tag:blogger.com,1999:blog-2766957177339141031.post4267399895031715449..comments2023-06-06T10:13:14.087-07:00Comments on Amanda Santer: Zen and Inspiration: A cash to committed campaign - and I don't mean Direct Debit.....Amanda Santerhttp://www.blogger.com/profile/15775571819595962482noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-2766957177339141031.post-84473116900410998762015-08-06T02:54:01.194-07:002015-08-06T02:54:01.194-07:00Hi Mike, lovely to hear from you and thanks ever s...Hi Mike, lovely to hear from you and thanks ever so much for your insight on the robustness. I am aware of its flaws and absolutely when we do it again we will look at making the testing more rigorous. As you'll know, sometimes it is more that you'll know a certain activity should have a positive impact or won't do any harm, but you are absolutely right and next time I look at this I will look at the criteria in more detail. If I am being honest I was a little surprised by how considered people were .. So though not statistically robust it appears to have had a positive impact. Amanda Santerhttps://www.blogger.com/profile/15775571819595962482noreply@blogger.comtag:blogger.com,1999:blog-2766957177339141031.post-54529347576761575142015-08-06T01:13:45.043-07:002015-08-06T01:13:45.043-07:00Hi Amanda,
Interesting post as ever and nice to s...Hi Amanda,<br /><br />Interesting post as ever and nice to see someone sharing results of something like this. However, I wondered whether you had thought about using a random control group of supporters who matched the overall selection criteria - where you did not supply this group to the agency for calling. This might have helped to interpret the impact of the calling and get over the inherit biases of the 3 groups that you mention:<br />i) Those pledging at the call – maybe these people are just warmer to your cause and therefore would have a higher response rate to subsequent mailed appeals<br />ii) Those who declined the pledge at the call – as above but in reverse<br />iii) Those who were not contacted – there are a multitude of reasons for non contact such as goneaway, not around when the calling was conducted, refuse to answer un-known calls etc all of which could have an impact on subsequent responses.<br />If you had a random control you would have been able to see if there was a real difference between those who you attempted to call (as one group) against those who you didn’t.<br /><br />Obviously available number may have been an issue. If so, widening the selection criteria to overcome this might have given you a chance to use a control group and hence get some more robust results.<br /><br />MikeAnonymoushttps://www.blogger.com/profile/09524289166486455516noreply@blogger.com