Now that we have more than a week of numbers, I thought it would be interesting to do a little number smoothing with the Kos poll. With a little spreadsheet work, it was possible to put together three day and five day numbers which show a smoother trend line than the single day values.
My results for the three day trend are slightly different than Kos's numbers but this is likely due to me not having the raw counts and having to extrapolate based on the rounded percentage and daily sample size. For the 8th and 9th, I'm having to estimate the sample size since it was not provided in the repot on the 11th. Make what you will of the numbers below. It should be even more interesting in another week or two.
Date |
M% |
O% |
Sample |
O-M |
|
3 day M |
3 day O |
3 day O-M |
|
5 day M |
5 day O |
5 day O-M |
9/8 |
47 |
46 |
362 |
-1 |
|
9/9 |
46 |
47 |
362 |
1 |
|
9/10 |
45 |
48 |
361 |
3 |
|
46.0 |
47.0 |
1.0 |
9/11 |
46 |
46 |
364 |
0 |
|
45.7 |
47.0 |
1.3 |
9/12 |
47 |
46 |
361 |
-1 |
|
46.0 |
46.7 |
0.7 |
|
46.2 |
46.6 |
0.4 |
9/13 |
45 |
48 |
359 |
3 |
|
46.0 |
46.7 |
0.7 |
|
45.8 |
47.0 |
1.2 |
9/14 |
45 |
48 |
362 |
3 |
|
45.7 |
47.3 |
1.7 |
|
45.6 |
47.2 |
1.6 |
9/15 |
44 |
48 |
363 |
4 |
|
44.7 |
48.0 |
3.3 |
|
45.4 |
47.2 |
1.8 |
9/16 |
43 |
48 |
358 |
5 |
|
44.0 |
48.0 |
4.0 |
|
44.8 |
47.6 |
2.8 |
If anyone is interested, I can publish the formulas used in these calculations. If Kos is willing to release the numbers (instead of just percentages), I'd be happy to update so that the picture is more accurate.