What's the relationship between "data association" and "loop closure" in SLAM?

Data association is to make sure that the sensor observations that have been token belongs to a certain real landmark not another one . So when you explore and do SLAM you can identify which observation from which specific landmark. So we associate the data given “observation” with its landmark

While loop closure is : revisiting places that was mapped before “looping again” which make the uncertainty goes very low and makes you more confident that what you expected to see”because you have revisited these landmarks before” is what really seen again so your confidence in the maps and your location goes very high and the uncertainty and noise goes very low.