Mars Close Approach
A number of people suggested that the close approach of Mars late in August 2003 would be a good time for a GCP prediction. On the 27th, our neighbor would be the closest in 60,000 years. One suggestion, by Prof. Adrian Patrut of Romania, had sufficient particularity for a clear prediction, which he felt would have a medium likelihood.
The formal prediction followed this prescription exactly, specifying the standard analysis. The outcome, with 49 eggs on the 27th and 50 on the 28th, is Chisquare is 6438.3 on 6480 df, and p = 0.641. Thus the test statistic is slightly in the wrong direction, despite a strong positive trend at the beginning of the period.
This event was also the first in an exploratory set of analyses designed to meet certain critical objections. Prior to the formal analysis, I gave the specifications to Prof. Dick Bierman of Amsterdam, who then extracted the data and four "decoy" datasets from elsewhere in the database. He provided me with the five datasets, which I was free to process any way I wished, with the task of ranking the five from most to least likely to be the real data. I did two kinds of analysis, and in both cases gave the target data rank # 2 of 5. The two following figures show the data from a Means-based analysis (similar to the standard formal approach) and a Variance-based analysis. The real data are the sequence plotted in yellow, set E.