Thursday, August 28, 2008

Quant bloggers, a precision division of the Steveosphere

Bruce Charlton (blogger identity BGC), writing in the science journal Medical Hypotheses, mentions a subsection of the Steveosphere, referred to as "quant bloggers". He singles out a few of us specifically--the GNXP crew, Half Sigma, and the Inductivist presumably for their needed "analysis and integration of [professionally gathered] data, for understanding its implications, and for speculating freely about the potential applications," and your correspondent for "chatty ‘opinion’ pieces".

The editorial is primarily concerned with the contemporary breakdown of the scientific Socratic method, with high status scientists no longer required to substantively defend their findings, address critics, or provide the underlying data employed to reach the conclusions they come to.

Quant bloggers come into play as a welcomed contrast to the new High Science. Quant bloggers, via their comment sections, are greeted with publicly viewable refutations, critiques, and suggestions from readers immediately upon publication. The data, if not presented in the body of the post or via links to the source, are available by request*.

It would be optimal to have professionals earning a living through the work undertaking what quant bloggers are currently doing. As BGC points out, however, the cultural climate doesn't permit it. If Steve struggles to keep his journalistic work economically viable, no one outside the protection of academic tenure is going to be able to pull it off.

So we'll make due with what we have. The best data sources are those able to be used for analyses unrelated to what the collector originally had in mind. Pew Research, for example, is an invaluable source not for the offered commentary accompanying the data (which I tend to skim) but for the raw numbers it consistenly makes available.

* If the Excel file I'm using lends itself to it, I make the numbers available via Swivel. If it exists online, I always link to the data source as well. And I have always and will continue to always make the data I'm using available to anyone who is interested in it. Just shoot me an email and it's on the way.

14 comments:

John S. Bolton said...

Congratulations on the recognition for being in a small group that some have high hopes for, as helping to advance learning where the speech code observant are too fearful.

tapeboy said...

Yes, congrats on the acknowledgment from the obverse other extreme.

I just wanted to tell you how much I enjoy the quant Audacious Epigone blog. I stumbled across your "Race Matters" article earlier this year when I was determining the correlations between "America's Safest and Most Dangerous Cities" and race. (BTW, correlations between race and violent crime for the 2006 list: Black = .5427; Hispanics = -.0416 [statistically insignificant]; Whites = -.5607) I would load it on Swivel, but I cannot find the Excel file at the moment :)

Providing references and data/spreadsheets is, I believe, critical to your commentary. It is the reason I now read your blog regularly. Keep up the good work.

Mike said...
This comment has been removed by the author.
Stopped Clock said...

I don't think Steve draws any significant income from his blog. I'm not clear if the ad money even goes to him, or if it's something that Blogger makes you add when you reach a certain number of clicks per day, i.e. to pay bandwidth. he remarked once about how irrelevant the ads were, and didnt seem too upset about it.

Audacious Epigone said...

JSB and Tapeboy,

Thanks. BGC is a bit of an insurgent, but he is operating amongst the professional scientists.

What cities did you run the numbers on and how did you determine them? It would be interesting to look at, but your last few blog posts spanning back into '07 don't seem to be on the topic.

Mike,

As Stopped points out, Steve has poked fun at the ads. I recall him drawing attention to an Afghani dating service at one point. But I think his primary income comes from TAC and VDare plus donations.

tapeboy said...

I still cannot find the Excel file containing what I did. I did, however, find a previously incomplete version. [2006_50_Most_Dangerous_Cities.xlsx] The file only has 50 of the 2006's "Most Dangerous Cities." I had originally done it for 100 off of the most dangerous listing. The results of the 50 cities for 2006 are:

[Correlations between "race(s)" and cities' crime ranking (murder, rape, robbery, aggravated assault, burglary, and motor vehicle theft)]

Black: r = 0.4854
Hispanics: r = -0.0750 (not significant)
Black & Hispanics: r = 0.5172
Whites: r = -0.5850

Of course, this is different from what I posted before since it is lacking 50 cities of data. I think it should be interesting to point out that the black and hispanic multivariate correlation actually was less than blacks alone for the 100 city analysis. The US Census definition of Hispanic or Latino is broad and would need to be broken down into further subgroups and/or additional variables in order to get a statistical significance. In contrast, blacks (as a group) do not require inclusion of additional variables to find a significant link to violent crime. I, too, have found that "variable" to be the single greatest predictor. When I added additional variables, my correlations often decreased or p-values increased. However, I must point out that I only played around with that data and in no way came close to even moderately analyzing the matter. (Although, I did begin to set up a database so that I could do so...but then I realized I was not being paid and it was not going to be published...so I stopped.)

