Jeff Lax and Justin Phillips posted this summary of attitudes on a bunch of gay rights questions:
They did it all using multilevel regression and poststratification. And a ton of effort.
P.S. My only criticisms of the above graph are:
(a) I'd just put labels at 20%, 30%, 40%, etc. I think the labels at 25, 35, etc., are overkill and make the numbers harder to read. And the tick marks should be smaller.
(b) The use of color and the legend on the upper left are well done. But they should place the items in the legend in the same order as the averages in the graphs. Thus, it should be same-sex marriage, then 2nd parent acdoption, then civil unions, then health benefits, and so forth.
6.13.2009
Gay Rights are Popular in Many Dimensions
@ 8:05 PM
Subscribe to:
Post Comments (Atom)
47 comments
It would also be good to have the graph go from 0 to 100%. As it stands, it looks like there is, for example, almost no support for gay marriage in Alabama when there is actually 22% support.
My eye keeps being drawn to Iowa, being right about in the middle of the graph and having every circle filled in. Be very interested to see how their state level politics play out over the next couple of years.
Another criticism is that the charts don't extend to 0 and 100%. The graph as presented overemphasizes the extremes of view. Housing antidiscrimination in Massachusetts enjoys an 83% approval rate, yet appears almost all the way to the right of the graph, implying near universal support. While on the other end, 23% of Alabamans support same-sex marriage, yet it's dot is almost all the way to the left, implying near universal condemnation.
The only question is, how hateful is Obama, this guy keeps licking gays to the curb.
Good visualization tool. It is fascinating to see how the 'gaysupportenedess' tracks in comparable proportions on the various dimensions such as hate crimes, discrim in housing, marriage, etc.
Any chance that we can see this on a flash visualization showing the curves moving over time?
It's irritating when people on a STATISTICS website gripe about the range on the graphs. It is STANDARD PRACTICE not to include any more range on the axis than necessary to plot the data clearly.
People need to learn to read graphs with the extremes of ranges cut off. It's a waste of space to show everything from 0% to 100%, especially if there are no values at the extremes. Also, if the full range is used, then it becomes harder to read the graph because everything will be smaller and closer together.
That said, I find this particular graph interesting, especially noting those states that haven't legislated basic housing and employment antidiscrimination. That seems so obvious to me, that it's surprising to see how many states DON'T have equal rights.
mclever said...
It's irritating when people on a STATISTICS website gripe about the range on the graphs.
Agreed.
When did those complainers last see a DOW chart go from 0 to xxx (what IS the upper limit to the DOW?) for a one day period, one week, one quarter, or one year? So why aren't they also complaining to Dow-Jones, the Wall Street Journal, and their local media about that?
Mike in Maryland
My Blogger ID is http://www.blogger.com/profile/02848893412251095965
Justin and I were worried our work might be controversial... but we had no idea things would get this ugly so fast...
durhambrad said...
It would also be good to have the graph go from 0 to 100%. As it stands, it looks like there is, for example, almost no support for gay marriage in Alabama when there is actually 22% support....
.... which is virtually no support.
One thing I'd add is a population weighted average of the entire country, and then one for each region.
Kind of pathetic that my state of Virginia ranks where it does. Of course if the chart broke out real Virginia from us communist slobs in the Northeast, we'd be up near the top...
By the way, the estimates plotted above come from Table 1 in our forthcoming paper here.
Justin and I were worried our work might be controversial... but we had no idea things would get this ugly so fast...
So far, the only controversy in the comments section is whether the graph should extend from 0-100% or not.
I think you guys did an excellent job putting all of this data together. I'm not a supporter of a lot of the "pro-gay" policies (like hate crime protection; I think violence is violence and should be punished as such no matter what the reason), but it's nice to see where each state stands.
Any idea of where Washington, DC would rank on this chart? Especially considering the referendum petition that is currently being debated?
Who exactly are these people in Utah and Idaho who support same-sex marriage but don't want people to be able to adopt their spouse's child?
