Lecture 2 - Applied clinical psychometrics Flashcards
If a distribution is Gaussian, it means it is…
A) A normal distribution
B) A skewed distribution
A) A normal distribution
True or false, when scores fall along a Gaussian distribution, it means there is a simple correspondence between standard deviations and percentiles.
TRUE
True or false, IQ tests are an example of a test that produces a non-Gaussian distribution
FALSE
True or false, it is important to remember that functions cannot be “impaired”. Only scores can be impaired.
FALSE
True or false, all tests suffer from the problem of multiple determination.
TRUE
Name the three levels of sequential inference involved in clinico-psychometric assessments.
ETS - PS - UDC
- Elevated test scores
- Psychological syndromes
- Underlying disorders or causes
In this lecture, we learned about two main types of scales that are used in psychometrics. What were they?
RS - PS
- Rating scales
- Performance scales
True or false, RATING SCALES are typically Likert style that represent symptom severity, frequency etc
TRUE
Is the DASS-21 an example of a rating scale or a performance scale?
A rating scale
True or false, PERFORMANCE SCALES typically measure an ability or skill
TRUE
Is the WMS an example of a rating scale or a performance scale?
A performance scale
Of performance scales and rating scales, which scores typically have a “real world meaning”
Performance scales
(Note there are exceptions to this, but I doubt they are material to our exam)
In this lecture, we learned about three types of performance scales. What were they?
S - A - F
- Skill scales
- Ability scales
- Function scales
What type of performance scale would be used to assess things like reading, spelling, or playing the piano?
(Skill, ability, or function)
A skill scale
What type of performance scale would be used to assess things like visuospatial function, processing speed, working memory?
(Skill, ability, or function)
An ability test
Do ABILITIES (in the psychometric sense) typically have Gaussian or non-gaussian distributions?
(e.g., IQ)
Gaussian
What type of performance scale would be used to assess things like being able to move your mouth to produce sound, or sensation on your skin?
A function test
Do FUNCTIONS (in the psychometric sense) typically have Gaussian or non-gaussian distributions?
(e.g., having sensation on both sides of your face)
Non-Gaussian
In this lecture, we learned about THREE clinical considerations when assessing any sort of test score. What were they?
NA - MD - PS
- Normative analysis
- Multiple determination
- Pathognomonic status
The Z score MEAN is always 0. What is the equivalent T score, Index score, and Scale score?
T score = 50
Index score = 100
Scale score = 10
For Z scores, the SD is always 1. What is the SD value for T scores, Index scores, and Scale scores?
T score = 10
Index score = 15
Scale score = 3
For a Z score of -2, what is the equivalent T score, Index score and Scale score… and percentile?
T score = 30
Index score = 70
Scale score = 4
Percentile = 2nd
For a Z score of 1, what is the equivalent T score, Index score and Scale score… and percentile?
T score = 60
Index score = 115
Scale score = 13
Percentile = 84th
What is the percentile that corresponds to a Z score of 3?
99.87th
Are measures of psychopathology typically Gaussian or non-Gaussian, and what does this mean for interpreting scores in SD terms?
Psychopathology is typically non-Guassian
This means we should interpret SDs differently, because being 1 SD above the mean is a sign of much greater elevation, relative to norms, than it would be for an IQ test (for example).
What is the SD for a T score?
T Score SD = 10
What is the SD for a Scale score?
Scale score SD = 3
What are you comparing when conducting an ipsative analysis?
You’re comparing different scores of the same patient
(i.e., what is elevated vs not elevated)
True or false, test score label descriptors should be based on pathology, not the frequency or commonality of performance.
FALSE
Descriptors should be based on the frequency or commonality of performance.
The AACN recommend SEVEN labels to describe scores on Gaussian-distributed tests, ranging from “Exceptionally low score” to “Exceptionally high score”. What are the intervening FIVE labels
(1. Exceptionally high score)
- Above average score
- High average score
- Average score
- Low average score
- Below average score
(7. Exceptionally low score)
The AACN recommend FOUR labels to describe scores on NON-Gaussian-distributed tests, what are they?
(Bonus point for knowing the associated percentiles)
- Within normal limits score (> 24th percentile)
- Low average score (9-24th percentile)
- Below average score (2-8th percentile)
- Exceptionally low score (<2th percentile)
What two properties of a person does Charles use to demonstrate the dangers of summing scores?
Height and weight
(Height and weight correlate, but we wouldn’t sum them together to produce a meaningful score)
In what situations is professional/clinical judgement preferable to actuarial judgment?
Diagnoses and the related acts of reverse inference
In what situations is actuarial judgment preferable to professional/clinical judgement?
Prediction about the future (I think this is right but I’m not 100% sure)
