Term
| List the levels of measurement |
|
Definition
Nominal
Ordinal
Interval
Ratio
*These are listed in levels of precision from weakest to strongest* |
|
|
Term
|
Definition
Putting something in categories (different in kind). Cannot do sophisticated stats because no math.
Order doesn't matter (male or female, for example)
Percentages are ok |
|
|
Term
|
Definition
Distance between ranks is unknown
*Likert scale |
|
|
Term
|
Definition
Beginning to know distance with some precision.
|
|
|
Term
|
Definition
| The most accurate method of measurement. |
|
|
Term
| Describing variables by measuring them |
|
Definition
*Correspondence *Standardization *Quantification *Duplication |
|
|
Term
|
Definition
| Taking an idea and converting it into a number. |
|
|
Term
|
Definition
| Repeating a study as closely as possible - but not as precise as replication (which is exact repeat of the study) |
|
|
Term
| Criteria for Selecting a Measuring Instrument |
|
Definition
*Utility
*Sensitivity to small changes
*Non-reactivity
*Reliability
*Validity
-Needs to be a good fit to your study.
-Should be sensitive to small changes
-Tool should not cause any changes, should only
measure change
-Consistency w/which variable is measured
-Tool should not change - should be the same |
|
|
Term
|
Definition
Consistency w/which the variable is measured.
*Test-retest method - Re-testing the same subjects to see if the results are the same
*Alternate-form Method
*Split-half Method
*Observer Reliability - Two different people observe the same thing. |
|
|
Term
|
Definition
Am I measuring what I really want to measure?
*Content validity
*Criterion validity
-Concurrent validity - "My tool is matching up
with that tool right now."
-Predictive validity - "Will my tool predict results of
another tool later?"
*Face validity - weakest kind. "It looks good, I think it makes sense." |
|
|
Term
| Reliability and Validity Revisited |
|
Definition
*A tool can be consistent, but not valid.
*A tool can be neither consisten nor valid
*A tool can be valid and reliable
*If a tool is valid, it is automatically reliable, but not
vice-versa. |
|
|
Term
|
Definition
| A good correlation rate is r=.80 or higher |
|
|
Term
|
Definition
*Constant errors
*Random errors |
|
|
Term
| What is a constant error? |
|
Definition
| Errors that stay constant throughout the research study. They stay constant because they come from an unvarying source (may be the measuring instruments used, the research participants, or the researchers themselves). |
|
|
Term
|
Definition
Random errors that are not constant are difficult to find and to make allowances or. Random errors spring out of the dark, wreak temporary havoc, and go back into hiding. May or may not cancel each other out.
Three types of random errors:
*Transient qualities of the research participant (fatigue, boredom)
*Situational factors (the weather, noise outside the window)
*Administrative factors - anything relating to the way the instrument is administered, or the interview conducted or the observation made. Include transient qualities of the researcher as well as sporadic stupidity like reading out of the wrong set of instructions. |
|
|
