Friday, February 22, 2013

Data link roundup (week of February 22, 2013)

The week's top data analysis links...


It appears network effects are statistically significant predictor of grade point average in high school students.


Using data from the U.S. Centers for Disease Control and Prevention, analyst Leon Farrant illustrates the effectiveness of vaccinations.
Source: Leon Farrant with data from CDC


Randy Krum, of Cool Infographics, analyzes the strengths and weaknesses of the informational graphics included in the "enhanced State of the Union" address. Krum's piece includes useful reminders for all data analysts and visualization specialists.
"One of the biggest risks with data visualizations and infographics is what I call the Risk of Negative Impression.  The idea is that while good visuals can quickly leave a good impression with your audience, if your visualizations are incorrect or flawed, you can leave a bad impression just as quickly and effectively."


"Missed Connections" are personal ads posted on Craigslist in hopes of connecting with someone seen but not spoken to. (Or spoken to with no way to follow up...) The ads generally list the day, time, and location of the missed connection.
Source: Buzzfeed

Dorothy Gambrell mapped the frequency of misses by location, based on the state where the ad was posted.

Tuesday, February 19, 2013

Contrary to popular belief: Births to older mothers not at historic high

Sometimes "common knowledge" is not knowledge at all...

There has been considerable hand-wringing in the press over births to women over the age of 35. According to the Telegraph, older mothers are "driving up the birth defect rate." The New Republic suggests that older parents will "upend American society." And Slate, generally progressive, bemoans the "feminist fertility myth."

As a demographer, I am aware that birth rates in age groups 35-39, 40-44, and 45+ have been rising since the late 1970s for ages 35-39 and 45+ and since the early 1980s for the 40-44 set.

But as the granddaughter of a woman who birthed her youngest child when she was nearly forty, my skeptic antennae went up. My grandmother may have been a singular woman, but she was hardly a statistical aberration.

As a historical demographer, who has some experience with fertility and mortality rate trends over the past century (and the century before), I suspected the evidence of a "major shift" might be lacking, and I set out to prove it.

While the fertility trend of delayed childbearing and concomitant increase in birth rate for mothers age 35 and older is occurring in nations around the world, I will focus on data from the United States.

The National Vital Statistics Reports, published by the U.S. Centers for Disease Control and Prevention, provide historical birth rate data by age of mother as far back as 1970. Earlier data must be compiled from a variety of other sources including the older, and often PDF-scan-only Monthly Vital Statistics Reports and the U.S. Statistical Abstract.

From those sources I collected data as far back as 1920, with complete annual data from 1935-2010 (the most recent year for which complete data are available). The historical birth rates (births per 1,000 women) are shown in the chart below.
Sources: NVSR, MVSR, Statistical Abstract, and author's calculations

Zeroing in on ages 35 and older, to show more detail on the chart scale, we see that birth rates are indeed higher in 2010 than they were in 1970. Specifically, birth rates are 45% higher for ages 35-39, 26% higher for 40-44, and 40% higher for ages 45-49 in 2010 than in 1970.

Sources: NVSR, MVSR, Statistical Abstract, and author's calculations
But we do not need to go back very much farther to find a time at which birth rates were higher than they are today: In 1965 birth rates were higher for all ages 35-49 than they were in 2010.

If we go as far back as 1920, birth rates were nearly twice as high for ages 35-39, were three times higher for 40-44 year olds, and were more than 5 times higher for ages 45-49 than the rates in 2010.

In short, the phenomenon of increasing birth rates for women in their late 30s and 40s is as much a return to historical patterns as a deviation from them.

To be clear, some pregnancy risks do increase with advanced maternal age:
"older age strongly increases a woman's chances of at least three untoward outcomes—namely, stillbirth, miscarriage, and ectopic pregnancy" (Stein and Susser 2000)
However, there is considerable evidence that socio-economic factors play the most important role in birth outcome - perhaps far outweighing the age effects (Cohen 2012, Azimi and Lotfi 2011, Stein and Susser 2000).

As for the current focus on trends in the past four decades, I suspect the ease of collecting data from 1970-present is partly the cause of the focus on a "huge" increase in childbearing at older ages... when, in reality, rates have fallen if you look back just 5 years more.

I am happy to share the raw data upon request. Feel free to contact me for more information.

Friday, February 15, 2013

Data link roundup (week of February 15, 2013)

The week's top data analysis links...


