Just a reminder, Andrew Gelman is now blogging at ScienceBlogs under "Applied Statistics".

Labels: Blogs

Andrew Gelman has started a new blog at ScienceBlogs, Applied Statistics. Someone should design him a header, perhaps a fancified Bayes' theorem?

Labels: Blogs

Earlier this year I put a reader survey. One thing that stuck out was now few children readers of this weblog had. Here's a comparison with other demographics from the GSS:

Mean Number of Children For Men By For Age Classes
			GSS, 2000-2008, Males Only
Age	GNXP N	GNXP	All Males	Bachelor's Degree	Advanced Degree	Atheists & Agnostics
18-25	84	0	0.24	0.09	(N too small, omitted)	0.28
26-35	140	0.21	1.06	0.53	0.68	0.8
36-45	97	0.9	1.79	1.73	1.56	1.29
46-65	108	1.57	2.22	1.93	2.06	1.82

Below the fold I've broken down by demographic for the GNXP male sample 46-65, which has an N of 108 total (though some questions were omitted for some individuals). I added the political extremes and centers together (e.g., Far Left + Center Left + Left = Left)

GNXP Males 46-65		# Of Children
	N	0	1	2	3	4	5	6	Mean # Of Children
Middle Class	43	0.3	0.23	0.3	0.09	0.05	0.02	0	1.4
Upper Middle Class	40	0.25	0.15	0.3	0.18	0.08	0.03	0.03	1.85
Graduate Degree	59	0.27	0.12	0.37	0.15	0.05	0.02	0.02	1.73
Undergraduate Degree	33	0.33	0.3	0.18	0.12	0.03	0.03	0	1.29
Left	38	-	-	-	-	-	-	-	1.76
Libertarian	32	-	-	-	-	-	-	-	1.74
Right	24	-	-	-	-	-	-	-	1.32
Christian	25	0.2	0.04	0.44	0.2	0.08	0.04	0	1.44
No Religion	72	0.35	0.24	0.25	0.1	0.06	0.01	0	1.33

Struck by the fact that GNXP male readers who are upper middle class have more children than those who are middle class (subjective definition of course), and, that those with graduate degrees have more children than those with only undergraduate degrees.

Labels: Blogs

The Worst NYT Trend Story of the Year?:

Here's an early autumnal contender: Virginia Heffernan's entirely anecdotal story about a massive Facebook Exodus. How serious is this Facebook exodus? Heffernan explains:

The exodus is not evident from the site's overall numbers.

The exodus is not evident from the site's overall numbers! Some trend!

This is a serious problem. Anecdotes add spice to real data. As illustrations of something real. Too often journalism involves finding five people on the street who can agree with whatever "trend" you've made up. This reminds of how I was taught to write essays in Middle School, make up a thesis and find n facts to support the thesis. Who cares if those facts are representative of the distribution of facts in the real world! You got your thesis and you know what you are looking for. And secondarily, there is the problem of trends are so widely accepted as to become background assumptions, but which turn out to be false upon even cursory examination.

Labels: Blogs, Media