Toying with numbers
I had some free time the other day, so I started looking at the corn production numbers and the Statistical Methodology at the end of the Production report. It states that chances are 9 out of 10 that the difference in yield from the August Production report will not exceed the final yield by 6.1%, so I wondered if a multiple(1,2,3...) of the 6.1% multiplied by the last year state yield would tend to show up in this August Production numbers. Starting out, I took 2016 Kansas yield of 142bu/acre times .939=133.34bu/acre. Ks production number was 133. (I got the .939 by subtracting .061 from 1.00) The 2017 Kansas yield/acre decreased by a factor of 6.1%. In S DAK, I took 2 times the 6.1% or 12.2% off of last years corn crop-161 x .878=141.36bu/acre. The report said 140. Missouri, the yield was the same. In Pa, I multiplied the 6.1% by 4 and took last years yield of 129 x 1.244=160.48bu/acre.. The report said 160bu/acre. In all, there were 32 states listed on the corn production report. Of that, 8 out of 32 were close to 0 difference when multiplied by a factor of 6.1% or had the same yield as last years. 6 out of 32 were plus or minus 1 bu/acre when multiplied by a factor of 6.1%. 8 out of 32 were around 2 bu/acre plus or minus when multiplied by a factor of 6.1%. The rest is 3, 4, or 5 bu/acre which will get pretty close to the middle of the spread with lower yield states. To note, a 6% spread in Iowa or Ill would amount to 12bu/acre or 6% x 200bu/acre. I am not a mathematician or statistician, so I have no way of knowing if this is significant or not-just playing with numbers I do find it interesting that the the numbers tend to cluster around a multiple of 6.1% as I would expect a more even distribution. I also find it interesting that the seven states with record yields are none of the ten objective states. 21,700 interviews with all the data sorting and field observations would entail a lot of work getting ready for a report. I'm not making any judgments, only presenting what I see.