Inner TRIM3 Masthead

Assume three different units -- A, B, and C -- all simulated to be eligible for $125/month of SNAP benefits under current law. Assume also participation probabilities of .5 for units eligible for $50-$100/month, .8 for units eligible for $100-$150, and .9 for units eligible for $150-$200. If unit A when surveyed reported receiving SNAP benefits, it will be assigned a random number between zero and .8. This ensures that it will decide to participate in the baseline run since its random number will never be higher than the participation probability specified for units eligible for $100-$150 (.8). If unit B did not report SNAP benefits and TRIM3 chooses it as a ‘correct responder’ its random number will be from the range .8-1. This ensures that it will decide to not participate in the baseline run since its random number will always be higher than the .8 participation probability. If unit C, like unit B, does not report SNAP benefits but is not chosen to be a ‘correct responder,’ its random number will be anywhere from zero to one. This allows the unit's participation decision to go either way, depending upon the value that is actually assigned as its random number.

If the actual numbers assigned are as follows, their participation decisions in the base run will be:

Unit

Random Number

Benefits Eligible For

Participation Probability

Decision

A

0.6

$125

0.8

participates

B

0.85

$125

0.8

does not participate

C

0.3

$125

0.8

participates

If the following changes occur to the benefits they are eligible for (and consequently the participation probabilities they face), their decisions will change:

Unit

Random Number

Benefits Eligible For

Participation Probability

Decision

A

0.6

$75

0.5

does not participate

B

0.85

$175

0.9

participates

C

0.3

$75

0.5

participates

Unit A is now eligible for lower benefits, and the participation probability among $50-$100 units is correspondingly lower than $100-$150 units (.5 vs. .8). Unit A's ‘taste’ or ‘preference’ for participation, as indicated by its random number (.6), while sufficient for choosing participation when eligible for $125, is not sufficient when eligible for only $75. Conversely, unit B is eligible for higher benefits, and is thus more likely to participate (its participation probability goes from .8 to .9). Its ‘taste’ for participation (.85), while insufficient for participation when eligible for $125, is sufficient enough when it's eligible for $175. Unit C, having a generally strong ‘taste’ for participation (.3), continues to participate even though its benefits take the same drop as unit A's.