I work in experimental economics, labor and housing market.

I am trying to answer the following question: why do researchers use matched pairs of applications in labor market correspondence tests (typically in hiring context), whereas they use a completely randomized approach in the housing market correspondence tests (typically rental housing market)?

I thought a possible explanation might be the following:

1) The completely randomized approach reduces the potential bias (i.e. similar probabilities for discriminators and non-discriminators to end up in the groups I compare).

2) The completely randomized approach also allows utilizing a larger sample, avoiding over/under representation of discriminators or non-discriminators.

3) The randomization process also shall be preferred to the matching approach (usually used in correspondence tests in the labour market), since the researcher might not be fully aware of which variables are relevant for the matching.

4)Finally, with the matching approach the detection probability will be limited.

What do you think?
Is the answer complete?
Anything to be corrected/added?