More details regarding the "Most Dangerous List" and results can be found at http://www.jesuswarehouse.com/blog/most_dangerous_cities. Embarrassingly, I never actually finished my "blog entry." Unfortunately, I believe I am too lazy to actually be a "blogger." What a shame :(

Audacious Epigone said...

Tapeboy,

I wonder if we could find stats on single mother households by city for comparison purposes (I haven't looked yet). At the state level, I found the black+Hispanic % to be the single best predictor (and the black % alone was the next best) of violent crime at the state level, but a reader named Antero ran the same analysis at the county level and found % of single mother houses to be a better predictor than the % black, which was second. However, his data table was missing figures for several huge counties, including some in the Chicago area and New York, if memory serves.

Anonymous said...

I'm afraid half sigma does not deserve to be counted as a "quant blog" after no less than 18 straight Sarah Palin gossip posts.

Steve Setzer said...

Here's another source for education data, by the way:
School Data Direct

Much of it is already available from the feds, but it's still interesting.

tapeboy said...

You piqued my curiosity. Just as a teaser, I did it for the cities in California.

I used the FBI's 2006 "Violent Crime" numbers per 100K and the 2006 American Community Survey's "POPULATION UNDER 18 YEARS: Total" and "OWN CHILDREN UNDER 18 YEARS: In other families; Female householder, no husband present" (Tables B09001_1 and B09002_15) to calculate an approximate percentage for children in single mom households. (Of course, this does not account for the other minors living in broken homes.)

Correlations b/w "Violent Crime" and...
Black Pop: r = 0.1374 (p = .20)
Black Pop/Single Mom (all races): r = 0.1351 (p = .23)
Single Mom: r = 0.0075 (p = .94)

The very messy Excel file is here.

When I previously played around with it, I was doing it by the Metropolitan Statistical Area (MSA). Cross referencing 300 Census and FBI MSAs resulted in a correlation of r = 0.5500 b/w blacks and violent crime. It was greater than Gini, poverty, 25+ years with no high school diploma, or Hispanic % of Pop. I would link to that file too, but it is a total mess right now.

I need to relearn how to use Access before I can analyze all US cities or MSAs since, of course, Census and FBI cities/MSAs do not match up. (The FBI provides information for 8,252 cities!)

P.S>
Steve, great link (very easy to find info and downloadable!). Primary public education is another topic that I am slightly obsessed about. It is the 800 pound gorilla in the room that is leading to the deterioration of the US. Why is not market education in the top five national political issues, damn it!?

Audacious Epigone said...

Steve,

Thanks for that. As Tapeboy said, it is very user friendly.

Steve Setzer said...

Tapeboy and AE, you're welcome.

We gather tons of educational data in this country. The problem is finding time to dig through it all.

Fat Knowledge said...

I think the idea of having more full time blogger researchers/armchair scientists is a good one. The question of course is how to fund them. I took a look a while back at 8 funding methods for digital content which I believe would be applicable here.

The best ways I see to fund blogger researchers would be advertising, donation and government funding (and of course volunteer). You might also be able to charge a subscription like a magazine for access to your blog or data, or write a book and get proceeds from that, or give lectures and charge for those, but those all seem unlikely.

The gov't funding might be a bit tough in your particular case given the controversial nature of your work. In general I would hope that in the future more gov't funding goes to basement dwelling bloggers given the lower overhead to support them over their ivory tower brethren.

Advertising seems a good way to go with Google AdSense, but I am not sure how large an audience you need to pull to make enough to go blogging full time.

Donation is another way. You can ask for small donations from lots of people, but I think the better way might be to find one rich person willing to fund a blogger for a year at a time. A rich person could fund a bunch of bloggers working at home rather than a think tank with building overhead. Or a think tank could finance bloggers themselves (Wil Wilkinson of Cato comes to mind).

I keep meaning to do a post on how science research needs to change to adapt to the internet/Google age. I need to get to that. But, I think the article you were mentioned in hits a lot of the key points about how things will change. Kudos to being part of the revolution.

Anonymous said...

...please where can I buy a unicorn?