First, I don't know of any credible person who argues that partisanship or other political attitudes influences charitable giving. That aside, there's a thread on the OMB citing some "study" that claims that Reps donate more than Dems to charities.

Among the sources cited is the General Social Survey. Now I just happen to have the GSS cumulative file (1978-2006) dataset on my harddrive and decided to poke around the data. What I found shouldn't surprise anyone who gives that claim a few seconds worth of thought...

Supporting the claim. If you crosstab party on financial donations to charity, it is true that Republicans are about 15-18% more likely to contribute to charity (depending on the question asked).

The problem is that this does not control for frequency of church attendance and income, both of which are known to be associated with party identification (as well as charitable giving). Any credible study (which the one cited on OMB is clearly NOT) would control for these factors. If partisanship then had an independent effect separate from religiosity and income, the claim might have more credence.

So I created a logistic regression model that controlled for several of these other factors. (Note: Logit regression simply allows us to estimate the probability that a person will donate to charity.) First, here are the results with ONLY partisanship included:

Iteration 0: log likelihood = -1402.7815
Iteration 1: log likelihood = -1374.1216
Iteration 2: log likelihood = -1373.4236
Iteration 3: log likelihood = -1373.4235

Number of obs = 2692
LR chi2(2) = 72.06
Prob > chi2 = 0.0000
Log likelihood = -1373.4235

COEFFICIENTS (P-value)
Democrat: -.390 (0.000)
Independent: -1.001 (0.000)
Constant: 1.672 (0.000)

Republicans are left as the baseline in the model, meaning that the coefficients tell us how much lower (or higher) Democrats and Independents are to donate than Republicans. The negative signs on the coefficients mean that both Democrats and Independents are less likely to donate than Republicans. Ignore the constant.

The numbers in parentheses indicate the statistical significance of these results. They are near zero, indicating that there is an almost no chance that the coefficients are zero (which would be the case if Democrats and Independents were identical to Republicans). In other words, the coefficients are statistically significant. Finally, the model predicts that 84.1% of Republicans will donate to charity while only 74.6% of Democrats will do the same.

NOW, here is what the model looks like when you control for income and religion...

Iteration 0: log likelihood = -1226.2211
Iteration 1: log likelihood = -1097.4675
Iteration 2: log likelihood = -1077.4941
Iteration 3: log likelihood = -1074.9701
Iteration 4: log likelihood = -1074.9137
Iteration 5: log likelihood = -1074.9136

Number of obs = 2386
LR chi2(4) = 302.61
Prob > chi2 = 0.0000
Log likelihood = -1074.9136

COEFFICIENTS (p-value)
Democrat: -.030 (.815)
Independent: -.499 (.001)
Income: .00000302 (0.000)
Attend church: .1998 (0.000)

Notice that the coefficient for Democrat shrinks considerably, and that the p-value of .815 is no longer statistically significant. What that means is that there is an 81.5% chance that Democrats are no different from Republicans in the probability that they will donate to charity. Independents, meanwhile, are still much less likely to donate (indicated by the negative and statistically significant coefficient) while those with high income and those who frequent church the most are more likely to donate (indicated by positive and significant coefficients).

Finally, just to make the point crystal clear, AFTER controlling for income and church attendance, about 85.7% of Republicans donate while about 85.4% of Democrats donate. Obviously, Democrats donate to charity just as often as Republicans.

This is something that anyone could have figured out on their own considering that there is no valid theoretical reason to think that party loyalty determines charitable giving. But it shows the need to avoid blindly trusting the crap that is spouted on blog sites and by partisan hacks on talk radio or cable news shows.

So, are trying to say, Dems are just better people contrary to popular belief? :unsure:

Nope. I'm saying that there is absolutely no difference in charitable giving between Dems and Reps. Church attendance and income are the real explanations for charitable giving.

Dems care more for humanity than Rebs.

When you say "after controlling for church attendance and income," do you mean that you only include Republicans and Democrats who attend frequently and have higher income, or what?

IOW, what kind of control are we talking about? It sounds like you are saying that people who attend Church frequently and have higher incomes donate more to charity (which makes sense). I just can't tell how you got there from here.

Explaining the intuition behind how regression models mathematically control for multiple variables can be tricky. But the short answer is "no", I do not limit the model only to frequent churchgoing and high income partisans. I include everyone -- low income, high income, non-church goers, frequent church goers, Reps, Dems, Ind's, and all possible combinations of these variables in the analysis.

Income is measured in terms of absolute income. Frequency of church attendance is an ordinal scale with 7 or 8 categories from never attend church to attend multiple times a week. So again, there are all possible levels of church attendance and income included in the model.