RCLCO's new report highlights growth hotspots around the United States based on job growth, working population, population growth, and income. Austin, TX tops the list for large metro areas, Raleigh-Cary, NC for midsize, and Bismarck, ND for small metros.
Source: RCLCO
And in a new analysis, documents the relative "age" of major metropolitan areas in the United States, using percent of housing built before/after 1940 as an indicator. Perhaps unsurprisingly, Boston and Providence rank as the nation's "oldest," while Las Vegas and Phoenix are the relative newborns of the bunch.


A recent article from the Los Angeles Times does an excellent job of explaining the link between sending-country birth rate and immigration... they just forget that citizens are migrants, not immigrants.

The Atlantic Cities piece on Bosnians in St. Louis is a clever blend of demographic concepts, personal anecdote, foreign affairs, public policy, social network theory, and assimilation theory.


And because it bears repeating...

New research published in the Journal of Personality and Social Psychology offers proof that there is more within-group variation between men and women than there is between-group variation.
Source: Science Daily

Fact Check: Citizens can't be immigrants

In what seems like a rather large blunder, the Los Angeles Times forgets a key fact in an otherwise well-written, demographically-focused piece on immigration policy...

Times writer Jonathan Last describes the association between fertility rates and international immigration* by using the flow of people from Puerto Rico to New York City as an example.

This flow, largely taking place between the 1920s and 1940s, does serve as an excellent illustration of the factors that influence population change.
In the 1920s, Puerto Ricans began to trickle into the United States. Their numbers accumulated slowly, and by 1930, there were 50,000 Puerto Ricans living in the U.S. (nearly all in New York City)... But from 1955 to 2010, the number plummeted. Even though the population of Puerto Rico had nearly doubled in that time, the total number of Puerto Ricans moving to the United States in 2010 was only 4,283...What happened is that Puerto Rico's fertility rate imploded.

The problem?

Puerto Rico has been part of the United States since the end of the Spanish-American war, and Puerto Ricans have been U.S. citizens since 1917. Puerto Ricans couldn't "trickle into the United States" because they were already in the United States.

Despite social commentary to the contrary, Puerto Ricans were migrants, but not immigrants.

*Using the term "international immigration" is redundant, of course, as "immigration" means moving to a foreign country, but I want to be clear in this point.

Thursday, February 14, 2013

Love and marriage

In honor of St. Valentine...


According to recent news, the longest-married couple in the United States tied the knot more than 80 years ago.

Valentine image source
While this may be an unusual feat, marriage duration has (on average) increased in recent years. This trend persists despite the fact that marriage rates are declining.

80 percent of marriages last at least 5 years, and 68 percent last 10 years or more, according to data compiled by the U.S. Centers for Disease Control and Prevention based on the National Survey of Family Growth (2006-2010).

This is an increase from the 2002 survey, in which 78 percent of marriages last at least 5 years and two thirds last 10 years or more.


According to many news sources, condom sales are highest in February (in the US and in India, for example). However, a National Institutes of Health study shows that increased condom sales do not necessarily translate to increased condom usage.

Nielsen research notes a corollary trend: an increase in sales for home pregnancy tests in March. From a February 2008 news release published by Nielsen:
First Comes Love, Then Comes...
Perhaps as a result of Valentine’s Day romance, more pregnancy and infertility test kits are sold approximately six weeks after Valentine’s Day than at any other time of the year. Consumers spend more than $15 million*on pregnancy and infertility test kits during the second, third and fourth weeks of March, with the third week of March ranking number one** in sales.
Notes: *Three weeks ending March 24, 2007 showed total sales of $15.4 million for pregnancy and infertility test kits in U.S. food, drug and mass merchandiser stores, including Wal-Mart. **One week ending March 24, 2007 showed total sales of $5.2 million for pregnancy and infertility test kits in U.S. food, drug and mass merchandiser stores, including Wal-Mart.
Despite Nielsen's sensational headline, births are actually highest in late summer and early autumn, as a result of pregnancies in late autumn and early winter of the prior calendar year. This trend, known as a "seasonal cycle in fecundability" is well documented in the scientific literature.

Wednesday, February 13, 2013

Valentine's by the numbers


An interactive map allows users to explore Valentine's customs around the world.
Interactive Valentines Map


Valentine's Day spending in the United States is expected to top $17.6 billion, according to new survey results released by the National Retail Federation.

Spending is expected to be highest on jewelry ($4.1 billion) followed by $3.5 billion on a date, $1.8 billion on flowers, $1.5 billion on candy, with the balance accounted for by clothing, gift cards, and other items.

