A 95th-percentile size cap left six-month Nino 3.4 cubic change imbalance uncertain in 1950-2025

We tested how large changes affect a prior negative six-month Nino 3.4 cubic change statistic in fixed NOAA ERSSTv5 data for 1950-2025. A symmetric size cap at the 95th percentile moved the statistic from -0.507 to -0.095. The conditional 95% five-year block interval for the capped value was -0.270 to 0.067. The paired change was 0.411, with interval 0.199 to 0.565. The largest 46 of 906 absolute changes supplied 121.83% of the net negative cubic sum; smaller changes offset part of it. A 99th-percentile cap kept a negative interval. All moving five-year omissions kept a negative uncapped estimate. Thus, size sensitivity did not imply dependence on one five-year window. A cap changes the measured quantity and does not refute the uncapped result. This is a follow-up on the same data, not independent confirmation. Fixed adjustments and thresholds, overlapping pairs, one reconstructed product and prior selection limit inference. Code, aggregate evidence, scalar draws, an internal audit and a static figure are included. Raw monthly data remain external.

ENSO · Nino 3.4 · extremes · influence · cubic change · ERSSTv5 · conditional uncertainty