At least one good thing will come out of the US election. I think we can do away with the argument that the media is biased towards Obama (or Romney for that matter). What I see in the media reporting is a clear attempt to generate a sense that the election is a tossup and that there is no indication of what will happen so, of course, you must stay tuned to the continual coverage!!! However the actual numbers show a clear lead for Obama, but with a chance that Romney could win.

To show this I want to look at some statistics for a minute (promise I will make this as brief as possible). If you recall several years ago, Phil Jones was quoted over and over when he said there was no Statistically Significant Warming for the last 15 years. What he said was accurate and his data backed him up, but it was taken by some to mean that “we can’t tell if it is warming or cooling” which is only partially correct. Others took it to mean there was no warming over that 15 years which is wrong. To understand why, consider the following graph.



Up and down the x axis you have the rate of warming in degrees C per decade. At the blue tick mark at the top of the graph the warming is 0.225 C/decade. When Dr. Jones did his analysis, he came up with a warming of 0.12 which you can see at the green tick mark. The blue tick at the bottom indicates a warming of negative 0.015 (actually a cooling rate). The purple curve represents the likelihood of the real or true warming rate being a particular value. For example, it is fattest at 0.012 which shows that the greatest likelihood is that the true warming rate is 0.12. As you move away from this central value, the purple curve gets thinner indicating that while possible, these values are less likely. At the bottom, you notice that the purple curve crosses into the negative before the minimum value of -0.015. At this point the purple line is very close to the axis indicating a very small chance of this happening, but there is still a possibility. Obviously the likelihood of cooling happening is much smaller than the likelihood of warming but it can’t be ruled out.

Now,lets turn our attention to the election. In my opinion the US election reporting is botching the statistics in a somewhat similar manner. A case in point - CNN released a new Ohio poll yesterday which shows Obama at 50% and Romney at 47%. The error on the poll is + / - 3.5%. They then go on to say how the results are within the margin of error so the race is a toss up and we don’t really know what is going on and anything could happen (but stay tuned until Tuesday).

I have drawn that poll result and while the probabilities are much more complicated for this case, the point is clear. Yes, the sampling error could all work in Romney’s favour and against Obama, but that is not likely. In fact, if the sampling error has Obama rated about ½ % low (so Obama is actually at 50.5%), then it is not possible for Romney to beat Obama. Or, if the poll has Romney ½ % high (i.e. his true poll is 46.5%), then there is no way he can beat Obama. In between is a bit of a no-man’s land and it depends on a number of factors. However I think the point is clear. Yes, based on the statistics there is a chance that Romney is actually ahead, but it is considerably more likely that Obama is ahead.



If the media was really interested in promoting Obama then they would be able to legitimately report the CNN poll as “giving Obama a clear advantage but it is within the margin of error”. Instead they are setting it up to appear to be a dead even race. This does not work to Obama’s benefit but it does help to keep ratings high. As I have long suspected, the media is somewhat biased, but towards the coverage that gives them the best ratings.

Note: Yes, to all those stats majors out there, I have taken some liberties with the stats, but I am trying to keep it somewhat readable. Yes, I ignored 95% levels and a lot more, but if you think you can find a better balance, feel free to write your own. My point stands.