Term
|
Definition
| Does it measure what it is supposed to measure? This can be broken down into (for example) internal, external, and construct validity. |
|
|
Term
|
Definition
| Clear specification of behavioral domains of interest. This is determined by domain sampling. |
|
|
Term
|
Definition
| Is the measurement consistent? That is, if given again would the same results be achieved. |
|
|
Term
|
Definition
| Calculated dividing the larger number by the smaller number and multiplied by 100. Recommended to use if being monitored by two observers for an entire day |
|
|
Term
|
Definition
| Calculated by: Agreement/(Agreement plus Disagreement) * 100. It is recommended to use this type of agreement because it provides a more statistically reliable source of IOA. It is also recommended to use if two independent observers are using a same sample. |
|
|
Term
| High and low rate problem |
|
Definition
High rates: may have high % of interval agreement because there were so many behavior occurrences that disagreements may not statistically show
Low rates: may rarely agree on occurrence of behavior and still have high % of interval agreement |
|
|
Term
|
Definition
Reactivity: people acting differently when being observed
Drift: moving from code definitions
Expectancy: coding what you are looking for
Complexity: too much to measure
Location of observer: may not be able to see everything
Rates of behavior: does it occur too often to code?
Episodic behavior: hard to catch extremely rare behaviors |
|
|
Term
| Pros & cons of co-plotting |
|
Definition
Pro: visual representation with data instead of just summary statistic, can actually see differences, better shows threats to internal validity, more closely related to decision ethics, good for team members without statistical background
Cons: no standard criteria, may clutter up graphs |
|
|
Term
|
Definition
| From ecological science, more concerns, more monitoring. If a behavior is improving, you can measure less frequently. |
|
|
