Carnal Reason

4 Responses to “53% of all statistics are completely made up”

1. F. James Kearney, Ph.D., Esq., on November 24th, 2005 at 11:58 pm Said:

   Author “pwyll” offers absolutely no evidence or argument supporting the connection between his headline and the clearly agenda-driven Guardian op-ed piece that follows. As a scientist who has fought the “don’t lie with statistics” battle for 20 years, this sort of it-speaks-for-itself argument is not helpful in conveying the point.

   As to a broader point, unintentionally supported by Ms. Bindel’s presentation, I’m reminded “[i]f the political and administrative system has committed itself in advance to the correctness of its reforms, it cannot tolerate learning of failure. …. We must be able to advocate without that excess of commitment that blinds us to reality testing.” Campbell, D. 1969. Reforms as Experiments. American Psychologist 24: 409-29.

2. pwyll, on November 25th, 2005 at 11:45 am Said:

   “… this sort of it-speaks-for-itself argument is not helpful in conveying the point”

   Ouch. All right then, let’s get more specific. What are some of the flags that signal crap statistics?

   The author cites no sources for the statistic. I am perfectly happy to give the source of my statistic: I made it up for rhetorical reasons. I suspect Ms. Bindel did pretty much the same thing with her statistic.

   The author presents no error bounds or confidence levels. My own statistic has error bounds of +-50% at a confidence level of 100%. Providing these bounds allows the reader to draw the conclusion that my statistic is crap. Failure to provide them suggests the author is hiding something, or does not understand statistics enough to know to provide them.

   The statistic is sloppily stated.
   “Violence against women is the cause of more deaths and disability around the world in 15- to 44-year-olds than cancer, malaria, traffic accidents or war.”

   Cancer, malaria, traffic accidents, and war affect both men and women. Does Ms. Bindel assert that

   a: violence against women is the cause of more deaths and disability around the world in 15- to 44-year-olds than cancer, malaria, traffic accidents or war cause among all people age 15-44

   b: violence against women is the cause of more deaths and disability around the world in 15- to 44-year-olds than cancer, malaria, traffic accidents or war cause among women age 15-44

   Option a: is extremely unlikely, and option b: has a severe ambiguity. Insofar as war is conducted via direct attacks against women (e.g. the current use of gang rape in the Sudan) do the casualties go into the war column or the violence against women column? Both?

   And what is the significance of the age 15-44 group anyways? Are the people who collect the violence against women data simply throwing away data outside this range? Not bothering to collect it? Does no violence occur outside this range?

   Another signal of the crap statistic: the statistic is tremendously broad in scope, so much so that collecting it would require resources simply not available. Is every nation in the world collecting and providing statistics on violence against women age 15-44? Where is the data coming from? How is it gathered? Do men fill out surveys and admit yes, I murdered my wife?

   Another signal: Ms. Bindel’s statistic is time-bound, but no dates are provided. War casualty rates spike during wars (oddly enough) and decrease in times of peace. Does Ms. Binder maintain that her statistic holds even during the height of the world wars? At the height of malaria epidemics? If not, when does it hold?

   Is that specific enough?

3. F. James Kearney, Ph.D., Esq., on November 25th, 2005 at 6:29 pm Said:

   The comment by “pwyll” (11.25.05 @ 11:45 am) is extremely edifying, and wholly consistent with the original purpose of the Internet: exchange of information among scientists.

   Moreover, to a statistician, the line “… [m]y own statistic has error bounds of +-50% at a confidence level of 100%. ….” is hilarious.

   Buonissimo!

4. F. James Kearney, Ph.D., Esq., on November 25th, 2005 at 6:33 pm Said:

   And, bye the bye, Ms. Bindel should should stick to carrying signs written in crayon and sparkle that say Stop Violence Against Women.