later, so you know the time order. Downside: expensive and time-consuming.
Card 3
Name 5 strategies to reduce random error (improve precision).
• Standardize measurement procedures
• Train the assessors
• Automate the instrument
• Blind the assessors
• Repeat the measurement
Card 4
What are the 3 problems with Bonferroni correction? Who argued this?
This was argued by Perneger.
• It tests an irrelevant hypothesis (the universal null): Bonferroni tests whether there is no effect at all across
ALL your tests combined. But you as a researcher want to know about each specific hypothesis individually
— the overarching 'is everything zero?' question is clinically meaningless
• The interpretation of a result changes depending on how many other tests you run — illogical
• It dramatically increases Type II errors: real effects are missed
Card 5
When IS Bonferroni correction appropriate? Who argued this?
According to Perneger:
1. Purely exploratory research: you have no specific hypotheses and are scanning the data for any interesting
pattern. Here the universal null ('nothing at all is going on') is actually a relevant question, because you
genuinely don't know what to expect.
2. Many unrelated tests: if your tests have nothing to do with each other (e.g. you measure 20 completely
different variables), each significant result is essentially a standalone finding. Without correction, you expect
1 in 20 to be a false positive by chance alone. Bonferroni protects against this.
3. Same hypothesis across subgroups: e.g. testing 'does treatment X work?' separately in men, women,
young, old. You're asking the same question multiple times, so the false positive risk accumulates and
correction makes sense.
, Card 6
What are the 3 strategies for the control group problem? Who proposed them?
These were proposed by Bonato.
• Cautious interpretation: acknowledge that groups differ on more than just the pathology. E.g. ADHD patients
may also differ from controls in sleep, motivation, education, medication use — so a difference on a cognitive
test is not necessarily due to ADHD alone. You must discuss these alternative explanations in your paper
• Fine-grained matching: match groups on relevant variables (age, education, IQ, gender) to reduce
confounds
• Control task: a task with the same setting, difficulty, and demands but tapping a different cognitive function.
E.g. if your experimental task measures attention, use a control task that measures language. If the patient
group scores lower on both, the problem is likely general (motivation, fatigue). If they only score lower on the
attention task, the deficit is specific
Card 7
Name 4 obstacles to computerized neuropsychological testing. Who identified these?
These were identified by Schmand.
• Psychometric: digitizing a test creates a new instrument
• Technical: technology outdates faster than validation
• Theoretical: traditional tests have decades of research
• Strategic: lack of standardization fragments the field
Card 8
What is the difference between precision and accuracy? What is the threat to each?
Precision (= reliability): reproducibility of measurements. Threatened by random error — unpredictable
fluctuations like a patient's concentration varying between sessions, or slight differences in test
administration. Reduce by: standardizing procedures, training assessors, repeating measurements.
Accuracy (= validity): closeness to the true value. Threatened by systematic error / bias — consistent
distortion in one direction, like a test that always underestimates IQ in non-native speakers. Reduce by: using
validated instruments, blinding, calibrating equipment.
Card 9
How should you deal with the pitfalls of significance testing? What is the key distinction?
The key distinction is between exploratory and hypothesis-testing studies:
1. Exploratory / hypothesis-generating studies:
You have no a priori hypotheses. You are scanning data for patterns. Here you should do no significance
testing at all — just describe patterns and generate hypotheses for future research.
2. Hypothesis-testing studies:
• Have a clear rationale and justification for each hypothesis
• Formulate hypotheses a priori (before collecting data)
• Do power calculations during study planning
• Perform significance testing as planned (no adding tests after seeing data)
• Do not more, but far less statistical hypothesis testing — only test what you planned
Clinical Neuropsychology
Flashcards — Study Guide
122 cards covering Weeks 1–7, Scenarios, Exam Twists & Diagnostics
University of Groningen — Dr. A. Fürmaier
,Week 1
Card 1
What are the 3 observational study designs?
1. Cross-sectional — snapshot at one point in time
2. Case-control — starts from outcome, looks back (outcome → exposure)
3. Cohort — follows a group over time (exposure → outcome)
Card 2
Describe the 3 main observational study designs: cross-sectional, case-control, and cohort.
