I tend to agree with Gavin, facts or no facts. Here's why -- the data we do have about giving shows that the percentage of their net income that individual American give has barely budged (ranging between 1.7 and 2.1 percent) over the last forty years. This over the course of a time period involving major tax changes, an explosion of new giving vehicles (including online giving), moments of intense media interest in giving, and the rapid growth of commercial ventures aimed at facilitating charitable giving. Oh yeah, and a period of wealth creation perhaps unparalleled in history.
But the bigger issue is the data we do have. Or, more accurately, the data we don't have.
There are lots of sources - my colleagues and I recently counted more than two dozen surveys, annual counts, facts/figures, and data sources about American giving. These range from highly sophisticated and longitudinally comparable analyses such as GIVING USA to one-off proprietary surveys of high-net worth individuals published by banks to specialty analyses of subsectors such as foundation or corporate giving. And then there are raw data available from the IRS, in the form of nonprofit reports and aggregated, anonymized samples of taxpayers.
However, when you try to triangulate these data you get a mess. Tim Ogden of Geneva Global told me he tried to reproduce the numbers reported in one major bank study, following the same methodology the paper described. His result was off from the reported result by tens of millions of dollars. And Tim triple-checked his math. We also tend to rely on a figure or number that, when you check the sources, all roads lead back to a single analysis. Instead of considering these numbers to be "reliable estimates from many sources," we need to understand that they may be "single estimates reported out many times." You can't cross-check data that all lead back to a single source.
So, if we can't count the simple things - how much do we give - we should figure out how to do so. It may simply be a case of explaining the differences between the different numbers, and why some data should be used for some analyses and other data for other analyses. Because, if we can't count the simple things (dollars), don't get me started on a discussion of measuring things like social impact.