The model does show that people who attend church frequently are more likely to give. It shows that those with higher incomes are more likely to donate. It also shows that Republicans are not significantly more likely to give than Democrats (after control for religion and income).

The point of showing the model with only partisanship included was to illustrate that the apparent effect of partisanship (the claim made in the OMB thread) was spurious to income and religion. We know that Reps tend to have higher incomes and they tend to be more religious than Dems. Religion and income, NOT party is what explains charitable giving, which is proven when the party effect disappears after adding income and religion to the model.

But then there is also a correlation between religion and income and party affiliation. That doesn't mean party is the driving factor in charitable giving, but it does indicate that those most likely to give (higher income religious folks) tend to be Republican, does it not?

IOW, while your point is well taken, aren't we in some danger of picking nits?

As ice cream consumption increases, the number of drowning deaths increase. Of course, ice cream consumption increases during the summer months, when people are swimming and therefore more likely to drown.

There's correlation between all three of those variables. The question is whether there is INDEPENDENT variation between ice cream consumption and drowning deaths after we control for the variation/correlation between summer months and drowning deaths. If summer months explains all of the variation between ice cream consumption and drowning, AND it furthermore explains a lot of variation that is independent of ice cream consumption, then we have to accept that the true explanation is summer months. IOW, we have to accept that all of the correlation between ice cream consumption and drown deaths is spurious (explain by) summer months, and that no real relationship exists.

Here's a quick diagram that hopefully illustrates this using three variables -- skill variety, absences, and autonomy. The red area (which includes the black area) is the correlation between "skill variety" and "absences". The blue area (which includes the black area) is the correlation between "skill variety" and "autonomy". The black area is the correlation shared by ALL THREE of the variables -- skill variety, autonomy, and absences. Regression will purge out or eliminate the black area, leaving only the red and blue areas (which are represented by the coefficients). In this example, there is correlation between absences and autonomy BUT there is also INDEPENDENT correlation between each variable and "skill variety".

Now imagine that the "absences" circle was larger and overlapped more with "autonomy" to the point where there was no blue area (e.g., the blue area was all black). The result would be a coefficient for "autonomy" that was near zero and statistically insignificant because that variable would have no independent effect...it's relationship to "skill set" would be entirely determined by "absences". This situation is almost exactly what we found when looking at charitable giving.

So yes, Republicans give more than Democrats on average BECAUSE Republicans tend to have higher income and attend church more often. It's the church attendance and income that CAUSES the charitable giving, and any relationship to party is illusory and meaningless.

I'm aware of the difference between correlation and causation, but the point is that you say "illusory and meaningless." I wonder if there is a reason folks with higher income and greater religious affiliation vote Republican.

And I don't have a dog in the hunt -- I'm not a Republican. But the point is, without looking BOTH at why they give AND why they vote Republican, isn't it a bit premature to say it's "illusory and meaningless."

Put another way, while the study referenced doesn't prove causation, the correlation could be more than coincidence (as with ice cream and swimming pool deaths, for example).

EDIT: your picture is too damn big.

My brain hurts. :huh:

Let me simplify it -- I want to know why greedy selfish lazy heathens want so badly to vote Democrat nr6 b667ur

High income people vote Republican because the GOP offers tax cuts. Frequent church-going Christians (especially self-identified evangelicals) vote GOP because the party offers social policies it likes. What does that have to do with charitable giving?

Btw, technically this isn't about the difference between correlation and causation. It's about shared correlation among multiple variables. I can explain ALL of the Party-Charity correlation with Income and Religion. There is NO independent correlation outside of the effect of those latter two variables. So how does partisanship help us explain charitable giving?

Let me put this another way--

High income, frequent church-going Democrats contribute to charity equally as much as high income, frequent church-going Republicans.

Low income, non-church going Republicans do not contribute a lot to charity, just like low income, non-church going Democrats.

IF there was something unique about party that explains charitable giving then you would expect that high income, frequent church-going Republicans would contribute MORE than high income, frequent church-going Democrats. That's not the case, which means that the effect of party is meaningless when explaining charitable giving.

All I'm saying, (and I've already admitted I'm not the sharpest tool in the shed on these boards) is that there is no way in Hades hen house I could have figured this out on my own, using the steps you took to get there. I only went as far as Geometry and simple calculus in school/college, and I b-a-r-e-l-y passed calculus. I get what you're saying, but I could not have done that math on my own if my life depended on it. :)













Smarty Pants. nr6

What do you base that on..any fact`s to back that up or just you`r opinion?

LOL! I could teach you the math if you took my undergrad statistics class.

But I wasn't talking about the math when I said anyone would have figured this out. I meant that it doesn't take any specialized knowledge to know that income and church attendance are the real explanation here. Someone at the OMB actually made this very point in that thread, and I just followed it up by showing the empirical results from one of the surveys that was cited by this "study".

