ChristopherGandrud

At least she ends on a positive note. Things could get better!

The whole Darien Scheme thing might not have worked out, but the Company of Scotland (founded 1698) had a pretty great flag.

From: The Verge

[Irish times](http://m.irishtimes.com/newspaper/breaking/2013/0207/breaking1.htm?via=latest) on IBRC liquidation legislation.

Pretty cool Google search picture today, though why is the [DLR](http://www.tfl.gov.uk/corporate/modesoftransport/1530.aspx) not in the legend? Did they merge it with the Victoria line?

Not the greatest example, because you can't actually distinguish between four spaces and a tab, but it's in there. I promise. (This gets us to the infamous tabs vs. spaces debate. See: [An Eternal Holy War](http://www.jwz.org/doc/tabs-vs-spaces.html)).

Nothing says a rehabilitated bank like a friendly new hippo logo: http://www.hypo-alpe-adria.com/115/home.nsf/pages/about_us-history-162

library(plyr) Data <- ddply(Data, .(country), transform, VariableLag1 = c(Variable[2:(length(Variable))], NA)

That's it. Just remember to have the variables in time order. Thanks to [this post](http://stackoverflow.com/a/8910583/1705044) on StackOverflow.

I really enjoyed [this working paper](http://gking.harvard.edu/gking/files/teach.pdf) by [Gary King](http://gking.harvard.edu/) and [Maya Sen](http://scholar.harvard.edu/msen/). It's called "How Social Science Research Can Improve Teaching". The paper describes how they have applied empirical social science findings to improve their teaching of the graduate social science stats class at Harvard. They point to three general findings: 1. social connections motivate, 2. teaching teaches the teacher, 3. instant feedback improves learning. Over the past two semesters while putting together courses at [Yonsei University](http://www.yonsei.ac.kr/eng/) I've tried to incorporate some of their suggestions and tools. The most successful technique so far seems to be making as many assignments and forms of assessment as possible involve social connections (the final exam is still sit silently and write). However, there are two general issues that I've hit that have sometimes made it difficult to use these techniques: - Many of my students' computers have difficulty running some of the tools that enable instant feedback and teaching other students. In particular many students were not able to successfully use the [NB](http://nb.csail.mit.edu/welcome) collaborative text annotation software. I am not entirely sure why this was, but it does seem that web technologies that are treated as standard in America and Western Europe are often not easily handled by computers in Korea (and vice versa, though that is a whole other post). I really need to do more research into this. - Probably a combination of bashfulness about English-language ability and experience with an education system that rarely affords opportunities for student-generated knowledge leads many of the students to refrain from fully engaging in the sort of activities that King and Sen promote for having students teach each other. Of course this is just a crack pot theory that I have little systematic evidence for. But it would be interesting to look into further, especially to help me better address the issue.

After writing my [last post](http://christophergandrud.tumblr.com/post/32353571459/what-do-standard-errors-mean-when-the-sample-is-the) I read [this article](http://www.buzzfeed.com/annanorth/majority-of-women-say-birth-control-keeps-them-fin) on a survey about why women use birth control. The headline for the article is: > Majority Of Women Say Birth Control Keeps Them Financially Solvent The results of the survey definitely fit my prior on the subject. But, since my mind was primed to think about samples vs. populations after the last post, the headline bothered me. The research the article is talking about surveyed "2,094 women at 22 clinics around the [United States]". So clearly the headline should have read something like. > Majority of Women Visiting Reproductive Health Clinics Say Birth Control Keeps Them Financially Solvent This may seem like a trivial point, but I think its very reasonable to assume that the original survey may have a sampling bias that makes it difficult to generalize the survey findings to all women. Especially since people who visited the 22 clinics may have different economic backgrounds than other women, and even other women that use birth control.

This is a [good post](http://andrewgelman.com/2011/10/how-do-you-interpret-standard-errors-from-a-regression-fit-to-the-entire-population/) by Andrew Gelman to keep in the back of my mind when when doing research with a sample that is the population (i.e. *all* Federal Reserve Chairman speeches).