Michele Margolis and Michael Sances write:
Conservatives and liberals are equally generous in their donation habits. This pattern holds at both the individual and state level, and contradicts the conventional wisdom that partisans differ in their generosity.
What about the claim by Arthur Brooks that conservatives give more? Margolis and Sances write:
We are not the first to ask whether partisanship affects giving. In 2006, Arthur Brooks made headlines with a provocative finding from his book Who Really Cares: despite stereotypes of liberals caring more about the poor, conservatives were purported to be more generous when it comes to giving to charities. These results stirred the political pot by taking “bleeding heart liberals” to task for their stinginess when it comes to their own money. . . . we demonstrate that these results are not robust, and appear to be driven by a non-traditional question wording for identifying liberals and conservatives. After correcting for this problem, there is no statistical difference between conservative and liberal giving, conditional on observable characteristics. Further, when we use partisanship rather than ideology to measure liberalism, there is no statistical difference in giving, regardless of whether we adjust for observable characteristics.
How did Brooks screw up?
One reason for this anomaly could be the unorthodox way in which the SCCBS [the 2000 Social Capital Benchmark Survey] asks about ideology, which differs from the standard phrasing used in the ANES and GSS. The ideology question wording in the 2000 SCCBS reads: “Thinking politically and socially, how would you describe your own general outlook–as being very conservative, moderately conservative, middle-of-the-road, moderately liberal or very liberal?” (emphasis added). It is likely that this wording compels many economic liberals to identify as social conservatives, and many economic conservatives to identify as social liberals; because social liberals tend to be wealthier, this would explain why liberals in the SCCBS are wealthier. Certainly the wording affects the distribution on the ideology question. Whereas 33 and 34% of respondents in the 2000 ANES and GSS respectively identified as conservatives, the percentage jumps to 43 in the SCCBS. For these reasons, it seems reasonable to ask whether these findings can be replicated using another dataset.
Where do liberals and conservatives give their money?
While levels of giving are roughly equivalent, liberals are much more likely to do- nate to secular organizations, and conservatives are more likely to donate to religious causes, especially their own congregation.
And when do they give?
Charitable contributions fluctuate based on the political landscape: Democrats (Republicans) donate less money when a Republican (Democrat) occupies the White House. Conversely, having a co-partisan in the White House increases the average and total donations to nonprofits at the state level.
I’m impressed by this work, partly because a few years ago when I saw Brooks’s claim, I contacted him and asked for details on what he did, and then I threw the problem to some students to replicate it. They got tired and never did it.
P.S. Recently, Arthur Brooks has been having some trouble with the General Social Survey. Working with data can be difficult!
{ 22 comments… read them below or add one }
“While levels of giving are roughly equivalent, liberals are much more likely to donate to secular organizations, and conservatives are more likely to donate to religious causes, especially their own congregation.”
Much of donating to one’s own congregation or even a denomination hierarchy is “charitable” only in a legal sense. Care and feeding of a pastor and facility from which one derives a personal “benefit” is hardly charitable. The YMCA is a tax exempt institution but my membership fee is not deductible because I derive a personal benefit from membership. The primary difference between that and a church is the special status regarded religious institutions.
“Care and feeding of a pastor and facility from which one derives a personal “benefit” is hardly charitable. The YMCA is a tax exempt institution but my membership fee is not deductible because I derive a personal benefit from membership.”
Here’s the thing. The facility and the pastor aren’t charging a membership fee to perform services. They do it for free, and won’t exclude someone from giving those services if they don’t want to donate. There’s also the slight problem that if you treat the donations as some sort of ‘de facto’ “membership fee”, you also have to realize that this “fee” is based on how much the individual in question wants to give the pastor/facility. Somebody might give him $10/month, others may prefer giving him more or less. The pastor/facility can’t really say otherwise if the masses pay less than their actual value.
Plus, the congregations that I go to also hold fundraisers in addition to their regularly scheduled donation boxes. Clearly those de facto membership fees aren’t raising enough money.
So pastor’s price discriminate and cross subsidize. Doesn’t affect the point that it’s mostly payment-for-services-rendered, not charity.
