"The data for steady-state-off measurements were long-tail skewed, with the top 5 stoves (9% of sampled units) emitting half (49%) of all steady-state-off emiss

"The data for steady-state-off measurements were long-tail skewed, with the top 5 stoves (9% of sampled units) emitting half (49%) of all steady-state-off emissions." So, they have a couple leaky stoves in the equation and they extrapolate the data. But they never bother to determine what was leaking, which is actually a fairly simple thing to do since there are commercial leak detectors that are quite sensitive (https://www.southerncrossinc.com/wp-content/uploads/2019/10/IRwin-brochure-9.25.19.pdf https://www.southerncrossinc.com/wp-content/uploads/2019/10/...) to at home units that detect NG/propane/etc. Plus there were quite a number of other WTF moments in the document that don't match with my understanding of current install practices around NG heaters/etc. Not the least of which, is that if 1% of the usage (which is a lot!) is leaking that will easily show up on peoples meters and be something that can be smelled. Which granted most people don't run outside and check their meter in the morning after having turned off all their gas consumers but people do smell gas. I've known at least one person who ended up ripping up a bunch of the piping in their attic because it was leaking, but i'm guessing its more likely someone forgot about an old appliance with an actual pilot light or the data was categorized improperly. Lets just say, that if you were honestly trying to run this study, don't you think you would want to go investigate some of the outliers to at least note that the top couple cases were broken valves, or rusted piping, or just debris in the gas line damaging/keeping the valve from closing (which is why modern gas codes require sediment traps) fully?