One of the unfortunate consequences of inappropriate American interventions in the Middle East, is that when an injustice occurs, folks like Jen go 180 degrees the other way and say that it's none of our business. It is the business of the international community to speak out in the fact of injustice.
It's a pet peeve of mine, but New York _does_ recognize same-sex marriage, as a result of Martinez v. County of Monroe. As things stand, you need to travel out of state to get the license, but once you do, New York affords you all of the rights that every other married couple in the state are entitled to.
I found the data fascinating on an array of attitudes.
If I was evaluating the placement a high technology business, there are several states that I would not consider. There are several states on the list that were a surprise and my focus was not on the gay marriage issue but the more broader feelings on social issues, that speak to the community.
This site is a treasure trove for data.
The top states on the list (New England plus NY & California) are no surprise, nor are the bottom ones (Deep South & Appalachia). What I found most interesting is how gay-friendly the Southwest (Colorado, Nevada, New Mexico, and Arizona) seems to be. This was (except for McCain's home state, which will be in play without him on the ballot) the breakout new blue region in 2008. It also seems to debunk the GOP's claim that "values" issues are their ticket to win over supposedly socially-conservative Latino voters. Interesting to see where this leads...
@ scott:
"I think violence is violence and should be punished as such no matter what the reason"
That's just silly. Would you suggest abolishing first and second degree murder as well as manslaughter, and making the punishment the same for all three crimes?
Gay Rights are Popular in many Dimensions
Share this Content
I am gay but is the above an April Day Fool's joke?
@jbailin My best guess on Utah (and similar, though less obvious trends in WY and ID, two other states with large Mormon populations) is that there may be some shallow support among LDS folks for the idea that gay people settle down and try to live as stable a life as possible, given their circumstance as gay people. Gay marriage may be seen as the best outcome of a bad situation for gays. After all, the church doesn't oppose the right of marriage to non-Mormons who drink alcohol to excess or cheat on their taxes or who ignore any other advice from the Word of Wisdom.
There may also be a very small number who are turned off by anti-gay marriage advocates who say it is a slippery slope to plural marriage, as if that would be the worst thing since the Manson murders. Saints know the Coalition for Marriage is not talking plural marriage in regards to Methodists, Anglicans or Catholics. They may see it as a subtle attack on a church that left plural marriage behind over a century ago. But, remember that more than a few Saints have an... ahem, complicated family tree back in the pioneer years. And more than few are still suspicious of anti-Mormonism among GOP evangelical protestants after some of the vicousness that came out around Romney's run for the nomination.
There is less support on a gut level for gays adopting kids. They may argue that while gays have the right to do as they wish, adopting kids and raising them in "that environment" would interfere with the child's free agency in regards to later moral choices.
Saints have a subtle and fairly complex set of beliefs, which does not lend itself very well to a simple and cartoonish "We hate queers!" sort of interpretaion.
That's just silly. Would you suggest abolishing first and second degree murder as well as manslaughter, and making the punishment the same for all three crimes?
Of course not. However, I would want to keep the punishments for "manslaghter against another person" and "manslaughter against a gay person because he was gay" the same. When you tack on extra punishment because the crime was borne out of hate, you are essentially punishing a thought or belief. As horrible as hate against gays is, that is unacceptable.
Scott...
You said... "...you are essentially punishing a thought or belief." You seem to suggest that the "thought" that goes into "premeditation", which does in fact add to the penalty of any crime, should be ignored as well?
I am on your side when it comes to frowning on cluttering up the legal codes with superfluities and nonsense. However a hate crime is in a very different class from ordinary violence, because frequently these spring from hatred that is organized and occasionally very well funded.
So an inner-city drug rip-off murder and the murder of Matthew Shepard are vastly different in both scope and consequence.
@Scott While theoretically, I strongly sympathize with your initial urge that all crimes be treated equally, regardless of the criminal's state of mind, I must disagree with your conclusion, because a hate crime has an additional criminal goal beyond the crime itself, and therefore carries an additional offense.
