Adventures in Onionism

Conservative Tests under Satisficing Models of Publication Bias , by Justin McCrary, me, and Daniele Fanelli. In one sentence for the lay-people: if you think there's a lot of publication bias, you might want to look for test statistics of 3 instead of 2 before you think results are statistically significant. With caveats, because nothing in life is simple.

The incoming class of Fellows at the Berkeley Institute for Data Science. My reaction, based on how awesome everyone else is: And now I'm tearing up because Maria .

For the first time in a month I stayed home this weekend and did nothing. Where nothing means grading finals and 51 miles and 12,000 feet of vertical gain, a bunch of it off-trail on Mt. Tam. Also, I got a research fellowship from the Berkeley Institute for Data Science . Starting in the fall I'll be half time at BITSS and half time at BIDS (confusing, I know). I spent Thursday and Friday at a BIDS conference on reproducibility that was pretty interesting, and people were super nice, and they're all better programmers than me, so it'll be interesting see what I can bring to the table as a social scientist. Importantly, it looks like I'll be good to stay in Berkeley for two more years.

I've been busy the last two days with the annual meeting of my employer, the Berkeley Initiative for Transparency in the Social Sciences ( BITSS ). Thursday morning we did a training session on registrations, pre-analysis plans, the Open Science Framework , R, RStudio, git, GitHub, writing a project protocol, and the Harvard Dataverse. (Maybe we bit off more than we could chew in that session?) Thursday afternoon we heard from some of the leaders in the research transparency movement: Ted Miguel (Berkeley), John Ioannidis (Stanford), Brian Nosek (UVA/Center for Open Science), Jack Molyneaux (Millennium Challenge Corporation), Victoria Stodden (UIUC). Friday we heard about recent research in transparency, including my presentation of a draft of the Manual of Best Practices in Transparent Social Science Research. People seemed to really like it. My slides and the draft manual are all on my github account. If you're into research transparency, we're encouraging open partic...

I'm in Charlottesville, VA at the moment, visiting at the Center for Open Science . (If you're interested in doing reproducible research, their Open Science Framework tool will likely be helpful.) We had a pretty fascinating meeting with lots of top researchers in the transparency movement earlier this week, which I blogged about it over on the BITSS blog , where' I'll occasionally be writing. 

Do not congratulate me yet--I haven't gotten the official lollipop of approval from the lady in Sproul Hall yet, but I uploaded my dissertation to ProQuest, and I have submitted the form with the signatures of final approval from my adviser and other committee members. In other news, I have a cold. I also thought I had a cavity, but the dentist said that too was also just my cold. I am hoping I'll be better in time for the PCT 50 outside San Diego this Saturday, but I can't guarantee it. ADZPCTKO was awesome as usual. Mattress and Heinz were sorely missed this year, but Ducky, Nano, Nitro, and I were keeping hope alive. Thanks to the high precipitation winter, the wildflowers are crazy this year. I used the day on Saturday to go scout out some of the San Diego 100 course in Anza Borrego and Cuyamaca Rancho State Parks. Gorgeous. Representing. I rode the train home on Monday from San Diego to Emeryville. I didn't think it's as pretty as other routes I've taken. M...

Presented today. I think it went well. You know it's going horribly if nobody bothers to make comments, and I was worried about that at the start, but I realized that was just the part where I was talking about non-economicsy background stuff ("Here's how you join the military"). It got lively with lots of comments when I actually started showing results, and the faculty was dropping knowledge bombs all around. And the suggestions are stuff I can actually do, so that's cool. I was exhausted, so I'm spending the rest of the day buying groceries, doing my laundry, fixing some bike spokes, and maybe watching Carrie or reading Harvey Pekar's adaption of Studs Terkel's Working . (And talking to a guy at RAND about the control functions approach to eliminating omitted variable bias.) The only problem is, I bought the freewheel removal tool (the Suntour 2-prong model) but I've only got an 8-inch crescent wrench which is not giving me enough leverage. At ...

Well, it's been a fascinating few weeks. When research is going well I actually sort of enjoy insanely long days in the office every single day of the week with no time to even do my laundry or buy groceries. Or cook. Or eat. Today I have a minute to breathe, but only because I'm all the sudden really worried that my data are too good to be true in that they might exhibit evidence of time travel. (You see how I made "data" plural there? I hate that. Next I'll be speaking only in passive voice or talking about myself with the the royal we.) Obviously this means I've goofed or my model is mis-specified or something. Anyway, my birthday was a while back so I ran a fun midnight 15-miler up to Inspiration Point with Gazelle. I also caught a free sneak preview of the financial collapse documentary Inside Job . It's worth seeing, but if you already dislike investment bankers and have listened to the This American Life episode called The Giant Pool of Money , the...

Can anybody out-google me to a list of the names of those who died in the Persian Gulf War? Actually, I don't need names, I need service unit, home county/state of record, and date for all US military fatal casualties from Jan 1, 1990 to October, 2001. DoD press releases usually have this information, but they don't always issue a statement, and an incomplete set doesn't help that much. Monthly recruiting goals by service branch from 1990-2006 would be dreamy too. Otherwise, I'm 1600th in the FOIA line.

And with that, I probably won't be posting much for a while. I'm hoping to be on the job market this year, which means military labor supply research 12+ hours a day, 6.5 days a week for the next couple months. I'm signed up to present in a lunch seminar in a month, then if all goes well, job applications go out mid-November. I unsubscribed from a ton of blogs, removed a bunch of sites from my bookmarks toolbar, cook less, and run only enough to stay sane. Wish me luck, and if you're a potential employer, say, at a great school in a great location, I would be an awesome colleague and an excellent teacher and researcher.

I got way more excited by this project yesterday when we finished teaching Stata and started talking about what we could actually do with the data. There was a pretty great question about the effect of community-based health insurance on health care utilization in Rwanda which I totally thought we could answer with our panel data using individual fixed effects. Only panel data isn't any better at getting you causal inference if there are unobserved shocks that are correlated with the first-differenced x's and y's. (That is, OLS is biased if there are unobserved e's correlated with both x and y; panel data instead regresses delta-y on delta-x, and if you use individual fixed effects you eliminate the e's. Only you still don't observe delta-e, and if that's correlated with delta-x and delta-y, you've got the exact same problem you had before. Ugh.) Still a cool idea worth looking into. ANYWAY, switched to a far swankier hotel this morning, hoping to get ou...

After reading a bunch of Wallace Stegner and watching John McCain lose Nevada, New Mexico, and Colorado after he said during the campaign that the Colorado River Compact might need to be renegotiated, I decided to read up on the matter. I just finished Norris Hundley's Water and the West: The Colorado River Compact and the Politics of Water in the American West . Normally when I try and think of research ideas I just read the newspaper looking for examples of bad journalism (if there's two of something it's a trend), or I just try and think of arbitrary situations or quirks that lead to natural experiments, from which I could get some sort of easily econometrically identifiable causal estimate or program evaluation. (Title IX causing arbitrary variation in girls' sports participation, catastrophic injury causing arbitrary variation in time a rookie QB rides the bench before starting, etc.) This time I thought I'd pick some large general topic I'm interested i...