nomis5 - 5:48 pm on May 29, 2012 (gmt 0)
You can at least try to test. For example if you take a couple of reasonably steady pages which in your case may have a 5% decline in income each month, you can then test a change on one page and guage its effect against the non-changed page.
But I undertand your difficulty because G is shaking everything around so frequently it's very difficult to judge cause and effect.
Recently I slowly migrated some pages from 336x280 to 300x250 because the 10% of the 300x250 i already had were starting to show improved income over the other format. I have learnt that the results from a change over the first few days should be ignored - it's almost as if G is trying to disguise the effect. But after a week or so the results were clear, 300x250 performed better on my site.
Volume is also key, if you only have a small amount of page inpressions it is almost impossible to judge cause and effect.