Editor’s note: Updated January 5, 2015 to add more details and highlight changes in the data since this explanation of methodology was published on October 5, 2014.
The initial version of this project used ZIP code data for states and metropolitan areas, which undercounted the amount filers gave for those areas. The interactive has been updated with county-level data that does not undercount the amounts. The methodology below reflects and further explains this change.
The Chronicle’s “How America Gives” report was based on Internal Revenue Service income-tax data.
The analysis is based on the tax returns filed by Americans who itemize their deductions, including charitable gifts. The Chronicle’s data for 2012, the most recent year available, cover about 80 percent of the money that individuals gave to charity in 2012, as reported by “Giving USA. ” (Data from 2006, presented for comparison, represent about 74 percent of charitable giving.)
The data were aggregated by income ranges for ZIP codes and counties to show giving patterns across the country.
The figures cited reflect total adjusted gross income (AGI), which is calculated by deducting certain expenses, including retirement-plan contributions, unreimbursed business expenses, medical expenses, and alimony.
Because the IRS provides aggregated totals for each geographic area and income bracket, we estimated the midpoint, or median, of each income bracket through linear interpolation, a statistical method often used to fill gaps in data.
The “giving ratio” is the ratio of itemized charitable contributions to adjusted gross income.
Religious-affiliation statistics come from survey data from the 2010 U.S. Religion Census: Religious Congregations and Membership Study, by the Association of Statisticians of American Religious Bodies, and from the Association of Religious Data Archives.
The political leanings of states and counties is from The Guardian’s election-results data. The project also uses some demographic information, like age and race, from the U.S. Census Bureau’s American Community Survey estimates from 2008 to 2012.
The IRS data do not paint a perfect picture of personal giving.
Adjusted gross income does not account for regional differences in the cost of living or other circumstances that can affect people’s ability to give. For example, some states have low or no income tax, while others have marginal tax rates exceeding 10 percent. As a result, some giving ratios could be lower as a result of tax policies or costs of living, rather than individual generosity.
The 2006 and 2012 aggregated ZIP code and county data provided by the IRS may include late or amended returns from previous years.
To calculate state and metropolitan-area totals, The Chronicle’s analysis used ZIP code data to generate the metropolitan and state information. However, to protect the identity of tax filers in sparsely populated ZIP codes, the IRS withheld some data where there were fewer than 20 returns. As a result, state and metropolitan totals shown may be lower than the actual amounts. While state- and county-level data are not subject to these privacy concerns, the 2006 data did not include AGI for those who itemized, so that data could not be used. [Editor’s note: On January 5, 2015, The Chronicle added 2012 state- and metropolitan-level data (and removed the 2006 comparisons) to give as full a picture as possible of giving at those levels.]
Because the IRS limits the charitable deductions filers can claim on their returns, it is likely that some filers gave more than they reported on their tax returns.
Because the data are based on income rather than net worth, some wealthy individuals with low or no annual income could fall in the lower income brackets or might not file income-tax forms at all. And any giving that is not reported on income-tax returns is not reflected in these data.