THIS AND THAT FOR OCTOBER

Here are a few interesting articles from the past few weeks:

Job Prospects for College Grads

In "Fear of a College-Educated Barista", Derek Thompson writes about the job prospects of college graduates and cites research about the prospects of STEM-educated graduates versus those in the liberal arts. Though he indicates that grads' prospects are better than those without a degree, he also says:

"This happy news comes with an important asterisk. A large chasm has opened between the fates of young liberal-arts majors and their peers in STEM (science, tech, engineering, and math) fields. The former are struggling to find work that pays, at least before their late twenties. The latter are mostly finding lucrative work after they graduate."

and:

"Indeed, the gap between humanities and STEM students is striking. Underemployment afflicts more than 50 percent of majors in the performing arts, anthropology, art history, history, communications, political science, sociology, philosophy, psychology, and international affairs."

Are Polling Margins of Error Accurate?

David Rothschild and Sharad Goel report in The Upshot on research done by Andrew Gelman and Houshmand Shirani-Mehr on election polling. The researchers examined the results of over 4000 polls in state-level presidential, Senate, and Gubernatorial races from 1998 to 2014. They found that the margin of error in the polls was twice what was reported with the polling results; instead of a margin of error of 3%, the actual error margin was 6-7%. 

This increase from the stated error margin is due to several factors, according to the authors:

• "voter screens" - often, pollster will try to correct error by asking about the likelihood of voting, but this will often increase the error margin rather than decreasing it.
• non-response error - often, voters supporting the trailing candidate will understate the likelihood of voting in the election, affecting the accuracy of the poll.
• analysis error - pollsters may analyze data very differently, leading to different conclusions drawn from the same database.
• survey working and interviewer bias - though these are viewed as less important in election polling, they may make for large differences in issue-based polling.

As the authors write: "...in reality, we find that the polling average can easily be two points off from the final vote share. For any given race, the polls are consistently too high for one of the candidates and too low for the other."

The Need to Revive Vocational Training

In Forbes, Nicholas Wyman writes about the desperate need to bring back vocational training in schools.

The author argues that, though 2/3 of high schools go on to college, only 40% eventually graduate, many carrying large debt loads. And, even then, those in particular majors are under-employed (see above post).

Wyman writes: "The U.S. economy has changed. The manufacturing sector is growing and modernizing, creating a wealth of challenging, well-paying, highly skilled jobs for those with the skills to do them. The demise of vocational education at the high school level has bred a skills shortage in manufacturing today, and with it a wealth of career opportunities for both under-employed college grads and high school students looking for direct pathways to interesting, lucrative careers."

If you are patient enough to get through all of the ads associated with the Forbes site, this is an interesting read.