Tuesday, April 26, 2016

The True Costs of Denying Public Access: Worse Health Outcomes for the Poor?

Vox had an interesting piece today examining the relationship between income and age expectancy. You can read it here. As one might expect, they report that growing income inequality is related to a growing gap in age expectancy. But there’s another factor seemingly related to how long one lives: Where one lives. Here’s a brief snippet explaining the importance of geography and longevity:

To Chetty's surprise, the strongest pattern in the data was the role geography played: Low-income individuals lived the longest (and had more healthy behaviors) in cities like New York and San Francisco with populations that are, on average, well educated, and high-income.

What leapt out at me, however, was not the connection between urbanity and life expectancy for the poor. Another geographic pattern became apparent as I scanned two charts reporting the life expectancy gap by income in each state: Access to public land. Here are both charts as reported by Vox. The one on the left looks at men; the second, on the right, at women. Click on them to enlarge.


Look at the states clustered at the top: many have substantial public land holdings. Montana is number one for men, and 10th for women. Like a good social scientist, I combined the data in the two charts and added a percentage of total public land ownership variable to the mix. I calculated the longevity gaps between the richest 25 percent and the poorest 25 percent of men and women. Finally, I ran a quick correlation between the gaps and the public land variable. The resulting correlation statistic is simply a measure of how well the two measures move together and in which direction. Here’s a screen shot of the results:

The correlation is negative and statistically significant for men and women, suggesting that the more public land in a state, the lower the longevity gap between the richest and poorest quartiles regardless of gender. Further investigation uncovered that the gap between the poorest 25 percent and the second poorest 25 percent also decreases significantly with greater percentages of public land, but the correlation is less strong (-.30 and -.33 for men and women, respectively, versus the -.47 and -.43 reported above).

It’s hard to make too much of this relationship, as I have not controlled for a whole host of other demographic and behavioral factors that explain health outcomes. Until I’ve done that, it’s hard to see how strong the effect of public land access is on public health outcomes and whether it withstands controlling for these other factors. Correlation is not causation. People who value public land access are very likely different in many respects from those who do not. Certainly, people who move away from a place without much public land in favor of a place with lots of public space are probably going to have different habits and leisure activities than those who stay put. And places with public land are also quite different from places without. Without examining all of these factors, the relationship between life expectancy and public land access is not completely clear.

With those caveats made, an important finding mentioned in the VOX article is the association between poverty and two key factors associated with lower life expectancy: obesity and sedentariness. Clearly, greater access to the great outdoors would help combat both—and this access may be even more important for the poorest among us to increase their life expectancy. Providing and protecting access to public land might be one of the best ways to mitigate one the most dire consequences of rising income inequality.

Thursday, January 28, 2016

Montana 49th in Wages? Not Exactly.

This Sunday, you’ll get to watch Republican gubernatorial candidate Greg Gianforte face off with me and Mike Dennison in the MTN studios on Face the State. I'll post a link once it has been uploaded. This is the first time I’ve interacted with Mr. Gianforte extensively, and I thoroughly enjoyed our conversation. It should be an exciting race.
Greg Gianforte: Founder of RightNow Technologies and Republican candidate for Governor

During the show, you’ll hear me express skepticism about Mr. Gianforte’s claim that Montana is 49th in wages. I wanted to provide some context for that discussion, and explain why I don’t think this number is a particularly good measure of Montana’s overall economic health--and how I think it undercuts the argument being made by Mr. Gianforte.

As a political scientist, I like quantitative data. At the same time, when we use quantitative data, it is important to know how the data are calculated and the potential ways in which that calculation can introduce bias into our measures. It is true that Montana is not as economically well-off as other states. But are we really only better than Mississippi? I found that hard to believe. Here are three other measures of the economy and how Montana ranks:

U.S. Census
Per Capita Household Income (2014): 38th
Per Capita Income (2014): 34th

Bureau of Labor Statistics
Unemployment (November 2015): Tied for 12th (with Kansas)

Not first or even in the top half for per capita or household income, but not nearly rock bottom as the 49th in wages number suggests. And the unemployment picture is stellar. So what gives? Do we really have such low pay?

Back in April, many Montana media outlets reported that according to tax data compiled by Transactional Records Access Clearinghouse (TRAC), Montana was 49th in wages. No one took any time, however, to actually do a deep dig into that number. Once I heard Gianforte using the number repeatedly during his statewide announcement tour, I decided to do exactly that. 

I clicked on the methodology portion of the website, reading carefully about how the wage number is calculated. TRAC took the aggregate total of wage and salary-based income as reported on individual tax returns and divided that number by the number of tax returns received from the state. That led to the determination that at $33,180, Montana is 49th in average wages—just above Mississippi.