Ugh, “pastors”, not “pastor’s”
Although it varies quite a bit from congregation to congregation, many churches spend a significant amount of their budget on helping poor and otherwise needy people in the congregation, or even out of it. They also financially support church volunteer efforts, educational efforts (usually at below cost), food drives, etc., which result in indirect charitable assistance that doesn’t show up as a budget outlay. E.g. Catholic Charities, the Salvation Army, etc.
GiT’s assumption that contributions to one’s own congregation are entirely, or even largely, for a “private” benefit to that congregation seems unlikely to me without data to support it.
The Brooks study was full of red flags. I noted how its definitions of religiosity biased the results: http://gnashq.blogspot.com/2007/09/do-religious-people-give-more-or-just.html. I never understood why anyone outside the conservative movement ever took it seriously.
The difference in composition of charitable giving is huge, probably the most critical thing, for deciding if there is a difference between the two groups’ charitability. Consider Giving What We Can or GiveWell, where by making a hard choice for a metric that mostly proxies for increased human well-being, disability adjusted life years, they can get rid of a lot of the wasteful and inefficient social signaling and status motivations for giving. When they examine charities that way, they see a many orders of magnitude difference in the efficacy of the most effective charities vs. the least. In my view, unless you get right down to the groups’ differences in terms of motivations and raw efficacy, then you can’t say much. Giving the same dollar amount hardly implies similar levels of charitability.
The problem with your critique is that “correcting for this problem” creates its own problems.
Additionally, Brooks’ conclusions are supported by other data. For example, The Chronicle of Philanthropy ranks states by charitable giving based on tax data. They find a strong correlation between conservative states and giving. There are quibbles with this (some people don’t deduct charitable contributions, not everyone itemizes, etc.), but the trend supports Brooks argument.
The best those on the left can say is that giving is about equivalent. Since the primary attack against the right is that they are “selfish,” hoping to tie conservatives in charity is ironic.
“Since the primary attack against the right is that they are “selfish,” hoping to tie conservatives in charity is ironic.”
I thought the “primary attack” was that liberals’ supposed concern for the less fortunate was hypocritical as shown by stinginess in theyr persoal giving.
I have a very storng Beysian proir that ideology would not be correleated with charitable giving if botth are well measured,
I was discussing the left’s “primary attack against the right.” Your comment relates to the right’s critique of the left. You may be right, but your response is non sequitur.
Robert Duvall (“Tender Mercies,” “The Apostle” has been critical of Hollywood (liberals?), saying that Hollywood does not understand religion or religious people in America (conservatives?).
I have always considered myself a social liberal and a fiscal conservative, although I am probably less liberal since I became a husband and a parent. I give mainly to my church and the Salvation Army as well as a number of smaller secular programs.
Religious people give more than money, they give time, they give encouragement, they listen. Non-religious give art to museums (at absurd mark-ups) and write it off on their taxes; they sponsor dance companies for their lovers/partners and write it off on their taxes; they initiate legislation to increase public funding for the poor, the sick, and the disabled who are not getting enough funding because of the taxes some people avoid because of their generosity to museums and the arts.
Non-religious people seem to have a mechanistic explanation for why religious people give: they “give” out of guilt, to appear righteous, to comfort their own feelings of inadequacy. Non-religious are generous to the poor, the sick, the disabled but often with taxpayer money.
BTW charitable deductions are grossly misleading. Years ago I worked alongside a man who claimed twice what I claimed for charity on our tax returns. We both made the same annual salary. The difference was that I actually gave to charity; he gave nothing – it was all a lie. And yes, he got away with it.
“Religious people give more than money, they give time, they give encouragement, they listen. Non-religious give art to museums (at absurd mark-ups) and write it off on their taxes; they sponsor dance companies for their lovers/partners and write it off on their taxes; they initiate legislation to increase public funding for the poor, the sick, and the disabled who are not getting enough funding because of the taxes some people avoid because of their generosity to museums and the arts.”
I’m sorry, but do you have a point that’s something other than a pack of baseless stereotypes?
The Table 1 column i Margolis-Sances 0.639 coefficient on the conservative variable is closer to the Brooks estimate of 1.03 than it is to the null hypothesis value of 0.00. So if Brooks’ estimate is incorrect, then isn’t it also incorrect to claim that “conservatives and liberals are equally generous in their donation habits”?
