CONTROLLING MY OWN POLITICIAL BIAS

Becky Quick on MSNBC this morning came up with a new thought, at least it was a new though for me this morning. She said something like the deviation from the trend line was small suggesting that a poll result was "probably" accurate. This may be a good way to analyze poll results. Large numbers of polls give different values. However, if observer averages the values of a time series and plots them on a simple x-y system of coordinates, for example a daily average poll result, this generates a wavy line. If the observer plots the average of averages, this dampens the waves—a graphic Fourier solution. For example the mean value of Monday, Tuesday, Wednesday is plotted, Then the mean value of Tuesday, Wednesday, Thursday are plotted and so one the resulting line or curve  takes on a smooth appearance that can  be projected to the next point. I believe her point was that any individual poll result that deviates markedly from that line would suggest it is inaccurate and those points that fall on that line are accurate. Where this count is when FOX News cites polls favoring Romney and MSNBC cites polls favoring Obama then the answer to which one the listeners should believe is the result that closest to the trend line. As a liberal with biases, I tend to believe MSNBC and disbelieve FOX but from now on, I will believe the poll result that falls closest to the projected trend line.