Most dangerous nations for women in 2018: US in top 10

Reuters recently published a news report, Factbox: Which are the world’s 10 most dangerous countries for women?. The results were the subject of a post at Feminist Philosophers blog, where I occasionally visit and even understand a bit, without being a philosopher nor much of a feminist. I reacted with incredulity to the results.

United States is the 3rd most dangerous country for sexual violence?

Sexual violence is defined by the survey as follows:

“This includes rape as a weapon of war; domestic rape; rape by a stranger; the lack of access to justice in rape cases; sexual harassment and coercion into sex as a form of corruption.”

How often was rape used as a weapon of war in the United States in 2018?

Yes, this fact check—based on investigative journalism that was funded, produced, and published by the Thomson Reuters Foundation—reports that women in the US are more at risk of sexual violence than in Syria! The US is considered more dangerous for women than nations where female genital mutilation is common, and untreated obstetric fistulas ruin lives.

Top 10
01 India
02 Democratic Republic of the Congo
03 USA
03 Syria
05 Congo
06 South Africa
07 Afghanistan
07 Pakistan
09 Mexico
10 Nigeria
10 Egypt
10 Somalia

India versus #MeToo

On the Foundation’s website, I did find a point of light, but bracketed by tragedy. Near Mumbai, two adult men raped a 7 year-old girl as she waited for her parents to pick her up from school. The attack was so brutal that she required hospitalization, but she will survive. The rapists were apprehended and are in custody.

The point of light is India’s Prime Minister Modi. He recently introduced the death penalty for acts of rape of girls under 12 years of age, in response to widespread public demand for such measures. Female children are 40% of the victims of all 40,000 rapes reported in India annually.

In contrast, #MeToo mostly involved women in America’s highest echelons of status and privilege.

To suggest that #MeToo revealed conditions in the US that are comparable to the sexual violence of India’s 16,000 child rapes annually is a harmful misrepresentation. Yet the Thomson Reuters survey did not hesitate to do so.

Sorry to see a respected foundation circulating such flawed & misleading data. The US is NOT worse for sexual violence than Somalia & Congo. Not even close. Why play games?
Women at risk around the world desperately need sober research—not hype & spin.

— Christina Sommers (@CHSommers) June 26, 2018

Non-sexual violence supposedly worse in US than Somalia

The survey defined non-sexual violence as “conflict-related violence; domestic, physical and mental abuse”. These are the survey findings.

Top 10
01 Afghanistan
02 Syria
03 India
04 Yemen
05 Pakistan
06 USA
07 Saudi Arabia
08 Democratic Republic of the Congo
09 Mexico
09 Somalia

The United States is deemed more dangerous for women than Somalia, Democratic Republic of the Congo, Mexico, and Saudi Arabia.

I am surprised to see Mexico mentioned. Maybe that is due to Mexico’s porous northern border, as violence against women flows southward from the United States?

Immigration sanity check

According to the poll, Guatemala and Honduras are safer, less physically dangerous places for women than the United States. If that is true, then most of the requests for asylum in the US made by women and children fleeing these two countries have no justifiable basis!

If levels of domestic abuse, gang violence, and sexual victimization of women and children in the United States is greater than in El Salvador, then likelihood of harm would be increased—rather than alleviated—by seeking asylum in the US.

The source of these findings about comparative geographic danger to women is a highly respected news organization. Regardless, the results are flawed. Faulty polling methods and shoddy statistical analysis of responses are implausible culprits: A subject matter panel of over 500 experts in women’s rights provided the data from which the final results were derived. Although I wonder; who are these experts?

The U.S. is now ranked among the 10 nations considered to be the most dangerous for women, experts say:

— CBS News (@CBSNews) June 26, 2018

Blame falls on Reuters. Editorial oversight would have been in place throughout the project. The featured results would not have been released prior to vetting and final approval. This was not a minor mistake, but rather, a breach of Reuters standards of accuracy and freedom from bias in journalism.

Why politically motivated surveys are bad

An unrelenting narrative that the United States is SO terrible has been running in the background for years. I have heard it since we invaded Iraq. During the Obama years, I think it was broadly referred to as the end of (maybe the myth of?) American exceptionalism. Outrage is particularly strident now, under the Trump administration. I believe that warped perceptions, combined with agenda-driven media can result in grossly inaccurate investigative reports such as this.

Protecting women’s rights is important! Appeal to emotion (e.g. Trump Derangement Syndrome) should be resisted. Failure to recognize situational bias hurts the effectiveness of advocacy efforts. Regaining credibility is often elusive.


I was disappointed at the lack of methodological detail provided, e.g. criteria to normalize results across disparate country populations and rates of violence. Who were the 548 experts? I will check for further details. Perhaps I will follow-up with a statistical survey design post if I can find more information.


Credibility and the Internet: Queuing Theory

Don’t believe everything you read on the Internet!

This was old news about queues back in 1985. Yet it was written up as a journal article, and received coverage as though a new finding in the June 2010 issue of ScienceDaily, an online publication owned by Reuters.

