Archive for the 'Methodology' Category

Most researchers use survey data, but more and more researchers are using “wild” data, which is defined as data not produced for research purposes. In fact, several PSC researchers are part of an NSF/Census project, which explores the usefulness of “wild” data ranging from administrative data (Social Security death index, Social Security earnings data) to data harvested from the web.

Below are several examples of informative posts based on web-based data:

The New Secessionists: Plotting whitehouse.gov secession petitions
Neal Caren | Big Data blog
November 14, 2012

This post shows the origin of each of the signers of the wave of secession petitions on the whitehouse.gov website via a county-based map. It also includes an explanation of how this was done. Many of the posts on Caren’s Big Data blog are excellent tutorials for the fundamentals of quantitative text analysis for social scientists.

It is also useful to refer to the history of secession petitions in the US, provided here:

10 facts about Secession
Kevin Robillard | Politico
November 14, 2012

The second example of an application of wild data, comes from a post about ‘mapping racist tweets’ based on content on Twitter immediately after Obama was re-elected to his second term:

Mapping Racist Tweets in Response to President Obama’s Re-election
floatingsheep.org
November 8, 2012

Note that a Harvard Ph.D. student used Google search data to study the under performance of Obama in 2008, which he atttributed to racial animus.

The Effects of Racial Animus on a Black Presidential Candidate: Using Google Search Data to Find What Surveys Miss
Seth Stephens-Davidowitz | Harvard
June 9, 2012

The popular press version of this is here:

Can Google Predict the Impact of Racism on a Presidential Election?
Garance Franke-Ruta | The Atlantic
June 11, 2012

And finally, the Google NGram project has useful data for researchers. Here’s an article from the Economist where the “data is” vs “data are” question is examined. More pertinent to researchers might be the evolution of “Negro man” to “colored man” to “African-American man” in common usage.

Data or datum?
K.N.C. | The Economist
July 13, 2012

And, here’s the link to the Google Ngram Viewer. Of course, you’ll want to have access to the data. Here’s the raw data is available for download link

Research is Hampered by New Limits on Death Records
Kevin Sack | The New York Times
October 8, 2012

A shift by the Social Security Administration to limit access to its death records amid concerns about identity theft is beginning to hamper a broad swath of research, including assessments of hospital safety and financial industry efforts to spot consumer fraud.

A quote by the senior project manager of the Nurses’ Health Study is apt:

the new policy ha(s) “thrown us back to the pre-Internet era where you’d start looking in the phone book for someone with a similar name and sending out a bunch of letters.”

On the first Friday of this month (October 5, 2012), the BLS released its job figures just like it does every month. This report had unemployment dropping from 8.1 percent to 7.8 percent. Immediately, there was an outcry that somehow BLS had cooked the figures to help the Obama campaign, the most notable of which was a tweet by Jack Welch, former CEO of General Electric:

Below are a few examples of that view or commentary on it as well as some more thoughtful posts on the noisy nature of the data.

Enabling the jobs report conspiracy theory
Brendan Nyhan | The Swing States Project Blog, Columbia Journalism Review
October 8, 2012

The jobs truther movement
Patrick Reis | Politico.com
October 5, 2012

Steep drop in unemployment rate spawns conspiracy
Scott Mayerowitz and Christopher Rugaber | AP
October 5, 2012
Great opening paragraph:

Sasquatch might as well have traipsed across the White House lawn Friday with a lost Warren Commission file on his way to the studio where NASA staged the moon landing.

Conservatives Jobs Conspiracy is Nuts
Robert Schlesinger | Newsweek
October 5, 2012

Don’t Trust this Number in This Jobs Report (or Any Jobs Report)
Derek Thompson | The Atlantic
October 5, 2012

How Bureau of Labor Statistics Tames Volatile Raw Data for Jobs Reports
Catherine Rampell | Economix Blog, New York Times
October 5, 2012

Canada is in a fix with its 2016 mid-decade Census on the horizon. It made a major change to its census operations with the 2011 Census, choosing to make The National Household Survey (long-form) voluntary. This was done on the behest of the Conservative government, not the advice of the statisticians in StatsCan [See here, here and here for links to this issue].

Final Report on 2016 Census Options: Proposed Content Determination Framework and Methodology Options
[Executive Summary] [Full report]

Stick with old-school census, says StatsCan
Jennifer Ditchburn | iPolitics
August 30,2012
Reaction from the press is pretty damning, showing that Canada still doesn’t know the full impact of the decision to go with a voluntary census. Here’s a comment on whether or not there is response bias in the National Household Survey:

“Are we totally off, slightly off, right on? That would be difficult to determine,” said Marc Hamel, manager of the census program at Statistics Canada.

Harper government’s assault on reason, scientists, ‘Orwellian’ and ‘alarming,’ warns pollster
Alice Funke | The Hill Times
September 10, 2012

This article is really belongs in a collection of “Death of Evidence” articles. But, it so nicely aligns with the above article as the first ‘assault on reason’ occurred when the government cancelled the mandatory long-form census.

With the departure of Bob Groves from the Census Bureau [See here for his exit interviews], it is important to reflect on the Census Bureau director appointment.

The Census Bureau director is a political appointment. But, Congress did accomplish something before their August recess. They passed the Presidential Appointment Efficiency and Streamlining Act. This act eliminates confirmation requirements for about 200 offices.

More important is that this legislation specifically addresses the Census Bureau director position. It still requires Senate confirmation, but it will now be a fixed 5-year term; one term spans the the census (2017-2022 while the other spans the planning stages (2012-2017, etc.). In addition, it has a Qualifications provision that states that “such appoinment shalll be made from individuals who have a demonstrated ability in managing large organizatons, and experience in the collection, analysis, and use of statistical data.”

Presidential Appointment Efficiency and Streamlining Act
See page 9 for the discussion of the Census Bureau director

As an aside, various friends of the Census Bureau have tried to have the Census Bureau director appointment be a non-political appointment – running for a 5-year fixed term. The last attempt was by Carolyn Maloney in the 111th Session of Congress. Her bill was just focused on the Census Bureau. Here’s some background materials.

In the end, this failed in the House, while the companion bill passed in the Senate.