Flashcards in Ch19Research Deck (19):
What are the goals of methodological research?
to document and improve the reliability, validity, and responsiveness of clinical and research measurements
Reliability of measurement is influenced by:
Sources of variability
Levels of standardization
Range of scores
Reliability in Non-methodological Studies
What are the two basic designs for optimization research?
standardization and mean designs
What are the three approaches to standardization?
nonstandardized approach, highly standardized approach, and partially standardized approach
What are Sources of variability?
Differences found in repeated measurements of the same characteristic can be attributed to instrument, intrarater, interrater, and intrasubject components.
What are Levels of standardization?
Degree of standardization is the number of sources of variability within a reliability component that are controlled
Include: Non-standardized, Highly standardized and Partially standardized
What is the non-standardized approach?
Controls none of the sources of variability and would establish the lower limit for the reliability component studied (real-world)
What is the highly standardized approach?
Controls many of the possible sources of variability to determine the upper limits of the reliability of the component (ideal)
What is the Partially standardized approach?
Describes reliability with moderate levels of standardization; Standardizes a few sources of variability while leaving others non-standardized (clinic)
What is the range of scores?
The reliability of a measure should be determined over the range of scores expected for that measure
What is standardization design?
Compares the reliabilities of measurements taken under different sets of conditions; level of standardization increases until reliability goal is met; or start with highly standardized and decrease until feasible
What is mean design?
Compares the reliabilities of single measurements and also compares the reliabilities of measurements averaged across several trials; appropriate for measures that are difficult to standardize
Influences external validity; who the results of the study can be generalized to
Reason for Optimization designs
Researchers found less than optimal reliability for rehab measures; documenting reliability does nothing to improve it, need to study ways of optimizing reliability
Types of Validity Designs
Construct, Content, Criterion
Constructs are artificial frameworks that are not directly observable; three ways to establish: persuasive argument, predictions about patterns, performance of "extreme" or "known" groups on test of interest
Involves documenting that the test provides an adequate sampling of the behavior that it is measuring; compares items in test against actual practice of interest; four issues: content's completeness, relevance, emphasis and sample on whom measure was validated
Determined by comparing it with an accepted standard of measurement; Considerations: Selecting criterion, timing the administration of tests and selecting a sample for testing