Lecture 3 Flashcards
Independent Variable
- Variable that is manipulated (assignment to treatment status)
- Presumed “cause” of an effect
- Characteristic that is being observed or measured that is hypothesized to influence an event
Dependent Variable
- Outcome (result) of an intervention or exposure, event that occurs or characteristic that changes
- Measured variable used to determine the effects of the independent variable
- Outcome whose variation we seek to explain or account for by the influence of the independent variables
Confounder variable
factor other than the independent variable that influences outcome
Attribute variables
Used to describe a sample (ie demographics)
Explanatory varaibles
Variable that causally explains the relationship or outcome under study
Predictor variable
Variable that is causally related to a future outcome
Internal validity
- Extent to which the study lacks BIAS
- How well does the study support a causal relationship?
External validity
Generalizability of results
-Can results be applied to the population? Can they be applied to your patient?
Bias
- A systematic error in the way a study is carried out that can lead to false conclusions
- Any trend in the data, analysis, interpretation, collection, publication or review of data that can lead to conclusions that are different from the truth
Threats to internal validity/sources of bias
History, maturation, testing, instrumentation, group assignment, loss to follow-up, treatment crossover, compensatory issues
Threats to external validity (generalization)
Subject selection, setting of research, passage of time since study conducted
Causation
An outcome or event will occur as a RESULT of a previous event
- If exposure A occurs, disease B will result
- If intervention X is provided outcome Y occurs
Association
A statistical RELATIONSHIP between two events that may or may not be causal
Measurement
Basic ability to:
- Describe and classify pts
- Demonstrate change
- Communicate info to others
- Means of evaluating pt/client condition and response to treatment
- Compare and discriminate between individuals or groups
- PRACTICALLY ALL CLINICAL DECISIONS BASED ON MEASUREMENT
Constructs
- Behaviors or events that are not directly observable/measurable
- Inferred by measuring associated behaviors or attributes
- Multidimensional
- Require development of an instrument to measure it
Examples of constructs
Disability, motivation, socioeconomic status, health related quality of life
Scales of Measurement
Differentiation can be accomplished with: Names, Numerals, Numbers
Categorical Variables
Nominal variables -2 levels vs multiple levels -Quantify by counts/frequencies and percents Ordinal Variables -Categories rank ordered -Quantitative analysis is ambiguous
Numeric Variables
Interval and ratio variables
- Zero value= True for ratio arbitrary for interval
- Equal distance between units
- May use mathematical operations for both
- NOT ALL NUMBERS NUMERICALLY MEANINGFUL
Types of Measurement error
Random= Chance variation in measurement
-Analytic (statistical) procedures are used to quantify uncertainty due to random error
Systematic= Results from identifiable source
-Bias dealt with in design of study and through good procedures
Sources of error
Tester/rater inaccuracy, instrument imprecision, subject response biases
Variance
A statistical measure of the variability or “spread” of observations in a sample
- True variability of the measure within the population
- Variability due to measurement error
- Related to standard deviation
Reliability
-Degree to which measurement procedure is repeatable
Validity
Degree to which a measurement correctly estimates the TRUE value of the object being measured
Types of reliability
Test-retest reliability, inter/intra-rater reliability, reliability coefficients
Types of validity
Face validity, content validity, criterion-related validity, construct validity
Measuring change
Numeric: Value of X at time 1 compared to value of X at time 2 (X2-X1)
Categorical: Change in an individual from time 1 vs time 2 (T2-T1)
Standardized Response Mean (SRM)
Expresses change in standard deviation units
SRM= mean of pre to post treatment change/standard deviation of change score