M/M/1 queues, Kendall notation, and models of balking behavior are certainly useful. However, the concepts, and their accuracy as models, were well-established for at least forty years. This is true whether applying queueing theory to modelling the performance of computer hard-drives e.g. random arrival times for seek requests, or to consumer behavior when switching lanes because of long lines at the supermarket checkout.

The Wiley text book, Fundamentals of Queueing Theory, was published in 1998.

Earlier editions were published in 1983, and explain in detail the theory and application of the concepts presented in the journal article reviewed by ScienceDaily.

A little more about M/G/1

On Math StackExchange, I noticed a rare inquiry. If you’re curious for more about queues, go read my answer to this question, Kendall notation’s “General distribution”, what does that mean?

I found this comment endearing:

Oh I thought that this stuff wasn’t even used in real life jobs… I thought it was merely theoretical, but seems that I’m wrong!

I’m okay with the G general theory [G as the general case when you just don’t know what sort of service time distribution to expect] since I’m not required to study it for now (I’m following an academic course), I just wanted to understand what the G meant and you helped me in that. Do you have any experience with multi-class queues too?


Statistics comes to Swarthmore College

Some years ago, I studied mathematics and statistics. At that time, there was only one statistician among the mathematics department members, maybe the entire Swarthmore College faculty, Gudmund R. Iversen. He was my academic adviser. Professor Iversen was grey, tweedy and Norwegian. He always addressed me as Miss Kesselman, which helped alleviate my shyness at the time.

Professor Iversen got his PhD in statistics from Harvard University in 1969. I noticed only one other familiar name on that very short list of all Harvard Statistics PhD alumni: Columbia University political science and statistics professor Andrew Gelman PhD in 1990.

Lunch with Tufte

Professor Iversen had a group of colleagues, all statisticians from other academic institutions. They would visit Swarthmore to give lunchtime talks, or more typically, late Friday afternoon presentations to mathematical statistics students.

I recall one particular guest statistician. Edward Tufte was on the faculty of Princeton University, and had recently written his first book, The Visual Display of Quantitative Information. The venue was a small private room in Sharples dining hall. I was one of maybe 20 attending.

Tufte was high-strung and slightly fussy, with occasional flashes of humor. He handed out hardback copies of his book, admonishing us “not to dip them in the gravy” from lunch (there was no gravy at lunch). Tufte explained that he had to take out a third mortgage on his house to finance the production and publication of Visual Display. The book was gorgeous. The statistical graphs were unlike anything I had ever seen before. Tufte spoke at length about Charles Minard’s famous map representing Napoleon Bonaparte’s doomed Russian campaign. In the summer of 1812, Napoleon set out for Moscow with 440,000 troops. Only 10,000 returned.

Tufte spoke well. After a mild question and answer session, he retrieved copies of his book from us. I badly wanted to keep mine. For a little more Tuftese see my Chart Art post.

Statistics moves up in the world

During my time at Swarthmore College, statistics was considered a marginal field of study, at best. The current math department chairman, James England, referred to it as “cocktail party math”. Professor Iversen had tenure by the time I arrived, yet he didn’t have an office with the rest of the mathematics department. Instead, he still occupied the same room in the 1st level basement of the engineering building as he had since 1973, and a ten minute walk from the rest of the department.  It was an almost windowless room, with woven wool rugs on the floors and hung on the walls, which kept the air warm and dry. Naturally, the furniture was mostly mid-century Scandinavian modern.

Given that background, I was surprised and happy when Professor Iversen became the new department head in 1992! In 1993, the department name changed. Now it is the Swarthmore College Mathematics and Statistics Department. As far as I can tell, Professor Iversen kept his original office even while he was department chair. After twenty years, it was their turn to come to him.


I happened upon a pleasant review of Statistics in Society: The Arithmetic of Politics written by Iversen for the MAA (Mathematical Association of America) in February 2000. The book is actually a collection of 47 essays. Overall, the review is positive. Excerpt:

The book grew out of activities supported by what is known as the Radical Statistics group, a twenty-five year old group unknown to me before reading this book. Radical Statistics is “a group of statisticians and others who share a common concern about the political assumptions implicit in the process of compiling and using statistics, and an awareness of the actual and potential misuses of statistics and its techniques.”


Statistical analysis of science fiction authors and fans

The classic science-fiction related excerpt that follows after the jump is neither up to-date nor analytically robust. I tidied it a bit, but to do a decent job would require re-running the data, not to mention collecting data with a more recent vintage. But it is entertaining, and the concept may be of use to others. To whom? Well, I have spent a fair amount of time on Stack Exchange sites recently. Let me tell you all about it.

What is Stack Exchange?

Question and answer websites are popular. Stack Exchange is a free, mostly user-run Q&A site. It was co-founded and managed by Jeff Atwood a.k.a. @Coding Horror and Joel Spolsky. EDIT: Joel now runs Stack Exchange, as The Coding Horror has departed.

The prototype version of the site was known as Stack Overflow, and continues to thrive. There are many stacks on Stack Exchange. Most are computing or analytically-themed e.g. programming, systems administration, website design, mobile applications development, mathematics and quantitative finance. Others are more eclectic, and thus of a more experimental nature. They are labelled as such, by a beta designation, and guided along by the whimsically named Area51 Stack Exchange site. Now that you’ve been enlightened by that tangential aside, I’ll get to the point. I was thinking of Literature Stack Exchange in particular.