But there are two problems with this method. First, consider that some people derive no income from wages in a tax year. Some of these people are retired, some are students, and others live off of capital gains or other sources of passive income. Yet, these folks need to file a tax return—but are included in the data to compute the wage average despite having zero wages. This will serve to drag down the average wages. You might argue that this is the case in every state—and that’s true. But, if a state skews older (which Montana does) or has a high number of retirees receiving income from passive investments (see Big Sky, for example), then this biases the wage average downward. 

(In fact, Florida, New Mexico, West Virginia, South Dakota, Arkansas, and Maine are all at the bottom 11 and they are some of the oldest states in the US) 

Second, consider the nature of Montana’s economy. In 2015, we ranked number 1 (and, indeed, have been number 1 for quite some time) in the Kauffman Index of Startups. Put simply, we are a state full of small businesses and sole proprietorships—with people eager and willing to take risks. Many farmers, ranchers, telecommuters, tax accountants, consultants, plumbers, tradespeople and the like do not report W2 wage income either. I know this because for many years my wife was a consultant and she had no W2 income. Instead, all of her income was reported as business income—a separate line of a 1040 form. That means that many Montanans, who had good jobs, also report zero on the W2 line of their 1040s. And, that means they drag down the state’s wage average, too. Again, this is a double hit. Their income is included as zero AND they their tax return is included in the denominator for the overall wage average.

I argue this number very poorly reflects on the state of Montana’s job picture and I am skeptical of its use. There is no doubt in my mind that Montana can have a stronger economy and there’s a path for an even better future for our state. That’s a debate we need to have in this Governor’s race. But it is important, while having a conversation about that future, that we use the best numbers to decide what policies we need to pursue to keep Montana the Last (but not 49th), Best Place.

Tuesday, October 20, 2015

Speaker Zinke Redux: Still Too Liberal?

This evening, Congressman Paul Ryan announced his willingness to serve as Speaker if House Republicans agree to unite behind him. As Congressman Ryan Zinke has already indicated he would support Ryan if he were to run for the Speakership, it would appear that Zinke's Speakership bid (such as it was) appears to be over. Unless, of course, the House Freedom Caucus still thinks Congressman Ryan is too liberal.

I wrote last week that Congressman Zinke appeared to be too liberal to obtain enough support from the House Republican Caucus. Using a common measure of ideology employed by political scientists, I demonstrated that the Congressman was to the left of both the House Freedom Caucus and the House Republican mean. In fact, among the many names bandied about for the Speakership, Congressman Zinke was the second most liberal.

The Congressman's spokesperson, Heather Swift, took to Twitter to dispute my claim. She pointed me to GovTrack's measure of ideology. Swift pointed out that it was important to not only consider votes taken on the House floor (which is the basis of the NOMINATE score), but legislation that is sponsored and co-sponsored but which has not (and very well will not) receive space on the House agenda, therefore, will not receive a formal roll call vote.

Swift is not entirely wrong on this point. In fact, as I argue in Battle for the Big Sky, bill sponsorship is a better measure of true preferences because of the substantial resources required to submit and advocate for a piece of legislation. I don't entirely buy the argument that co-sponsorships is a good indicator of real ideological preferences, however, because a co-sponsorship could represent a member's true, revealed preferences or it might simply represent "cheap" talk--an attempt to look more left or right than one actually is.

In any case, I was happy to investigate the data that Swift pointed me to in order to see if I had missed something or had misrepresented the Congressman ideologically. I went to GovTrack, read their account of their ideology measure based upon bill sponsorships and co-sponsorships, and downloaded it. The folks at GovTrack note that their measure of ideology, while based on different information than Keith Poole and Howard Rosenthal's NOMINATE scores, correlated well with those scores. You can read about the methodology here and grab the scores yourself.

The first thing I did was merge the NOMINATE data with the GovTrack Ideology measure and ran a simple correlation in Stata. Sure enough, the measures correlated robustly at .90. In other words, the measure of ideology based on roll calls was very similar to the measure of ideology based upon co-sponsorship patterns. In the political science business, this is what we call a high degree of face validity.

Next I simply ran a quick scatterplot of the NOMINATE and ideology scores for each member of the 114th Congress. Again, visually the data show that the two appear to measure the same thing. The upper right corner represents the Republicans while the lower left represents the Democrats. Here's that plot with the X-axis the Poole-Rosenthal NOMINATE and the Y-axis the GovTrack Ideology measure.

