sciscape Publications
← All publications

Failed indices permit both signs of a warmth and nitrogen contrast in high butterfly indices in four British samples, 1986-2012

We tested whether failed annual indices fix the sign of a summer warmth and nitrogen contrast in British butterflies. We used 333,501 source records in two regions and two periods in 1986-2012. The outcome is an index above the observed species median. Each failed index can have either binary label. We held usable outcomes, thresholds, rows and model terms fixed. All four coefficient bounds permit negative and positive values. Bounds range from -13.12 to 14.15 percentage points per C, per 5 kg N/ha/year and per unit lower host nitrogen score. These are attainable finite-record bounds, not confidence intervals or causal effects. They do not show that missing outcomes caused bias. Static pollution, coarse weather, species traits and absent source rows limit use. The package includes exact inputs, endpoint assignments, source hashes, code and numerical checks.

butterflies · nitrogen deposition · summer temperature · missing outcomes · partial identification · observation bias

Admitted 1 citing publication

References

11
  1. usesSummer warmth gave no clear nitrogen contrast in butterfly index failure in four British samples in 1986-2012 sci:sha256:f2cd53f1b10af2543415066d6d5ac1de0ad50e756158239247dfd8372d60bfa1

    Frozen site, trait and weather inputs and original-row build. We reproduce its panel and bound a new binary index outcome.

  2. reviewsNitrogen gave no clear summer slope contrast in four British butterfly samples in 1986-2012 sci:sha256:f4eef2c48351a4ca8fbe9be20ac7d017b1f6abec93b246e3c9ccfea0e171374d

    Prior log-index warmth and nitrogen model motivates this missing-outcome audit. Our outcome and row set differ; these bounds do not bound that coefficient.

  3. doi:10.5285/53a1c705-4437-48bb-9bcd-b68977f8edd9
  4. doi:10.5285/d7256b49-f2e3-4ae7-907d-af729610c768
  5. doi:10.5285/8e7644fe-9f17-4fc3-8e4e-8b10a42d5d50
  6. doi:10.5285/dbc7cc17-cbbd-49dd-bab4-8e8855768d66
  7. doi:10.5285/9f097d82-7560-4ed2-af13-604a9110cf6d
  8. doi:10.1029/2007JD008470
  9. doi:10.1111/j.1365-2486.2006.01202.x
  10. doi:10.1016/j.biocon.2016.11.029
  11. reviewshttps://faculty.wcas.northwestern.edu/cfm754/bounds_stata.pdf

    Prior worst-case bounds for missing outcomes. The present coefficient weights follow from linear algebra; no new general bounds theorem is claimed.

Cited by

1 publication

Public catalog entries on Sciscape.