A simple crime which is a non-hate crime goes something like this: Victim has a $100 watch on her wrist in a dark alley. Criminal wants watch. Criminal kills victim, takes watch. Terrible tragedy that should be punished, but it's a simple murder. It had no further social goal than sheer greed and no further social impact than perhaps raising the fear of dark alleys.
OTOH, in a hate crime, it runs more like this: Criminal decides to discourage behavior he does not like, but is not illegal, such as gay people holding hands in public, a synagogue operating, or a black family moving into a white neighborhood. The criminal attacks the "offending" person (the gay couple, the congregant leaving the synagogue or the mother of the black family) and kills them.
All are murders, and should first be prosecuted as such, yes. But hate crimes have an additional component. That component is the criminal's attempt to enforce his own personal view of social policy on his community in an extra-legal manner. So, although the community has decided that it's not illegal for two women to hold hands in public, or for a synagogue to open, or for a black family to live wherever they can afford to buy a home, the individual criminal decides that his personal judgment supersedes than that of the community.
This makes a hate crime more than just a plain old murder. A hate crime is a direct attack on our civil decision making process, an attack on the very notion of a democratic republic.
So, in addition to punishing the initial crime, hate crimes legislation punishes *that* act. Hate crimes legislation is a way for communities to re-affirm that, though the criminal has the legal and constitutional right to campaign to criminalize gay behavior, oppose the operation of a synagogue when such is proposed before a zoning board, or to re-establish segregation through accepted political, electoral and even organizational (boycotts, picketing, etc) means, he does not have the right to extra-legally overturn those decisions once they have been reached.
@Nosimplehiway:
Well put.
Murders are not all the same. Many states have increased penalties for murder of a cop. And you can bet that assassination of the president would be treated much more seriously than the murder of a wife's lover.
This is because the policeman and the politician represent civil society. An attack on them is an attack on all of us. Similarly, hate crimes are, fundamentally, terrorism. And they should be treated as such.
wv: oustr: what Ahmedinejad may deserve but will not get.
I think that graphs that save range space should make the cut explicit. E.g.
0--//--20--25-30...80--85--//--100
to draw attention to the fact that there is a natural range that's being truncated. Yes, if you read carefully and think about it you don't need this, but people don't do that reliably.
As for the Dow, seeing full-range graphs more often might well be beneficial.
The complaint about the order of items in the legend vs. the graph is a good one.
But otherwise it's well-done and interesting, I bet Tufte would be proud. What jumps out at me is the democratic deficit, the large number of states with strong support for unimplemented policies. Housing anti-discrimination has supermajority support in almost all states, but lots of empty circles. Conversely, Iowa in particular and 2nd parent adoption in several states has been pulled ahead of popular opinion.
Very nice. I agree that showing cuts from 0-20 and 85-100% would be helpful. Only other comments on the graph itself would be to use a typeface other than Arial (which always looks very "default" and unconsidered.. except for Times or Comic Sans almost any other face would make this pop a bit more), and to perhaps ditch the border around the legend.. there are already enough black lines (all the same thickness) as it is.
What would be really great here is another sheet showing seven small maps, one for each policy. Not difficult to do at all, and I think you'd see some interesting patterns. If you want to send some data to bill@radicalcartography.net, I'd be happy to give a quick look.
If I get the chance I will correlate this data to the the last several national elections.
I am curious how this tracks to popular vote and margins for Democrats and Republicans.
Is it possible that there's some typo in the entry of the data for Utah? On this table, support for gay marriage is listed as 25%, but on the previous chart, support for gay marriage was below 20% at all three time points that are broken out. Assuming the values here should be some sort of (possibly weighted) average of the values from the other chart, this doesn't make sense.
My thought at first was that the second-parent adoption dot and the same-sex marriage dot were accidentally switched, but even the second-parent adoption dot has a value greater than 20% here, which means that it also couldn't be the average of the values in the other chart.
Alabama born and raised.
It is STANDARD PRACTICE not to include any more range on the axis than necessary to plot the data clearly.
