Whatever method Blackstone uses. Been using them for 10+ years, well aware that there are other labs that use different methods, probably should give one a try. However, my 2019 Civic was reliably consistent @ about .5% or less through many samples, all with 5k change intervals, the CRV, operating in essentially the same manner, hasn't had a sample with less than 2.5%, including one taken at 4200 miles showing 3.5% with the OLM indicating 50%, thinned just below grade as a result of the dilution. Perhaps there are more accurate tests, but the OLM showing 50% life remaining under those circumstances, indicates to me that I certainly wouldn't want to put another 1-2k on that oil. Honda's MM is an algorithm, in reality is only an estimate of oil condition and in the case of (at least mine) the CRV is seriously flawed.
Also of interest, my CRV forum won't allow further discussion of oil dilution, insisting that "Honda has solved that problem", but when my Honda dealership indicates "oil level high on dipstick" on the Multipoint Inspection form and has a pre-printed space for that, suggests that there still is a problem. I don't know of other manufacturer's small-displacement Turbocharged DI engines having oil dilution issues, but I'm fairly sure it's happening.