Section 3 Flashcards
0
Q
Reliability
A
- are we getting a consistent measurement (regardless of whether it is wrong/right)?
1
Q
Validity
A
- are we measuring what we think we are measuring?
2
Q
Measurement Errors
A
- differences in values that are a result of flaws in the measurement process
- Systematic: getting the same error each time
- Random: getting a different value each time
3
Q
Pragmatic Validity
A
- Use of multiple sets of indicators and values to test the validity of a given concept
- Concurrent: involves come pairing the results of one indicator to another
- Predictive: involves using the indicator to predict other behavior
4
Q
Construct Validity
A
- involves relating an indicator to an overall theoretical framework
- External Validation: based on the knowledge of the concept we seek to measure, we can postulate relationships between that concept and other concepts
- Convergent Validity: different methods of measuring the same concept should produce similar results
5
Q
Discriminant Validity
A
- inferring validity according to the indicators ability to differentiate across concepts
- the degree of which it is unrelated to indicators for other concepts
6
Q
Content Validity & Face Validity
A
Content: concerned with the content of what is being measured
Face: can knowledgable people be persuaded of this validity?
7
Q
Assessing Reliability
A
- An empirical matter; three different methods of assessment
- Test-Retest Method: Can we get the same results?
- Alternative/Parallel Forms Method: Using two ways of measure for the same case
- Subsample/Split-Half Method: Dividing sample into sub-samples; similar results should be achieved by all sub-samples
8
Q
Probability Sampling
A
- every member of population has a known and non-zero probability of being included in the sample
- this avoids unconscious bias
- allows use of inferential statistics
9
Q
Non-Probability Sampling
A
- no way to specify probability of inclusion; some pop. may not have any chance of inclusion
- used when convenience and economy out weight risk of bias
- also when no pop. list is avaliable
10
Q
Simple Random Sampling
A
- every member of pop. has same probability of inclusion
- Could result in extreme samples
- also can be time-consuming
11
Q
Systematic Random Sampling
A
- dividing the pop. size by size of desired sample to achieve sampling yield (denoted at ‘k’)
- then researcher will take every ‘kth’ person on the pop.list
- if the first individual is selected randomly, there is no restriction on inclusion
- can potentially produce extreme samples depending on how pop. list is ordered
12
Q
Proportionate Stratified Random Samples
A
- ensure key groups in pop. are represented in their correct proportions
- requires information about each person in the pop. before conducting study
13
Q
Disproportionate Stratified Random Sample
A
- same as proportional, except that this deliberately over-samples certain pop. groups
14
Q
Multi-Stage Random Cluster Samples
A
- used when no pop. list is available
- Convenience: whatever people happen to be available
- Volunteer: people self-select to participate
- Purposive: researcher uses their judgement to pick people of a pop.
- Snowball: start will small group of participants and ask them to reach out and ask more participants
- Quota: select sample that represents microcosm of pop.