Actually, I just added this example to my undergrad course! It gives me a certain pleasure knowing that a bunch of 19-20 year old students will be debunking this "study" each and every semester. fgtb76

Yeah, I know. I saw the thread title, but I didn't read any of it. Only so much of the banter I can take every day. Mostly, I just go on personal experience, and from people I know. And to be honest, the subject of charity associated with the topic of political party affiliation really does not come up in conversation very much. That said, when I was going to church regularly, I knew people from both parties that attended my church, and I knew some that tithed and others that didn't.
At the end of the day JDave has a point. Generally speaking, it seems that Reps usually have higher income, and an easier time of tithing or giving to charities vs. Dems. But if you're going off of tithing, then it's supposed to be 10% regardless of income. 10% is 10% no matter how much you make. But the more you make, I would think the easier it would be to make your tithing commitment every month, regardless of party affiliation.

And frankly, tithing shouldn't be included as "charity" in the sense that we're discussing it here. Some of the tithes definitely go to charity, but a lot of it goes to upkeep of the church and other administrative purposes. Theologically, it is mandated by the Bible. That strikes me as a qualitatively different type of decision than choosing to donate to a charity on a more (theologically) voluntary basis.

On the other hand, excluding tithing could skew the results in favor of liberals because I'm sure there are a lot of people who view their tithes as their charitable giving. Non-churchgoers (who are more likely to be Democrats and liberals) could be more likely to engage in these other forms of giving.

:dunno:

Yet another reason why it's silly to focus on politics and charity when the simple, straightforward explanation involves income and church attendance.

Agreed. But were you using tithing as part of charity in this example, or just churchgoers? I may have misunderstood. I know you considered both churchgoers, frequent churchgoers, and non-churchgoers. Did the OPs point on the OMB have anything to do w/all of this? I think I lost the point somewhere.

Anyway, I would say (again generalizing) that those that attend church regularly are more charitable, as it is a taught principle in Christianity. At least in most all of the churches that I've attended.

** leaves to go give the other thread a gander**





**throws up a little in mouth at the thought of it**

iy88767

The examples in the thread you're referencing, the two *ahem* samples aren't even in the same GD state!! I haven't even read that whole article yet, but how can you compare the land of fruits, nuts and flakes to a midwestern SD city?? The laws in the 2 places are completely different. Sorry, but this alone makes me want to raise the BS flag. There is a reason SF is called the city of bum poop, and it doesn't have to do w/rdonham alone. If the city I was living in was condoning/enabling homelessness the way SF seems to be doing, and STILL couldn't keep bum poop off the streets, then I'd probably have a hard time giving money to charities too. Might still do it, but those charities may be vastly different causes. Pet rescues and the like...

The dependent variable includes all forms of charity, so if the person considers tithing to be a form of charity then it would be included.

Yeah, this is pretty much a response to the OMB thread.

You're right that those who frequently attend church are more charitable as taught in the Bible. That's kind of my point...politics has nothing to do with it.

To be fair, in that instance correlation and causation may overlap :D

Yeah, there's so much wrong with that article and the cited "study" that it's hard to take any part of it seriously. Another problem is that it also does the red state/blue state comparison, which is questionable because we're talking about why individuals choose to donate.

That's why I wanted to show the individual level data using one of the datasets that was cited, to get a nice, clean model for what causes charitable giving.

What's discouraging to me is that I cranked out this model is just a few minutes and wrote it up in less than 10 minutes. A good researcher can answer these types of questions very easily, and yet some clueless boob gets paid to spout a bunch of nonsensical crap on Townhall.com.

What passes as "research" these days is scary.

Did I miss something about rdonham?

:lol:
touche! Give him my regards over at LKP... 678hnn77

He got banned from LKP.

Damn, you gotta be pretty bad to get banned everywhere (INCLUDING LKP).

By the way, the guy that did the meat of the study is NOT the guy that wrote the Townhall article, it is Arthur C Brooks:

---------------------------------------------------------------------------------------
Arthur C. Brooks is Louis A. Bantle Professor of Business and Government Policy at Syracuse University’s Maxwell School of Citizenship and Public Affairs and Whitman School of Management. He is also a Visiting Scholar at the American Enterprise Institute.

http://www.arthurbrooks.net/author.html
---------------------------------------------------------------------------------------

You may begin your character assassination now Ramen...

I tend to agree that tithing shouldn't be included as a "charity" in this sense because of the administrative nature of the use of tithes. While most Churches set aside money for mission and almsgiving, I don't regard paying the light bill as "charity."