Standard practice and good practice are not necessarily the same thing.
The chart would be so much improved if it extended from 0 to 100 per cent! Otherwise very interesting and useful.
Mr. Gelman,
The chart is very readable AS IS, IMO.
My only quibbles would be agreeing with too many labels (10% breaks would be better), and to have a couple more vertical lines at maybe 30 or 35, and 65 or 70. Be a bit easier on the eye to guesstimate whether a state in the middle of the graph is at one percentage or another for a particular without scrolling up and down, up and down.
However, maybe to please some of the whiners and crybabies about the 0% - 100% not showing, maybe you should create a zero line on the left side, then squiggles showing the gap to 20%. On the right side, squiggles and then a line to indicate 100%.
If the whiners and crybabies can't read such a slightly revised chart, then don't even think of entertaining such complaints in the future.
And/or show a one day movement on the DOW, with the lower limit set to the lowest value of the DOW (July 8, 1932) when it stood at 41.22, with the upper limit set for the same amount over the historical high (14164.53 + 41.22 = 14205.75), and see if the whiners and crybabies could read it then, and/or have comments.
Mike in Maryland
My Blogger ID is http://www.blogger.com/profile/02848893412251095965
This might be obvious to anyone used to doing statistical analysis but what do the filled-in vs. non-filled in circles mean? Does this speak to the quality/quantity of the data that dot is based on? Or does it show some sort of rounding or something?
@CJ:
I wondered about Benford's Law as well. Apparently a professor from the University of Michigan has done the analysis on the data and concluded that it complies with Benford's Law. The math is beyond me, but maybe you'd like to take a look at it. There's a link to a PDF here.
I believe the graph should have rainbow colors. Those colors simply aren't gay enough.
I somewhat agree that showing a full 0-100 range would make it a better representation of the data for the general user, though it is a very good chart as is. Definitely a keeper.
I can read this chart fine, but then I'm used to reading charts and I get the reason for cutting off parts of it. But in this case, allowing 20% of the chart to be unshown on one end and 15% to be unshown on the other subtly shifts the visual center of the debate toward the anti-gay side of things. A full chart would show the majority of the range of data to be right of the 50% line. None of this was intentional, I believe, but any time statistics are shown graphically, there can easily be some slight distortion. With the data graphed on a full-range chart, the takeaway for this posting might easily have been: "22% of Utahns Support Second Parent Gay Adoption, While 18% of Massachusetts Residents Think It Should Be Legal For Landlords To Deny A Lease To Gay People."
Looking the chart over again, btw, I noticed something I hadn't seen before. Towards the bottom of the chart there's a split. Views on marriage, civil unions and adoption (private, family issues) shift hard to the conservative side, while views on public issues (employment, housing, hate crimes, even health benefits for same-gender partners) holds pretty steady above 50%. (With the exception of UT, which is usually an outlier, and to a much lesser degree OK and AR.) Maybe this is just because marriage is the issue in the news lately, so the pump has been primed. But I wonder if it is a more intrinsic philosophic difference... public sphere v. private sphere, or maybe "special rights" vs "discrimination just ain't right".
It certainly seems from this chart that the nation as a whole is ready for ENDA. Perhaps a good strategy for the pro-marriage side would be to put marriage on the back burner for a bit until opposition ages out of the electorate (maybe 5 years, certainly no longer than 10), and focus on ENDA or inclusion in the 1964 Civil Rights Act. Once a victory is won on that, as it seems it could be, and the conservatives don't experience pillars of fire destroying homosexual riddled cities because of it, then the marriage fight may just be that much easier.
@MikeInMD The stock market graph you describe is a less than perfect parallel. The fact that you are capable of proposing a ridiculously bad graph for the stock market does not therefore mean that showing the full range of possible results might not improve this chart slightly. It's not that those who are offering a full-range chart as a constructive suggestion are whiners and cry-babies, and it's definitely not that they are incapable of reading it.
