LECTURE 3 Flashcards

1
Q

one way repeated measures test

A

1 group over 2+ time periods

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

2 groups (exp and control) post-tested after intervention

A

post test only randomized group
post test only nonrandomized group

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

factorial fully independent

A

2 or more independent groups
2 or more independent variables
*diff participants in each cell

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

factorial fully repeated

A

2 or more independent variables
subjects participate in each cell!

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

factorial mixed

A

2+IV, with or without randomization
groups move through repeated conditions (time)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

crossover with or without randomization

A

groups get tested after intervention…then they switch! (one group gets placebo, other intervention)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

measurement theory

A

psychometrics!
foundation for evaluating tests
reliability and validity are the most fundamental measurement theory concepts

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

any study investigating reliability or validity of clinical and research measures is called…

A

methodological study!

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

variables like cognition, pain, coordination…must be specifically defined. This is called

A

operational definition

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

MDC

A

minimally detectable change (amount of change in instrument beyond error)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

MCID

A

minimal clinically important difference
amount of change in instrument resulting in clinically relevant change to patient

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

consistency or degree of association between two variables is called

A

reliability
*high if there is relative/actual agreement btwn two variables

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

accuracy/degree of measuring concept that is intended to measure

A

validity
*can be reliable but not valid, but not valid without being reliable

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

test-retest reliablity

A

can instrument be consistent?
can patient be consistent?

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

reliability between 2 ratings measures…

A

do they vary in similar way? (10, 20, 30 vs 100, 200, 300 = reliable/good co-variation

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

agreement is…

A

if 2+ ratings, do scores agree/match exactly?
*both reliability/agreement measure REPRODUCIBILITY

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

parallel forms reliability

A

equivalent forms!
questions are divided into 2 equal sets.
Two sets of questions are given to the same people. SATs are made this way

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

split-half reliability

A

questions split into two sets and BOTH sets given to same people

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

does instrument measure what it is supposed to measure? ex. does depression scale seem to have questions about depression?

A

face validity

20
Q

does IQ scale cover all areas of intelligence? Does instrument cover entire domain to be measured?

A

content validity (what experts agree it measures!)

21
Q

degree to which outcomes of one test correlate with outcomes on gold standard

A

criterion-based validity

22
Q

degree to which outcomes of one test correlate with outcomes on another test
(neither are gold standard)

A

concurrent validity

23
Q

can instrument predict future outcome? ex. BERG less than 45 predicts fall risk

A

predictive validity

24
Q

degree to which theoretical construct is measured by an instrument
how many different constructs are measured? are extra items not needed?

A

construct validity ex. factor analysis

25
Q

extent to which significant changes in participants are reflected in substantive changes in observed values

A

responsiveness to change
*ceiling and floor effects both limit responsiveness

26
Q

the more items/responses, the more

A

responsive!

27
Q

true positive rate

A

how often test will be positive if person HAS the disease (sensitivity)
*LOVES to be positive, so if its negative, RULED OUT

28
Q

true negative rate

A

how often test will be negative if you do not have disease (specificity)
LOVES to be negative, so if positive, prob have it.
RULED IN

29
Q

true positive/condition positive (all who have it)

A

sensitivity
true positive/ true pos + false neg

30
Q

true negative/condition negative

A

specificity
true neg/ false positive + true neg

31
Q

false negative is a type ___ error

A

2

32
Q

false positive is a type ___ error

A

type 1

33
Q

PPV is

A

true positive/ test outcome positive (true+ false positives)

34
Q

NPV is

A

true neg/ test outcome neg (false + true negatives

35
Q

If specificity of a test is 87%, then test was able to correctly identify 87% ___(with/without) fracture, but incorrectly gave false_____ to 13% of people.

A

SPIN: RULE IN test was able to correctly identify 87% without fracture, but incorrectly gave FALSE POSITIVE to 13% of people.

36
Q

What is the probability that a person with a positive clinical test truly has the condition?

A

positive predictive value
increases as prevalence of disease in pop increases

37
Q

What is the probability that a person with a negative clinical test truly doesn’t have the disease?

A

negative predictive value
increases as prevalence of disease in pop decreases

38
Q

What are the odds of a patient having a particular disease/condition?
OR
What is the chance that a given test result would be expected in a patient with the disease/condition compared to the chance that this same result would be expected in a patient without the disease/condition?

A

LIKELIHOOD RATIO

39
Q

higher positive likelihood ratio is

A

better!

40
Q

sensitivity/ (100-specificity) is

A

POS LR

41
Q

probaility person with condition tests positive (sensitivity) / prob a person without condition tests pos (false positive)

A

positive likelihood ratio

42
Q

probaility person with condition tests negative (false negative rate) / prob a person without condition tests negative (specificity)

A

NEGATIVE LIKELIHOOD RATIO

43
Q

(100-sensitivity) / specificity

A

NEG LR (lower is better)

44
Q

LR of 10 = 10x increase in the odds of having a condition with a positive test result.

A

positive likelihood ratio of 10

45
Q

LR of .1 = 10x decrease in the odds of having a condition with a negative test result

A

negative likelihood ratio