Definition of 'relevant' In an Interview
This topic describes the definition of what constitutes a 'relevant' attribute. Relevancy can be used to control whether or not a screen is displayed during an interview and whether information is shown or hidden in an explanation and decision report. For more information on how to use these features, see:
Note: For interviews that use decision services, all inputs are presumed to be relevant for the outputs because decision services do not support relevance. That is, the decision service makes a decision based on a set of inputs that resulted in a set of outputs and there is nothing available from outside the decision service to distinguish which inputs were used for which outputs.
Rule 1: An attribute's value is relevant if changing it could cause the conclusion of the rule to change
Example 1:
A if
B and
C
If B is true and C is false, then A is false. In this example:
- B is not relevant because no matter what you change it to, the false value of C keeps A false.
- C is relevant because you could change it to true, and A would become true .
Example 2:
the total income of adult household members = InstanceSumIf(the household members, the household member is an adult, the household member’s income)
With the following sets of conditions and values, the result is $500:
- Household member 1:
- the household member is an adult = true
- the household member’s income = $500
- Household member 2:
- the household member is an adult = false
- the household member’s income = $100
For household member 1, the household member is an adult and the household member’s income are both relevant. For household member 2, the household member is an adult is relevant but the value of the household member’s income for household member 2 is not relevant because no matter what it is set to, it has no effect.
Example 3:
A if
B is unknown or
C
If B is unknown and C is unknown, then the conclusion is true. In this example:
- B is relevant, because B being unknown is causing A to be true.
- C is not relevant because changing its value it does not affect the outcome of A.
Rule 2: Where a set of values are not relevant individually (through Rule 1) but are equally responsible for the value of the conclusion, then all values in the set are considered relevant
This is intended to cover situations where attributes are equally relevant to the conclusion, with no single attribute value independently having an effect if it changes. Using Rule 1 Example 1 above, if B and C are both false, then A is false. Changing either B or C independently does not change the conclusion, so Rule 1 does not apply. However, you could change both of them to true, and it would change the conclusion, so because of Rule 2, they are both considered relevant.
Rule 3: All values that could be relevant if unknown values became known, are considered relevant
Example 1:
A = B + C
If B is unknown and C is 5, then the result is unknown. In this example:
- B is relevant because if it changed (to become known), it would affect the outcome (Rule 1)
- C is relevant because if B became known, it would be relevant to the outcome (Rule 3)
No special consideration of uncertainty is required - handling for uncertainty falls naturally out of the above rules.
Example 2:
A = B + C
If B is uncertain and C is unknown, then the conclusion is uncertain. No matter what value C becomes, A will always be uncertain. In this example:
- B is relevant, because if B changed to be unknown, then A would become unknown (Rule 1).
- C is not relevant, because even if it becomes known, it cannot become relevant.
Rule 4: In Latest Version Interviews, an attribute is only considered relevant if it is editable
That is, an attribute is only considered relevant if it is visible and not read-only.