@phred Circles filled in mean that that particular policy has been put into effect. An empty circle means it has not. So, Iowa's solid purple circle means IA has a law against housing discrimination, while Pennsylvania's open purple circle means that a landlord can legally say, "Nah, I don't rent to homos." The circles only represent statewide policy, though, so in PA, housing discrimination is illegal in some cities, Philly, for example. :-)
Nosimplehiway said...
The stock market graph you describe is a less than perfect parallel.
IMO, yes it is.
The whiners and crybabies want to see 'the full range', even if it includes a lot of white space that shows nothing. A DOW chart would also show 'the full range', including a LOT of white space that shows nothing.
I'm not saying show the daily closing average for the entire history of the DOW - I'm saying show the particular day's trading on that graph, and the whiners and crybabies would then get what they want. A graph that shows the full upside and downside the DOW has ever shown, but the data is represented in such a way that it is totally unreadable and unintelligible.
Or maybe the whiners and crybabies should just learn how to read charts and graphs.
Mike in Maryland
My Blogger ID is http://www.blogger.com/profile/02848893412251095965
I was surprised that Pennsylvania is as high as it is for gay rights. I really expected it to be closer to Ohio and Indiana - it leans culturally conservative on other issues. I was surprised so many people favor the anti-discrimnation and housing stuff. Why shouldn't people be able to hire and rent to who they please?
It's also that New Jersey and the desert Southwest states are almost as liberal on these issues as the weirdo liberal states of New York, Massachusetts, and Vermont.
It's an excellent chart, never said it wasn't.
@Nosimplehiway The split becomes apparent partly because the states are ranked by their degree of support for gay marriage. It would be interesting to see how things looked if they were ranked by another series, such as job discrimination.
So empty circles to the right of 50 would be politically feasible things to vote on? i know it's prolly not that simple...
@Matt No it's not that simple, but in broad brushstrokes, yeah.
Remember, these are polls in most cases of the general public, rather than likely voters. Likely voters tend (broad stereotype alert) to be older, less liberal and less transient than the general population.
Also, how a referendum is on the ballot can make a big difference. Presidential election years, odd numbered years, a particularly charismatic candidate of the state's minority party, national swings, or even just the state's largest city having a not heavily contested mayoral race can all shift the ideological composition of the electorate. It's unlikely that that shift would be more than maybe ten or twenty points, though, even in a perfect storm. So, 60-70% would be more the safe level for a vote.
More important than referenda (which I personally despise as a means of governance, but that's a whole other story) though, is what these polls say to elected officials. If a statewide candidate (Governor, AG, Controller, whatever) sees numbers like we see in LA, AZ or WV in support of outlawing housing discrimination, it may give them courage to act even in theoretically red states. I could definitely see even fairly conservative GOP candidates running for Gov in any of these states on a platform of no to marriage, but supporting ENDA.
BTW, I used to know a Matt Gilbert in Easton, PA. Any relation?
@Nosimplehiway Nope, different Matt.
Thanks for the details. I was just worried about straight up swiftboating and the like, but you laid out many more complications.
I went and made some maps based on this data, for those who are interested.
This being politics, I think people do have a point that it would be nice if there were a visual cue where 50% was aside from the tick marks. Please don't enlarge the axes ranges, but perhaps slightly different shading, with the 20% around 50% one shade, the 10% to either side of that another shade, etc would visually show that most gay rights are well above 50% in support might make it clearer.
Also, tick marks for states at intervals of 25% would help, I had to count up to locate Florida as the middle state so I could think about it in terms of presidential election repercussions.
One more comment..."filled circles for existing law" might be clearer than "filled circles for pro-gay policy."
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店經紀,
酒店打工經紀,
制服酒店工作,
專業酒店經紀,
合法酒店經紀,
酒店暑假打工,
酒店寒假打工,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店工作,
酒店打工經紀,
制服酒店經紀,
專業酒店經紀,
合法酒店經紀,
酒店暑假打工,
酒店寒假打工,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店工作,
酒店打工經紀,
制服酒店經紀,
酒店經紀,
菲
梵,
Post a Comment