{"id":"arxiv/2610.23504","url":"https://arxiv.org/abs/2610.23504","tags":["math.ST","stat.AP","stat.ME"],"score":0,"title":"Weighted spatial averages with missing observations: Convergence and accuracy of delta-method variance approximations","domain":"math.ST","source":"arxiv","authors":["Ashwin K Seshadri","Abhishek Pal Majumder"],"reviews":0,"subject":"math.ST","summary":"Weighted spatial averages from incomplete networks are ratios whose variance is approximated by truncating a Taylor series (the delta method). Three questions remain open: whether the series converges, how large a truncation's error is, and how it behaves as the network grows. We address these for observations missing completely at random, independently with a common probability $α$. First, the truncations, which count corrections for the denominator (the total reported weight) and for its coupling with the numerator, converge for a given network only if $α>\\tfrac12$ and both corrections…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-10-16","reactions":{}}