Simpson's paradox in trend analysis: An example from El Leoncito airglow data

J. SCHEER; E.R. REISIN

Buenos Aires

Workshop; 7th Workshop on long-term changes and trends in the atmosphere; 2012

IAGA / ICMA / CAWSES-II

The correct interpretation of trend results does not only depend on the statistical characteristics of the data time-series. In spite of all usual efforts to remove the impact of any other geophysical and instrumental variation, the validity of all the underlying statistical assumptions can hardly ever be ascertained. We show from the example of our mesopause region temperatures at El Leoncito (31.8°S, 69.3°W), how the increased length of the data set alone does not simplify trend analysis. Quite the contrary, it may make unexpected features stand out which require an explanation before definite conclusions on long-term trends can be drawn. While the rotational temperatures derived at El Leoncito from the OH(6-2) airglow band appear rather homogeneous, the O2 temperatures measured with the same instrument and optical filter exhibit features strongly reminiscent of Simpson´s classical statistical paradox, in that straight-forward trends derived from parts show signs opposite to those of the complete data set. The resolution of this paradox requires intercomparison with other instruments and the elimination of new uncertainties discovered in the act.