Validity & Reliability
Measurement & Threats
Measurement & Threats
Results from an assessment can predict a future outcome
Predictive validity
A test is not administered as the chance of the condition is remote
Test threshold
The chance that an individual with a prognostic factor will develop an outcome
Odds ratio
Which type of design is considered to have the least amount of bias
Experimental designs RCT best!
Systematic reviews that use quantitative methods are referred to as
Meta-analysis
Scores between different therapists are consistent
Interrater reliability
A test is not administered as immediate tx is indicated
treatment threshold
Range of scores within the true score for a variable is estimated to lie within a specified probability
Confidence interval
Best practice to implement when researchers lose subjects overtime
Intention to treat analysis
Not considering gray literature and non-English language publications are ways to counteract
Selection and publication bias
A researcher loses subjects during the course of a study
Attrition
A diagnostic test is capable of correctly classifying individuals with the condition of interest ( true positive)
Sensitivity
p value < .05
What number represents a large effect size
.80 and above
To strengthen the validity of a review, researchers should be sure to
Identify search terms, databases used, and the selection process
How well a new assessment correlates with the gold standard
concurrent validity
A test can correctly identify individuals without the condition (true negatives )
Specificity
The process of operationally defining the sample in a study comes from
setting clear inclusion and exclusion criteria
The minimal level of change required in an intervention before the outcome would be considered worthwhile
Minimally clinically important difference (MCID)
What is the next step after a study is selected to be included in the review process
It should be reviewed for quality and bias using standardized processes and or tools
changes in subjects characteristics overtime
maturation
What likelihood ratio would indicate that a test result is no better than a coin flip ( 50-50) chance
LR= 1
A retrospective design that is often used when researchers cannot easily follow subjects over time
Case control design
When determining the difference between groups, what determines the data analysis to be used ( i.e. T-test vs chi square)
The type of data represented by the outcome Parametric (ordinal/ratio) vs nonparametric (ordinal/nominal)
A method to depict the results of a systematic review that show differences in likelihood, odds, risk ratios and effect size
Forest plot