Margolis and Sances were working with five percent of the observations that Brooks worked with, so if their data cannot let us reject the null hypothesis of no effect, then maybe their data cannot let us reject the Brooks claim of a 1 percentage point difference, either.
Therefore, if we want to prove Brooks incorrect, wouldn’t it be better to directly test whether the GSS data permits rejection of Brooks’ estimate of a 1 percentage point difference?
Exactly. The “replication” muddles the issue by confusing regression coefficients (liberals and conservatives are equally generous _holding constant religiosity_) with simple descriptive claims (“Conservatives and liberals are equally generous”), and by confusing weak evidence in the GSS data with evidence of no difference. Brooks may be wrong for a variety of other reasons, but nothing here provides any good reason to doubt that conservatives give 1% more of their income to charity than liberals do.
Thanks. Based on the coefficients and standard errors reported in Table 1, the 95 percent confidence interval for the conservative variable in Table 1 column i should be about [-0.2, 1.5] and the 95 percent confidence interval for the conservative variable in Table 1 column ii should be about [-0.2, 1.3]. The Brooks estimates of 1.03 and 1.10 fall within these 95 percent confidence intervals.
Brooks’ work was pure ideology masquerading as research, as is obvious from the number of serious questions it begged about the meaning of the evidence it worked with. But beyond these evidential questions about the giving of money there is also the larger problem of defining giving as what one does at leisure with money or time that one has earned. There is another kind of giving that liberals do far more frequently than conservatives: giving one’s life to a nonprofit or public sector career. This amounts to sacrificing income one would otherwise accrue, income one would otherwise have available to then give. I haven’t done the research, but I’m willing to bet that had Brooks simply counted foregone income as giving he would have found that liberals are far more giving than conservatives.
Brooks’ idea of what counts as giving is misguided; his concept of selfishness is misguided as well. He does not recognize a basic economic principle: that like any other choice, all giving includes some element of selfishness. The difference between the giving of one’s free time, or of one’s money, and the giving of one’s life is arguably that one gets more selfish pleasure in the multiple punctual givings of things that one owns than in the giving of one’s self that a teacher or a doctor does.
But Brook’s primary finding wasn’t that conservatives give more than liberals but that the religious give more than seculars. The fact that outside of the black community there are virtually no religious liberals is what is most interesting to me.
I suspect that if the study were rerun just on social values it would find a e difference in social conservatives’ favor.
I’m a religious liberal. We’re not THAT rare. Google “religious left” sometime.
“It is likely that this wording compels many economic liberals to identify as social conservatives, and many economic conservatives to identify as social liberals…”
It is more likely that this bit of subterfuge and deflection is the most disingenuous thing that I’ve read in months, if not longer. Conservatives–social or fiscal–would never self-identify as liberals. Liberals of any variety would never self-identify as conservatives. This is a toothless and lame attempt to dismiss Brooks’ findings and it is obvious that no amount of data or analysis could convince liberals that they are stingy with their own lucre, yet very, _very_ generous with mine.
It wouldn’t impress me if it would be found conservatives donate more to charity.
Charity is itself a somewhat conservative way to see redistribution, the giving away of those capable of producing wealth to help those who are less capable.
Liberals don’t see redistribution as an act of kindness and generosity, but of fairness, and not at an individual level of charity, but at a colective level of sharing the social product to allow the worse of to have minimun conditions of living.
I believe liberals would be more participative in voluntary, social and community work and stuff alike, while could be conservatives more generous in actual direct charity.
“One reason for this anomaly could be the unorthodox way in which the SCCBS [the 2000 Social Capital Benchmark Survey] asks about ideology, which differs from the standard phrasing used in the ANES and GSS … It is likely that this wording compels many economic liberals to identify as social conservatives, and many economic conservatives to identify as social liberals”
What’s the evidence that this claim is true? That the ANES and GSS use different wording doesn’t mean that their wording provides a more accurate result.
Couldn’t you consider NOT writing off a deductible expense as charity? I don’t really consider writing off charity as charity, as you are not actually paying anything out of pocket, and thus don’t write it off myself.
I think it would fair to compare “donations” to government in this way(stereotypically liberal) to donations to churches (stereotypically conservative), as while both benefit others, they provide personally beneficial services, have management costs, and can both be abstractly considered membership fees.
It would be interesting to see the numbers adjusted to subtract donations written off, as a comparison.