Cross-sectional: measures variables at a single point in time — a snapshot. Example: measuring ADHD
symptoms and sleep quality in 300 students at one moment. Strengths: fast, cheap, good for hypothesis
generation. Limitation: cannot establish causality (no temporal sequence).
Case-control: starts from the outcome, looks back at causes (outcome → exposure). E.g. patients with
dementia vs. healthy controls — did they have head injuries? Risk of recall bias.
Cohort: follows a group from exposure to outcome over time (exposure → outcome). E.g. follow 1000 rugby
players for 20 years — who develops dementia? Stronger temporal inference: you measure exposure first
and outcome later, so you know the time order. Downside: expensive and time-consuming.
Card 3
Name 5 strategies to reduce random error (improve precision).
• Standardize measurement procedures
• Train the assessors
• Automate the instrument
• Blind the assessors
• Repeat the measurement
Card 4
What are the 3 problems with Bonferroni correction? Who argued this?
This was argued by Perneger.
• It tests an irrelevant hypothesis (the universal null): Bonferroni tests whether there is no effect at all across
ALL your tests combined. But you as a researcher want to know about each specific hypothesis individually
— the overarching 'is everything zero?' question is clinically meaningless
• The interpretation of a result changes depending on how many other tests you run — illogical
• It dramatically increases Type II errors: real effects are missed
Card 5
When IS Bonferroni correction appropriate? Who argued this?
According to Perneger:
1. Purely exploratory research: you have no specific hypotheses and are scanning the data for any interesting
pattern. Here the universal null ('nothing at all is going on') is actually a relevant question, because you
genuinely don't know what to expect.
2. Many unrelated tests: if your tests have nothing to do with each other (e.g. you measure 20 completely
different variables), each significant result is essentially a standalone finding. Without correction, you expect
1 in 20 to be a false positive by chance alone. Bonferroni protects against this.
3. Same hypothesis across subgroups: e.g. testing 'does treatment X work?' separately in men, women,
young, old. You're asking the same question multiple times, so the false positive risk accumulates and
correction makes sense.
, Card 6
What are the 3 strategies for the control group problem? Who proposed them?
These were proposed by Bonato.
• Cautious interpretation: acknowledge that groups differ on more than just the pathology. E.g. ADHD patients
may also differ from controls in sleep, motivation, education, medication use — so a difference on a cognitive
test is not necessarily due to ADHD alone. You must discuss these alternative explanations in your paper
• Fine-grained matching: match groups on relevant variables (age, education, IQ, gender) to reduce
confounds
• Control task: a task with the same setting, difficulty, and demands but tapping a different cognitive function.
E.g. if your experimental task measures attention, use a control task that measures language. If the patient
group scores lower on both, the problem is likely general (motivation, fatigue). If they only score lower on the
attention task, the deficit is specific
Card 7
Name 4 obstacles to computerized neuropsychological testing. Who identified these?
These were identified by Schmand.
• Psychometric: digitizing a test creates a new instrument
• Technical: technology outdates faster than validation
• Theoretical: traditional tests have decades of research
• Strategic: lack of standardization fragments the field
Card 8
What is the difference between precision and accuracy? What is the threat to each?
Precision (= reliability): reproducibility of measurements. Threatened by random error — unpredictable
fluctuations like a patient's concentration varying between sessions, or slight differences in test
administration. Reduce by: standardizing procedures, training assessors, repeating measurements.
Accuracy (= validity): closeness to the true value. Threatened by systematic error / bias — consistent
distortion in one direction, like a test that always underestimates IQ in non-native speakers. Reduce by: using
validated instruments, blinding, calibrating equipment.
Card 9
How should you deal with the pitfalls of significance testing? What is the key distinction?
The key distinction is between exploratory and hypothesis-testing studies:
1. Exploratory / hypothesis-generating studies:
You have no a priori hypotheses. You are scanning data for patterns. Here you should do no significance
testing at all — just describe patterns and generate hypotheses for future research.
2. Hypothesis-testing studies:
• Have a clear rationale and justification for each hypothesis
• Formulate hypotheses a priori (before collecting data)
• Do power calculations during study planning
• Perform significance testing as planned (no adding tests after seeing data)
• Do not more, but far less statistical hypothesis testing — only test what you planned
