Friday, August 30, 2013

Data link roundup - visualize it (week of August 30, 2013)

The week's top data analysis links...
This week's theme: data visualization (overused term, but an important concept)


VISUALIZATION TOOLS

Over the past few weeks I've been working on data visualizations for a project at my day job. For prior projects, I had been using the "motion charts" Google gadget to make interactive graphics, but that tool (sadly) has been disabled.

In my search for a replacement, I learned about a wide variety of free online data visualization tools. A few resources worth noting are: Weave, Tableau, and Datawrapper. All three allow users to develop interactive charts and graphs. Tableau and Weave also offer mapping capabilities.

And if those options still don't fit your needs, try this directory of about a dozen other tools worth further exploration.


VISUAL EXPLORATION OF DATA

Sinan Aral's article in Harvard Business Review explains how the New York Times uses visual representations of data to understand patterns that a simple table might not illuminate...



BEST CHART OF THE WEEK

I'm a fan of images that convey a wealth of information with a simple concept. Population-adjusted country size maps, when done well, can do just that...

Source: Gizmodo


Friday, August 23, 2013

Data link roundup (week of August 23, 2013)

The week's top data analysis links...


RANDOM BITS


IN CASE YOU MISSED IT...


BEST CHART COMIC OF THE WEEK

For those of you preparing for the Fall 2013 semester...
Source: PhD Comics

Thursday, August 22, 2013

Odds for betting on a birthday

Earlier today I tweeted about my maternity leave memo...
My maternity leave memo made finance staff laugh. What? Don't all moms incl a table w/ statistical likelihood of birth by week of gestation?
— Beth J (@DataGeekB)
The tweet generated as much commentary (and laughter) as the memo did, so here are some highlights of the data and answers to reader questions.

While plenty of practitioners will tell you that a "full term" birth is considered to be 40* weeks (280 days) of gestational age, the numbers simply don't bear that out. In recent years, both the median and the mode place a baby's most likely birth date at week 39 with a standard deviation of 7 to 9 days (depending on the source).
Source: CDC Births: Final Data for 2011, compiled by author
Note: Data shown here are for singleton births only, as multiple births have a higher likelihood of preterm delivery
I suspect the misconception about the most common length of pregnancy is a matter of old data.

In 1990 the median delivery date for a singleton** birth was in the 39th week, but the mode was 40. Perhaps most shockingly, in 1990 more than 10 percent of births occurred at 42 weeks or later!

Birth timing is not entirely predicted by biology. Medical interventions abound, as one in three singleton births are via cesarean delivery, so the distribution of births by gestational age is highly influenced by current medical standards.

Research from the mid-2000s suggests that mothers and infants are at higher risk of birth complications if pregnancy is allowed to extend beyond 42 weeks, so current practices encourage the use of medical interventions to ensure delivery before 42 weeks. This likely explains the substantial drop in post-term births between 1990 and 2011.

The data: Distribution of singleton births by gestational age
2011 2010 2006 2005 2000 1990
Under 28 weeks 0.6% 0.6% 0.6% 0.6% 0.6% 0.6%
28-31 weeks 1.0% 1.0% 1.0% 1.0% 1.0% 1.1%
32-33 weeks 1.2% 1.2% 1.3% 1.3% 1.2% 1.2%
34-36 weeks 7.3% 7.5% 8.1% 8.1% 7.3% 6.8%
37-38 weeks 25.7% 26.7% 28.9% 28.3% 24.4% 19.4%
39 weeks 30.1% 29.1% 26.2% 26.0% 24.9% 22.0%
40 weeks 19.9% 19.7% 19.4% 19.8% 21.9% 23.0%
41 weeks 8.6% 8.5% 8.6% 8.9% 11.3% 14.4%
42+ weeks 5.7% 5.6% 5.8% 6.0% 7.5% 11.5%
Source: CDC, National Vital Statistics Reports, Births: Final Data for 2011

Keep these stats in mind the next time you're betting in an office "guess the birth date" pool.

Have questions about the data?
Send me a message!