The next two graphs separate out the seventeen Republican speaker candidates mentioned in the news. The first chart replicates the scatterplot above, while the second simply graphs the speaker candidates from left to right using the GovTrack ideology score alone. In both charts, I add the mean value of the House Republican caucus, the mean value of the House Freedom Caucus, and the value for one standard deviation above and below the mean.

What do I find?

Essentially the same story I found when looking at the NOMINATE data alone. Congressman Zinke is now the fourth most liberal among those mentioned as Speaker candidates when looking at the GovTrack ideology measure. NOMINATE had him at second most liberal. He's still to the left of the Republican House Caucus mean. And he's still considerably to the left of the House Freedom Caucus. Using these data, Paul Ryan, Ryan Zinke, and Darrell Issa are all essentially the same ideologically--and all are to the left of the House Republican Caucus.

The take away point? Congressman Zinke, whether using roll call votes or co-sponsorship patterns, is on the left ideologically among his House Republican colleagues. In fact, there are 62 other members who are closer--using the GovTrack ideology measure--to the House Republican mean.

And, again, if Congressman Ryan is too liberal to be Speaker according to the House Freedom Caucus, than isn't Congressman Zinke?  #data

Thursday, October 15, 2015

Congressman Ryan Zinke: Too Liberal to be Speaker?

Montana's lone Congressman, Ryan Zinke, has recently been floated as a possible candidate for House Speaker. Apparently, the Congressman is taking the possibility of a campaign for the Speakership seriously and may throw his hat in the ring.

Zinke's leadership as Navy Seal propelled him to victory in 2014, and he--and others--think his leadership skills are exactly what are needed to right the Republican House Caucus.

I think this idea is crazy for lots of reasons, and I've gone on the record as to why I don't think it's a good idea or feasible (here and here). But to be clear: Zinke's leadership skills are not the problem.

One item I did not mention in my interviews is the Congressman's ideology relative to this peers in the House Republican caucus because I had not reviewed the data. Now that I have, I can say with some confidence, that the Congressman's ideology would present a serious problem in any Speakership bid. And, contrary to a recent press release from the Montana Democratic Party, it's not because Congressman Zinke is a "Tea Party sympathizer." Instead, Congressman Zinke might be too liberal to be Speaker of the House.

On this blog, I have long referenced the DW-NOMINATE scores produced by political scientists Keith Poole and his colleagues over at voteview.com. Using roll call votes cast, Poole and his fellow number crunchers develop a measure of ideology arrayed along a left-right dimension that plots each member of the Senate and House from most liberal to most conservative, roughly constrained at -1 for the most liberal member and +1 for the most conservative. In the past, these data have only been available after a congress has concluded. Now, however, they have developed a measure of ideology which they update weekly once roll call votes have been concluded.

I downloaded these weekly DW-Common Space scores earlier this week (available here), stripped out the Senate, and then calculated the mean ideology score for the House Republicans and then the House Freedom Caucus (using wikipedia's membership list to determine who was in the caucus). I then plotted those values along with the ideology scores of every conceivable speaker candidate from least conservative to most conservative. The results are reported below. Click on the chart for the best view.

Table:  Ideology of House Speaker Candidates in the 114th Congress

As is plain to the reader, Congressman Zinke is well to the left of the House Freedom Caucus, slight to the left of the average Republican member, and--among all the possible Speaker candidates mentioned--the second most liberal to Congressman Greg Walden, the NRCC chair from Oregon. He is more liberal than Congressman McCarthy, who withdrew from the speakership race. And, he is certainly more liberal than Congressman Ryan, who, according to The New York Times, is possibly too liberal for some members of the very conservative House Freedom Caucus.

Congressman Zinke will not be Speaker of the House. He's simply too liberal for House Republicans.

Monday, August 17, 2015

Hate Forest Fires in Montana? Then Let's Overcome Cognitive Dissonance and Address Root Causes

Summer is my favorite time of year in the Gallatin valley, perhaps because it is so fleeting (which makes it ever-so precious). I hike, bike, and run about town in the wonderful sunlight, surrounded by blue skies and hot, dry heat.  Of course, this weekend, my beloved Bozeman has been choked with smoke from a wildfire out in Three Forks. One of the costs of living in Montana, and the West more generally, is the fire season which is so essential to life and renewal here. Of course, that fire season has been getting worse and longer, which makes it harder on me when I exercise outside due to my asthma, a condition I've learned to manage throughout my life.

