Sample-size slope intervals crossed zero in four ozone invertebrate samples from a 2024 data release

We tested whether smaller reported sample sizes were linked to more negative ozone response ratios for invertebrate abundance and reproduction. We used 145 records from 34 papers in a public 2024 data release. We balanced endpoints within papers and gave each paper equal weight. Four tests split Europe and the Americas by publication years through 2005 and later years. Slopes per doubling of sample size were -0.427, -1.216, -0.006 and 0.086. All four corrected intervals crossed zero. The later Americas test had only four papers. The pooled adjusted slope was -0.084, with ordinary 95 percent interval -0.311 to 0.142. Endpoint, field-exclusion and facility checks did not give a clear pooled association. This result does not establish that publication bias is absent. Reported units, mixed endpoints, shared facilities, unmeasured study differences and selective reporting limit the tests. No climate interaction was tested. The package includes code, source hashes, aggregate results, a figure and independent arithmetic checks.

ozone · invertebrates · sample size · observation bias · regional replication · null result