NOTES:
*Births prior to 37 weeks are considered to be pre-term (or premature). Births at or after 42 weeks are considered to be post-term. The wide range in the middle, from 37 to 42 weeks, truly reflects full-term gestation.

**Singleton births are births to an individual child, and excludes twins, triplets, or other multiple births.

Friday, August 16, 2013

Painfully obvious research results - data link roundup (week of August 16, 2013)

The week's top data analysis links...
This week's theme: The Duh Files.

Did we really need a study to "know" this?


BIRTH CONTROL BLUNDERS

Culling statistics from the National Survey of Family Growth, researchers found that the "withdrawal" or "pull-out" method of birth control is surprisingly common. 31 percent of surveyed women used withdrawal as a form of birth control at least once.

But (not surprisingly) the method is terribly ineffective. Unintended pregnancy rates were 21 percent among those who used the withdrawal method.

In an interview published by the National Institutes of Health, study author Dr. Annie Dude notes:
"Our study showed that use of withdrawal for contraception is very common, but it doesn't work as well as other methods"
The full study will be published in the September issue of Obstetrics & Gynecology.


RUNNING IN HEELS

New analysis published in the Research Journal of Applied Sciences, Engineering and Technology shows that when running in heels, energy expenditure increases with heel height... at least for the dozen women who participated in the study.

No word on the effects of heel high on the risk of breaking an ankle...


AND THERE'S MORE...

And other entries into the "painfully obvious results of research studies" category:


Now don't get me wrong, I'd much rather see silly, and seemingly painfully obvious correlations than paid-for-by-the-producer clinical trials that later are proven to be flawed, or research that shows a strong correlation and jumps to the conclusion that correlation means causation (when, in fact, confounding factors are to blame).

And we all know that most of these studies include findings that are deeper, more complex, and more useful than the titles imply.

For example, the University of Georgia's findings that abstinence only education does not work at reducing teen sexual behavior or teen pregnancy seems obvious. Well... it seems obvious to most of us, but some public policy decisions would benefit from wider distribution of the the evidence.

Then again, sometimes it's just fun to wonder what on earth the researcher was thinking (and how on earth the project got funded).

And really, did we need the BMJ to publish an article to know that sword swallowing is a dangerous profession???

Friday, August 2, 2013

Data link roundup (week of August 2, 2013)

The week's top data analysis links...
Last week we focused on mortality stats, so this week we're looking at the other end of the life course spectrum: babies and mommies.


PAMPERING A BABY (not as easy for some families)

New research published in the journal Pediatrics reveals a harsh reality for low-income families: diapers cost more than many low-income families can afford.

Lack of adequate diaper supply can result in negative outcomes for caretakers and children, including: family stress, anxiety, depression, and child health problems (e.g. increased risk of urinary tract infection).

Data were...
...derived from a cross-sectional study in 877 pregnant and parenting women. Mothers completed surveys on topics related to mental health, basic needs, and health care use. Logistic regression was used to estimate the relationship between diaper need and psychosocial correlates.
The study found that nearly one third of mothers in the survey reported a diaper shortfall. Hispanic women were significantly more likely to report diaper need than African American women.

(See also: diaper report summary on The Atlantic CITIES)


BREASTFEEDING BY THE NUMBERS

Breastfeeding rates have been climbing slowly but steadily for the past several years, according to new data released by CDC.

Rates of breastfeeding at 6 months are highest in Idaho, California, and Oregon according to the CDC's 2013 Breastfeeding Report Card. Rates in those three states exceed 70 percent. Rates are lowest, below 20 percent, in Mississippi.


SEX SELECTION AT BIRTH

New analysis from PRB highlights the problem of son-preference sex selection at birth in Armenia, Azerbaijan, and Georgia,


BEST CHART OF THE WEEK

Partly based on fertility and mortality trends, completely based on the latest United Nations population projections, The amazing, surprising, Africa-drive demographic future of the Earth, in 9 charts is worth a look.

Here is a sample (1 of the 9 interactive graphics):

Source: Max Fisher, Washington Post


IN CASE YOU MISSED IT...