Over the weekend, I had time to reflect on forest fires and how sound public policy might help bring them under some control. I revisited Senator Daines’ recent op-ed in the Billings Gazette discussing wildfire and forest reform, which I read with interest. First, Daines hits the nail on the head concerning a critical challenge facing the US Forest Service and other federal agencies responsible for managing healthy forests: money. The US Forest Service’s budget is burdened with ever-increasing fire-fighting costs, which drain its ability to spend on other important activities such as trail, campground, and facility maintenance. Daines’ solution to this problem is his Wildfire Disaster Funding Act “which ensures large forest fires are treated and funded as the true natural disasters they are, similar to hurricanes or tornadoes.” I hope the bill—which Daines is co-sponsoring along with Senator Tester—finds a solid reception among critical allies of both parties, particularly those senators representing East Coast states hard hit by Hurricane Sandy. (Montanans also should hope that they don’t hold then-Congressman Steve Daines’ very first roll calls against him, given that he voted against a bill funding Sandy relief efforts.)
But what troubled me profoundly was the blame the senator placed for the “deteriorating conditions” that are responsible for increased wildfire risk. Senator Daines notes that we are at risk due to beetle kill in our forests “being left untreated”—a risk that he says is compounded by “years of inadequate forest management practices, spurred by obstructionist litigation from fringe groups and excessive regulations.”
Really? Well, I guess it’s time for me to whip out some “fancy” social science and give a little lesson in the funnel of causality. This is a theory developed by Campbell et al in their path-breaking work The American Voter, published in 1960. Essentially, Campbell et al argue that while the proximate decision influencing how someone votes is a person’s issue position, those issue positions are the product of a person’s party identification, which itself (often?) is a function of how the person was socialized into politics by his or her parents.
In other words, there’s a causal chain one needs to follow to understand the best and most powerful predictor of voting behavior, and that predictor is partisan identification—a bundle of attitudes and beliefs that is not immediately proximate to the voting decision. Issue positions don’t really matter—it’s the partisanship behind those positions that do.
Now let’s apply the funnel of causality to Daines’ argument on wildfires: According to his op-ed piece, the proximate causes for the risk we face, which are the tired trope of “government mismanagement” and “fringe environmentalists,” are the real problem Montanans face and the ones that require attention and redress.
But one must ask: Why is there more beetle kill in the first place? And why has the size of wildfires been on the rise in the United States, Canada, and globally—both as the charts below indicate and a recent study demonstrated? Oh, right: Because the warmer winters associated with global warming mean fewer Pine Bark beetles are dying off, enabling them to leave behind more dead trees strewn about waiting to burn up. All of this is well-documented Andrew Nikiforuk’s Empire of the Beetle. And, again, a recent study demonstrates that increasing temperatures lead to more wildfire activity globally.

Total Hectares Burned by Wildfires in Canada, 1970-2013. Source: Canadian National Forestry Database

Total Acres Burned in the US (in millions of Acres), 1960-2014. Source: CRS Report, "Federal Funding for Wildfire Control and Management," July 5, 2011 and National Interagency Fire Center.

Daines is giving far too much credit to proximate causes: lawsuits and mismanagement. The root problem is global warming. We need to address that if we really want to get a handle on our wildfire risk. And to address that we, as global citizens, need to come to grips with our role in global warming via our insatiable appetite for carbon emissions.
Unfortunately, Senator Daines voted against an amendment earlier this year during the Keystone XL debate acknowledging the human role in climate change and, when running for the House, indicated in an interview a few years ago that the “jury is still out” on the role of CO-2 emissions in our ever-warming world.

The jury is not out. Ninety-seven percent of studies unambiguously endorse the notion that rising carbon dioxide levels, the result of human activity, are an important and substantial contributor to global warming. Check out the information yourself at NASA.
How many of you, when running a business or a household, place your bets on the 3 percent versus the 97 percent? If Senator Daines did this during his business career, I assure you, his career would have been much shorter and far less successful than it was. If I said it is only a "theory" that how people vote is a product largely of their partisanship, I'd be drummed out of the profession with good cause. The problem is that it is very hard for people to accept information that conflicts with their priors. Humans don't like cognitive dissonance, so instead, we reject information when it doesn't fit our beliefs. Worse, we search for justifications to confirm why that information is wrong and why what we believe is right. It is hard to overcome cognitive dissonance, but we--and those whom we elect to serve us--must.

The senator is a bright, articulate public servant. He is a graduate of Montana State, and has a degree in engineering. One would think that he would, as a man of science, make public policy on science and not advanced smoke-and-mirrors arguments about a very real threat our forests face from wildfire. How we fund fire fighting absolutely must change: Daines is right on that. But if we are really going to make our forests safer today and for future generations, we need to stop blaming red herrings and, instead, face the facts staring us straight in the face with regards to climate change.