Internal Validity Flashcards
Observed relationship between two or more variables should remain unambiguous as to what it means rather than being due to “something else”
Internal Validity
observed differences on the dependent variable are directly related to the independent variables and not due to some other unintended variable
Internal Validity
Selection of people for a study may result in the individual differing from one another in unintended ways
Subject Characteristics
Subject Characteristics is sometimes referred as
Selection bias/Subject characteristics threat
Loss of subject is also known as
mortality threat
When subjects or individuals drop out for one reason or another
Loss of Subject
true in most intervention studies since they they take place and time
Loss of Subject
A problem in questionnaire studies
Failure to complete instruments
Loss of subjects not only limits generalizability but also
introduce bias
A common misconception in loss of subject
that threat can be eliminated simply by replacing the subject
Attempt to eliminate the problem of mortality
provide evidence that the subjects lost were similar to those remaining on pertinent characteristics
Particular location in which data is collected or in which intervention is carried out
Location
the best method of control for location threat is
hold location constant (keep it same for all participants)
the way instruments are used may also constitute a threat to internal validity
Instrumentation
Alternative name for instrumentation threat
Instrumentation decay
Often the case when the instrument permits different interpretation results resulting in fatigue of the scorere
Instrumentation
Principal way to control instrumentation decay
schedule data collection and/or scoring
Demographic (age, gender, etc.) of the individual who collects the data in the study and how it may affects the nature of the data collected
Data collector characteristics
primary ways to control data collector characteristics
using the same data collector(s) throughout, analyzing data separately for each collector
data collector or scorer may unconscious distortion of data in ways to make certain outcomes more likely
Data collector bias
two principal techniques for handling data collector bias
Standardize procedures (require some sort of training) Collector lacks information
means the use of any form of instrumentation, not just “test”
Testing
Common practise before and after interventions
Testing
One of more unanticipated or unplanned evers may occur which can affect responses of subjects
History
