ACS 5-year vs. 1-year estimates - which to use?

Are there some helpful rules of thumb for determining whether to use a 5-year or 1-year estimate of American Community Survey data?

I understand that the 5-year data can be a little more precise on a granular level, and that the 1-year data is a bit more current- but I always find myself just using the 5-year data when it is available.

I am looking at large cities in particular - would getting population data on the city-level already be a task that is better to use 5-year estimates for? What about when looking at specific age groups on the city level?
To my understanding, once one is looking at PUMAs, the 5-year estimates are more reliable and should be focused on, is this a decent idea?

As I’m looking at change over time, and making comparisons between different data, the precision is pretty important to me, as I don’t want to ‘find’ a trend that is mostly explained by statistical noise.

Anyway, very new to understanding the arguments for use of the 1-year vs the 5-year, and would love to hear any of your thinking on it, tips, or rules of thumb.

Thanks a ton.

There are no hard-and-fast rules when it comes to choosing between 1-year and 5-year estimates. As you mentioned, there is a tradeoff between currency and reliability when it comes to using 1-year data (more current, less reliable due to smaller sample size) or 5-year data (less current, more reliable due to larger sample size). If you are studying a small population, the 5-year estimates are your only option. Studies with a large population (defined by the U.S. Census as > 65,000) will work with either estimate. Since you are more concerned with precision, I suggest you use the 5-year estimates. For more information, refer to these guidelines and Section 3 of the General Data Users Handbook by the U.S. Census Bureau.