Latent constructs help us think carefully about what we want to measure and how we define our constructs. They give us several views into the behavior and performance of the system we are observing, helping us eliminate rogue data. They make it more difficult for a single bad data source (whether through misunderstanding or a bad actor) to skew our results.
