Term
|
Definition
A way of clarifying how a variable will be USED.
Important for replicating a finding
Remember our lesson on empirical questions – the more specific your definition, the better. |
|
|
Term
|
Definition
| states that the finding results from sampling error (i.e., size too small), rather than a reality of the population. |
|
|
Term
|
Definition
“...measures what it is designed to measure and accurately performs the function it is purported to perform” (Patten, p. 61)
Validity is always...
Determined relative to a purpose
A matter of degree
Operational definitions may be very specific, but have poor validity for a particular purpose. |
|
|
Term
| How can you tell if an operational definition (e.g., a survey) is valid? |
|
Definition
Compare it to criteria logically
It looks like it’s pretty good.It addresses all the main concerns...
Compare it to other (more established) operational definitions
Validity coefficients
Correlation coefficients Range:-1to0to+1 |
|
|
Term
|
Definition
Consistency
Over time
Between observers
Over items (within a survey instrument)
Measured by comparing scores
Cronbach’s alpha (over items)
Pearson correlation (over time, between observers)
A reliability coefficient of .70 or higher is
considered "acceptable" in most social science research |
|
|
Term
|
Definition
Optimum performance
Achievement tests
Aptitude tests
IQ tests |
|
|
Term
|
Definition
How do I convert my hypothesis into a better one?
Break up complex hypotheses into several simple ones.
Based upon your instrument, figure out if your variables are likely categorical or continuous.
Phrase hypothesis accordingly. We hypothesize that...
Variable 1 will be positively/negatively correlated with
variable 2.
People in Category 1 (variable 1) will have higher mean
score of variable 2 than people in category 2 (variable 1) |
|
|
Term
| 3 common ways of getting a sample |
|
Definition
1. Convenience
2. Volunteer
3. Random |
|
|
Term
| Random is not a random word |
|
Definition
It means something very specific in a
research context:
Every member of a population has an equal chance of being in the sample.
E.g., drawing names from hat,sing random #
table.
Requirements: You must...
Know how many are in your population
Have access to every member,
Have complete agreement (i.e., no refusals).
In reality, a random sample is almost impossible to generate. |
|
|
Term
| What if you have some information about the population size? |
|
Definition
If you know the population size:
Can use systematic sampling (e.g., every nth
person on an alphabetic list)
If you know some proportions, can reduce
sampling error and bias by stratified sampling
Stratify and randomly draw within strata
Helps with big-picture variables that you know will influence your study variables.
Stratification variables must be relevant and
independent of one another.
Stratification is not for comparing subgroups. |
|
|
Term
|
Definition
Cluster
Purposive (qualitative especially)
Snowball |
|
|