The problem at hand

Literature Stack Exchange was initially overrun by book-recommendation inquiries. This was unfortunate. Why? Because suggestions about subjective matters are nearly impossible to provide to friends and relatives, let alone on an online forum. Fortunately, the issue has resolved itself for the time being, through better site administration.

Update – The issue has resolved itself permanently, because the site was closed due to a general lack of interest in early May of this year. Stack Exchange does have a thriving Science Fiction community, which enjoys a great deal of activity! So let us continue, along the same, still relevant theme.

Perhaps the following approach might provide inspiration for those seeking reading material recommendations.

Classic science fiction writers and reader politics

Politics is the horizontal dimension, with the right-wingers at the right and the left-wingers at the left. Hard-science science fiction is the vertical dimension, with hard-science authors at the top and anti-hard-science authors at the bottom. While hard-science tended to be somewhat Righty, New Wave was strongly Lefty, having a correlation of -0.51 with politics. Not surprisingly, there is a correlation of -0.25 between hard-science and New Wave.

We learned all kinds of odd facts about fans and the things that inspired them to like different authors and styles.

  • Student fans like both Ellison and Heinlein more than  average, and like Vance and McCaffrey less.
  • Female fans are more likely than men to prefer McCaffrey and sword-and-sorcery fiction.

SourceNew Maps of Science Fiction by William Sims Bainbridge and Murray M. Dalziel, first published in Analog Yearbook 1977, pp 277-299.

To summarize, the chart captures the political leanings of sci-fi fans circa 1977, not the authors. H.P. Lovecraft is a good example, see the lower left quadrant of the chart. Lovecraft fans tend to be liberal sorts, supportive of all manner of progressive liberal ideology. H.P. Lovecraft has lain dreaming in R’yelah (or on Pluto, or in New Englander heaven) since 1937. If he were with us today, he’d probably support the Tea Party.

I find the heavy use of negative correlations rather confusing. (There are ways of remedying that though.)


Statistics Themed Everything

Would you ever leave on a road trip without a map or GPS? It is foolhardy and unwise.

Dr. L. Leemis published the paper Univariate Distribution Relationships in The American Statistician available here:

In the paper is the most wonderful chart any statistician has ever seen.

With Dr. Leemis’s permission we are able to bring you a full sized poster of the Univariate Distribution Relationships so that even though you may wander amongst the distributions, you are never lost.via

All original work by NausicaaDistribution viaEtsy

My-my-my-my data hits me so hard
    makes me say oh my Lord
Thank you for blessing me
    with SAS to use and proc MC MC
It feels good
    when you know you're done
A consistent sample from a target distribution
    And its known as such
And this is a property u can't touch

I told you homeboy
    u can’t touch thisYeah that's how we're samplin' and you know
    u can't touch this

Look at my steps man

    u can't touch this

Yo, let me reach the distribution
                                               u can’t touch this


Correlation Extravaganza

From the Stats With Cats blog, here is the chart accompanying the post Secrets of Good Correlations which offers the most comprehensive collection of correlation coefficient possibilities that I have ever encountered!


Porcine Joy

Spring showers bring pig flowers

Well, maybe not, but I haven’t featured any pigs recently. That could be in violation of the administrivial oink’s website charter.

The Ides of March are past, and we head into spring. We also approach the one-year anniversary of this blog on March 21 or thereabouts, right in time for the vernal equinox. I would like to take this opportunity to thank my loyal subscribers, all three of you, and my other readers, whether frequent or occasional. Never hesitate to leave comments, especially if it isn’t spam!

Statistics, probability and applied quantitative methods reading recommendations

Finally, I wish acknowledge subject matter experts with nice blogs in my fields of professional interest, including applied probability, (mostly) frequentist methods, and due diligence for purposes of financial and security-focused anomaly detection.

Stats with Cats I don’t like cats, but you can just ignore the photos. This is an accessible, frequently updated blog covering descriptive and inferential statistical methods, mostly explained through charming examples

Data Genetics This blog has excellent graphics (without gratuitous interactive data visualization!) accompanying posts demonstrating statistical, probability and mathematical methods for engineering as applied to a wide range of real world concerns e.g. using Benford’s Law to detect accounting fraud, Hamming Codes for error correction and solving combinatorics problems to demonstrate the poor odds for winning dice and card games.

Error Statistics Philosophy Error statistics quantifies how frequently and reliably different statistical models can be used to detect errors.Error probability statistics uses frequentist error probabilities, not frequentist probability. Frequentist error probability is the relative frequency of errors within a statistical model. Frequentist probability is merely the use of relative frequency of occurrence to infer probability of events. The introductory post, Frequentists in exile acknowledges the long-held perception that only Bayesian methods have respectable statistical foundations. Error Statistics Philosophy focuses on the defensible use of frequentist methods for probability and statistical models, especially in circumstances of limited information and high error avoidance requirements.