That works out to $126 per person celebrating the holiday. This is an 8.5 percent increase over spending in 2012, and the highest in the survey's decade-long history.
"...the average person planning to spend $74.12 on their spouse or significant other, up from $68.98 last year. Additionally, consumers will spend and average of $25.25 on their children, parents or other family members and $6.92 on friends. Valentine’s Day is a great day for pet owners to show their furry friends just how much they mean: the average person will spend about $4.52 on their pets."
Source: National Retail Federation 2012 Valentine’s Day Consumer Intentions and Actions survey


U.S. producers sold more than $18 million in cut roses in 2011 (estimated wholesale value for all operations with $100,000 or more in sales).
Source: USDA National Agricultural Statistics Service Floriculture Crops 2011 Summary

To support holiday spending, shoppers can choose from nearly 24,000 jewelry stores and more than 16,000 florists nationwide.
Source: U.S. Census Bureau, County Business Patterns


To satisfy millions of sweet-toothed customers, the NECCO candy company produces approximately 100,000 pounds of Sweethearts conversation hearts each day for 11 months of the year.

This adds up more than 8 billion heart-shaped candies annually.
Source: New England Confectionery Company

A 2009 study from Nielsen research showed that Valentines week accounts for more than 5 percent of annual chocolate sales.
Source: Nielsen

Tuesday, February 12, 2013

Race, culture, and data of Mardi Gras

Laissez les bon temps rouler!

Sociological Images has a series of several intriguing posts about Mardi Gras from a social and cultural perspective. Here is a sample of their posts:

The Atlantic Cities profiles "the largest street fair on the planet," with an estimated 6 million revelers at Carnaval in Brazil:
"There will be 14,454 policemen and 985 traffic guards on duty this Carnaval. 1,050 urban cleaning workers will collect over 600 tons of trash... 'Operation Dry Law' ticketed 786 drunk drivers; and the 'Shock of Order' seized 3,700 cans of beer, 1,100 bottles of water, 71 coolers, costumes, spray foam and cigarettes from unlicensed street vendors."

And the Census Bureau is tweeting (@uscensusbureau) about king cake and other Mardi Gras facts and figures.

Friday, February 8, 2013

Data link roundup (week of February 8, 2013)

The week's top data analysis links...


In an effort to make cycling safer, analysts and officials in Minneapolis are using a data-based approach to understanding the causes for bike crashes.


The Amish are the fastest growing religious group in North America, according to the 2010 U.S. Religion Census. High fertility rates and low rates of religious attrition are cited as key reasons for the rapid growth.


Decade-to-decade population growth, by county, in the United States from 1910-2010:
Source: U.S. Census Bureau

Thursday, February 7, 2013

Men aren't from Mars

Men and women are more alike than they are different, after all! (Go figure...)

New research published in the Journal of Personality and Social Psychology offers proof that there is more within-group variation between men and women than there is between-group variation.

From Science Daily...
"People think about the sexes as distinct categories," says Harry Reis, professor of psychology at the University of Rochester and a co-author on the study to be published in the February issue of the Journal of Personality and Social Psychology. "'Boy or girl?' is the first question parents are asked about their newborn, and sex persists through life as the most pervasive characteristic used to distinguish categories among humans."
But the handy dichotomy often falls apart under statistical scrutiny, says lead author Bobbi Carothers
Source: Science Daily

For more information see: "Men are not from Mars," a more thorough summary of the publication "Men and women are from Earth: Examining the latent structure of gender."

Friday, February 1, 2013

Data link roundup (week of February 1, 2013)

The week's top data analysis links...


This week three links about the foreign-born population in the United States stood out for their clarity:
  1. The Pew Research Center released new data, from the 2011 American Community Survey, on characteristics of the foreign-born population in the United States.
  2. Slate explained how demographers develop estimates of unauthorized migrants.
  3. NPR news explained the difference between the terms illegal immigrant, undocumented, and (my preference) unauthorized migrant.


A new online mapping tool illustrates the persistent problem of urban decline in Detroit. The map shows properties that are delinquent (yellow) in foreclosure for delinquent taxes (orange) or foreclosed (red).

A summary in The Atlantic Cities reports that more than 150,000 (39%) of the city's 384,861 properties are in tax distress. The map illustrates the magnitude of the problem.


In honor of Super Bowl Sunday...
NFL allegiance, by county, based on each county's number of team "likes" on Facebook.