Under "Statistics - Competency" you can see three statistical indicators of the ratings given for the entire competency:
Under "Statistics - Behavior Anchors" you can see various statistical indicators of the ratings given for the individual Behavior Anchors of a competency:
The indicators mentioned are explained in detail below.
They are also displayed in color in the Insights, depending on how good the values are. This is to give you a rough guide for interpreting the results.
In the Applysia Insights, the Observer Agreement is automatically calculated (for all statisticians: the intraclass correlation) - both for the ratings of all competencies and the ratings of the Behavior Anchors.
The correlation is thus a measure of the extent to which the ratings of different Observers match, e.g. for a Behavior Anchor. Differences are averaged across Candidates to ensure that only general tendencies and not individual profiles influence the data. If, for example, an Anchor has a very low agreement, this could be an indication that it is not formulated clearly enough and thus leads to misunderstandings among the Observers as to what exactly is to be rated. It can happen that Observers give very different ratings, although they have observed the Candidate at the same time. Here, the highest possible level of proficiency is desirable.
Observability indicates whether a competency or Behavior Anchor can be observed at all in an exercise. If, for example, half of all Observers do not rate a Behavior Anchor in an exercise, this could indicate that this Behavior Anchor often simply cannot be (clearly) observed in this exercise. This is then a good starting point to specifically evaluate the design of the exercises, to revise them if necessary, and thus to (clearly) rate the desired competencies. The goal here should therefore be to achieve 100% proficiency.
The internal consistency (for all statisticians: here Cronbach's Alpha) helps to assess to what extent the Behavior Anchors are suitable for assessing a related competency. A high value means that the Anchors are strongly related and thus probably suitable for rating the same competency. If the value is low, at least one of the Anchors probably does not fit the others. This can be the case in particular with quite broad competency classes and "combined" competencies such as "Strategy and Action competency". In principle, lower values do not mean that the Behavior Anchors are "bad", but rather that one should take a closer look here, because it could also make sense to divide up the competency, for example.