However, I would disagree that "theologically, it is mandated by the Bible," at least if by that you mean giving exactly 10% of your income. GIVING is mandated, but giving is NOT to be done as some sort of lawkeeping. Rather, it is done as an act of love and in furtherance of the Gospel.

Which is to say that the widow's mite is worth more than Bill Gates' 10% by far. To whom much is given, much will be required, etc.

The point is, just because the Bible instructs us to give, that does not make it some sort of compulsory law that we must keep such that our giving ceases to be an act of faith and love. Rather, it is out of faith and love for the Gospel that we are moved to give. If you give out of obligation, fear or compulsion, you are not pleasing God.

There was some thread on the OMB about San Francisco and bums crapping in the streets or something.

Sorry for the sub-reference. I thought everybody had read his tales of bum poop on the SF streets. He was quite enamored w/it for quite a while.

Anyway, nite all. I'll be back tomorrow. :mf_sleep:

Rdonham got run from this place around 8 months ago, not that he was dismissed by admin or anything. :unsure:
The posters here give him hell every time he spoke and he finally just left on his own accord.
He was a real glutton for punishment but even he had his limit. ;)

Either his methods suck because he did not use multivariate models to control for the different effects, or the Townhall.com rag piece you cited did not accurately describe his results.

First, I used the same data set he cited (the General Social Survey) to debunk the findings reported in the article by running some fairly standard regression models. His conclusions about the Democrat/Republican split do not hold up when other factors are controlled. That's statistics 101 right there, and again my freshmen and sophomore stats students would rip those conclusions apart.

Second, Brooks' own words confirm the models I posted here:
--------
Finally, Brooks says one thing stands out as the biggest predictor of whether someone will be charitable: "their religious participation." Religious people are more likely to give to charity, and when they give, they give more money -- four times as much.

But doesn't that giving just stay within the religion?

"No," says Brooks, "Religious Americans are more likely to give to every kind of cause and charity, including explicitly nonreligious charities. Religious people give more blood; religious people give more to homeless people on the street."
--------

In other words, by Brooks own admission religion is the dominant factor. The multivariate analysis shows that ALL of the variation in party identification is explained by church attendance and income.

You claim to be a professional researcher and yet you can't see basic methodological flaws in the articles you cite. Any competent researcher would see through the nonsense in that Townhall.com op-ed, which makes me question your claim.

Rd was funny. Back in the day lkpers did a lot of lemonparty posting, but Rd nuked them with the most disgusting pic of anal warts. I threw up a little, but I think that pretty much tops even that time when I got halcon to lemon party them all
678hnn77

You're arguing the theology of tithing. How many churches teach the gospel as you have articulated it, though? I think a large number present tithing as an obligation by the congregation, even if that obligation is stated implicitly. I'm concerned about how the individual views the tithing (for the purpose of understanding charitable giving), not with the correct theological interpretation. [The latter is important, don't misunderstand me. I'm simply saying that no small number of churches view tithing differently.]

Think about all of the televangelists who con people into believing they are pleasing God by sending them a $100 check. Some even offer "miracles" in exchange for those gifts. For a lot of people (though admittedly I don't know precisely how many), it's a fundamentally different mindset from other forms of charity.

All that said, excluding tithing presents its own problems for the analysis, which is why interpretation and strong theoretical arguments are needed in order to explain the causes behind charitable giving. Income serves as a suitable proxy for the ability to give while religion captures (to a large degree) an increased motivation to give. Viewing charity as a combination of motive and ability is a more theoretically sound (and more straightforward) explanation than some convoluted argument based on personality differences of people with different partisan tendencies.

The claim that politics has a strong effect on giving is simply a weak argument. That's my only point here.

No problem. I had forgotten about a lot of that stuff, honestly. I do remember that...er...person very vividly. A nasty human being he is.

He listed several other sources besides the "General Social Survey". Are you taking those into account?

I agree that religion is a big factor. That point was made in the OMB thread.

And just to beat this dead horse a little more, if charity were primarily a function of ability (income) and motivation (religion) then we would expect a strong interactive relationship. No matter how motivated, people with the lowest incomes would lack the ability to give. The effect of church attendance should be very low among the low income earners, and motivation should have the strongest effects among people who are not low income earners.

I went back and reran the models to include an interaction term (lowincome*attendance, where lowincome is defined as income below the 20th percentile). The interaction term was significant, indicating that there was indeed a conditioning effect. Church attendance had a small (and statistically insignificant) effect among the low income earners, and had a large effect for everyone else.

The party variables became even smaller and less statistically significant once the correct functional form of the other variables was specified -- theory tells us that income and attendance have a nonlinear relationship to giving. That's a fancy way of saying that the results are strongest exactly where theory tells us they should be, and the political explanation becomes even weaker after making the model more consistent with the better theoretical explanation.