There are now over 400 responses to the survey. Here is a link to the responses in CSV format. If you import this into R, an extra parameter in regards to encoding may be necessary:
responses=read.csv("responses.csv",sep="t",header=TRUE,fileEncoding = "UCS-2LE")
I decided to separate the respondents into two categories, biologists and non-scientists (therefore, excluding other types of scientists from further analysis). You can see the filtered responses for biologists and non-scientists yourself. Below are some comments on interesting differences.
I can’t send you to a direct link, so here are the results. Some of the results need to be crunched by me unfortunately. But here’s a sample (yes, I didn’t do much more than the basic R density plot).
It has been brought to my attention that Discover Magazine has a reader survey as well. Here’s the first page:
All responses to this survey will remain completely confidential and will be studied and interpreted only in combination with all other responses received.
As a thank you for participating, you will have an opportunity to enter into a drawing to win one of ten $50 Amazon Gift cards.
During the survey, please answer each question and avoid hitting your Web browser’s Back button.
For a copy of the sweepstakes rules, click on the link rules page.
It took me <2 minutes to complete.
We’re well north of 500 for the reader survey. Thanks to everyone who participated. I’ll let it run until Saturday morning and then close it. I figure if you aren’t reading the blog at least once a week right now you’re not a core reader anyway (yes, I know that people get busy!). But I did promise to release the raw results at some point. I’m doing so now. It’s in Google Docs.
We’ve moved north of 400 responses on the reader survey. I think the goal of an N of 500 is totally viable. In the past I’ve actually pushed it well north of 600 by leaving the survey open for a while. I know there are some people who drop in once a week or so, or don’t have time or inclination to participate initially. If you want to participate: just click here! It will take ~10 minutes, and no answer is mandatory.
There are already a few robust findings though. GNXP readers are well educated and smart. About 60-70 percent aver that they are irreligious, and 85 percent reject the existence of the supernatural. Over half have backgrounds in the natural sciences. None of this is too surprising. I’ve been taking surveys of the readership since 2004, and the main change has been in politics. Whereas in the mid-2000s libertarians were the largest contingent, now Left-liberals are, though there remains a sizable libertarian minority.
One of the most consistent findings on this weblog has been the sex ratio: the proportion of female core readership is on the order of ~15%. Since moving to Discover it looks like that ~20% is the new set point. This shouldn’t be too surprising…there are very few explicitly female handles in the comments. Though because of the male bias in the readership there’s obviously going to be a natural tendency toward assigning implicit male identity to anonymous or gender ambiguous commenters when a substantial number will be female. Currently the most prominent female in the comments is Michelle, who is a prominent science blogger in her own right (pictured above). The very fact that I could type the previous sentence is a commentary on the sex ratio imbalance!
A major reason I want to go north of 500 responses is that I can compare across two classes more easily. The smaller a sample size the greater the error. I’d be a lot more confident comparing those who believe in God vs. those who don’t if I had more respondents who actually believed. But now that I’m at nearly 100 female respondents I thought it would be interesting to compare across the two sexes in terms of similarities and differences.
First, let’s compare the cross-tabs of sex by other variables in a table. You see below the percentage of males and females who fall into a particular class. So below you can see that 73 percent of males support abortion on demand vs. 82 percent of females. While 36 percent of male respondents have made a non-trivial edit to Wikipedia, only 12 percent of female respondents have.
My sample size for the reader survey is now ~200. I’m aiming for ~500. If you are a regular reader of this weblog, please consider filling out the survey. The software is telling me that the average reader is taking about ~10 minutes. All questions are optional, so you can quickly skip over confusing ones or those which you don’t want to divulge.
The results so far are here. At least for the questions which weren’t open ended. I added a lot of open ended numeric questions so that I could run some scatterplots and more natural statistics (i.e., I don’t have to convert categorical responses into numerics and so forth).
To give a taste for the kind of stuff I’m running on the nerd-heavy data set I thought I would explore how # of sexual partners relates to age and IQ. First, let me admit that I assume that the IQ distribution of the readership here is somewhat artificially shifted upward (to the right of the distribution). Those with higher IQs are more likely to know their IQs. And whether unconsciously or consciously individuals will almost certainly self-report results which are drawn from the higher range of their results distribution. Additionally, since the readers are ~85% male there’s an expectation that there’ll be a slight padding of the # of sexual partners. But since I’m really curious about correlations I’ll assume that these biases are independent across the two result sets.
Below are some simple summary statistics, density distribution plots and scatterplots. Also some information on the virgins for those who are curious.
Since the reader survey is topping out in response, I though I’d report some of the results. Since I’ve been doing these surveys my readership has exhibited a few patterns, and I was curious as to any changes since moving to Discover. Not too much has shifted. Instead of 15% female, as was the case for years, the readers are now 25% female. It looks like ~10% of the readers know this website only through Discover. Feel free to browse the results yourself.
I think the most interesting aspect for many is the political diversity. Generally the readership is split between Left liberals and libertarians. Though there are a small number of conventional conservatives, it is very rare to find those who are socially conservative and fiscally liberal. These “populists” tend not to be as intelligent as the other combinations, and so I suspect that’s why they’re not well represented on the web, among my readership, or the political elite of the United States in general (for what it’s worth, I’ve been moving in a more populist direction over the years, starting from a libertarian stance).
First, a few summary statistics. I asked readers their index of liberalism, with 0 being as conservative as possible, 10 as liberal, and 5 in the middle. I asked on two dimensions, social and economic.
About six months ago I did a survey of the readership of my two Gene Expression blogs (before moving to Discover). The N was around 600. You can view the raw frequency results here. One of the issues which I was curious about: did the disciplinary background of readers have any major correlates with responses? So I created three categories from the data on disciplines:
Social science had its own section, but for science I amalgamated those who studies Math, Engineering, Natural Science and Medicine. The balance were under “Not science.”