Survey Results

Scott Alexander12 May 2009 22:09 UTC
62 points
212 comments7 min readLW linkArchive

Followup to: Excuse Me, Would You Like to Take a Survey?, Return of the Survey

Thank you to everyone who took the Less Wrong survey. I’ve calculated some results out on SPSS, and I’ve uploaded the data for anyone who wants it. I removed twelve people who wanted to remain private, removed a few people’s karma upon request, and re-sorted the results so you can’t figure out that the first person on the spreadsheet was the first person to post “I took it” on the comments thread and so on. Warning: you will probably not get exactly the same results as me, because a lot of people gave poor, barely comprehensible write in answers, which I tried to round off to the nearest bin.

Download the spreadsheet (right now it’s in .xls format)

I am not a statistician, although I occasionally have to use statistics for various things, and I will gladly accept corrections for anything I’ve done wrong. Any Bayesian purists may wish to avert their eyes, as the whole analysis is frequentist. What can I say? I get SPSS software and training free and I don’t like rejecting free stuff. The write-up below is missing answers to a few questions that I couldn’t figure out how to analyze properly; anyone who cares about them enough can look at the raw data and try it themselves. Results under the cut.

Out of 166 respondees:

160 (96.4%) were male, 5 (3%) were female, and one chose not to reveal their gender.

The mean age was 27.16, the median was 25, and the SD was 7.68. The youngest person was 16, and the oldest was 60. Quartiles were <22, 22-25, 25-30, and >30.

Of the 158 of us who disclosed our race, 148 were white (93.6%), 6 were Asian, 1 was Black, 2 were Hispanic, and one cast a write-in vote for Middle Eastern. Judging by the number who put “Hinduism” as their family religion, most of those Asians seem to be Indians.

Of the 165 of us who gave readable relationship information, 55 (33.3%) are single and looking, 40 (24.2%) are single but not looking, 40 (24.2%) are in a relationship, 29 (17.6%) are married, and 1 is divorced.

Only 138 gave readable political information (those of you who refused to identify with any party and instead sent me manifestos, thank you for enlightening me, but I was unfortunately unable to do statistics on them). We have 62 (45%) libertarians, 53 (38.4%) liberals, 17 (12.3%) socialists, 6 (4.3%) conservatives, and not one person willing to own up to being a commie.

Of the 164 people who gave readable religious information, 134 (81.7%) were atheists and not spiritual; 5 other atheists described themselves as “spiritual”. Counting deists and pantheists, we had 11 believers in a supreme being (6.7%), of whom 2 were deist/​pantheist, 2 were lukewarm theists, and 6 were committed theists. 14 of us (8.5%) were agnostic.

53 of us were raised in families of “about average religiousity” (31.9%). 24 (14.5%) were from extremely religious families, 45 (27.1%) from nonreligious families, and 9 (5.4%) from explicitly atheist families. 30 (18.1%) were from families less religious than average. The remainder wrote in some hard to categorize responses, like an atheist father and religious mother, or vice versa.

Of the 106 of us who listed our family’s religious background, 92 (87%) were Christian. Of the Christians, 29 (31.5% of Christians) described their backgrounds as Catholic, 30 (32.6% of Christians) described it as Protestant, and the rest gave various hard-to-classify denominations or simply described themselves as “Christian”. There were also 9 Jews, 3 Hindus, 1 Muslim, and one New Ager.

I didn’t run the “how much of Overcoming Bias have you read” question so well, and people ended up responding things like “Oh, most of it”, which are again hard to average. After interpreting things extremely liberally and unscientifically (“most” was estimated as 75%, “a bit” was estimated at 25%, et cetera) I got that the average LWer has read about half of OB, with a slight tendency to read more of Eliezer’s posts than Robin’s.

Average time in the OB/​LW community was 13.6 ± 9.2 months. Average time spent on the site per day was 30.7 ± 30.4 minutes.

IQs (warning: self-reported numbers for notoriously hard-to-measure statistic) ranged from 120 to 180. The mean was 145.88, median was 141.50, and SD was 14.02. Quartiles were <133, 133-141.5, 141.5-155, and >155.

77 people were willing to go out on a limb and guess whether their IQ would be above the median or not. The mean confidence level was 54.4, and the median confidence level was 55 - which shows a remarkable lack of self-promoting bias. The quartiles were <40, 40-55, 55-70, >70. There was a .453 correlation between this number and actual IQ. This number was significant at the <.001 level.

Probability of Many Worlds being more or less correct (given as mean, median, SD; all probabilities in percentage format): 55.65, 65, 32.9.

Probability of aliens in the observable Universe: 70.3, 90, 35.7.

Probability of aliens in our galaxy: 40.9, 35, 38.5. Notice the huge standard deviations here; the alien questions were remarkable both for the high number of people who put answers above 99.9, and the high number of people who put answers below 0.1. My guess: people who read about The Great Filter versus those who didn’t.

Probability of some revealed religion being true: 3.8, 0, 12.6.

Probability of some Creator God: 4.2, 0, 14.6.

Probability of something supernatural existing: 4.1, 0, 12.8.

Probability of an average person cryonically frozen today being successfully revived: 22.3, 10, 26.2.

Probability of anti-agathic drugs allowing the current generation to live beyond 1000: 29.2, 20, 30.8.

Probability that we live in a simulation: 16.9, 5, 23.7.

Probability of anthropic global warming: 69.4, 80, 27.8.

Probability that we make it to 2100 without a catastrophe killing >90% of us: 73.1, 80, 24.6.

When asked to determine a year in which the Singularity might take place, the mean guess was 9,899 AD, but this is only because one person insisted on putting 100,000 AD. The median might be a better measure in this case; it was mid-2067.

Thomas Edison patented the lightbulb in 1880. I’ve never before been a firm believer in the wisdom of crowds, but it really came through in this case. Even though this was clearly not an easy question and many people got really far-off answers, the mean was 1879.3 and the median was 1880. The standard deviation was 36.1. Person who put “2172”, you probably thought you were screwing up the results, but in fact you managed to counterbalance the other person who put “1700″, allowing the mean to revert back to within one year of the correct value :P

The average person was 26.77% sure they got within 5 years of the correct answer on the lightbulb question. 30% of people did get within 5 years. I’m not sure how much to trust the result, because several people put the exact correct year down and gave it 100% confidence. Either they were really paying attention in history class, or they checked Wikipedia. There was a high correlation between high levels of confidence on the question and actually getting the question right, significant at the <.001 level.

I ran some correlations between different things, but they’re nothing very interesting. I’m listing the ones that are significant at the <.05 level, but keep in mind that since I just tried correlating everything with everything else, there are a couple hundred correlations and it’s absolutely plausible that many things would achieve that significance level by pure chance.

How long you’ve been in the community obviously correlates very closely with how much of Robin and Eliezer’s posts you’ve read (and both correlate with each other).

People who have read more of Robin and Eliezer’s posts have higher karma. People who spend more time per day on Less Wrong have higher karma (with very strong significance, at the <.001 level.)

People who have been in the community a long time and read many of EY and RH’s posts are more likely to believe in Many Worlds and Cryonics, two unusual topics that were addressed particularly well on Overcoming Bias. That suggests if you’re a new person who doesn’t currently believe in those two ideas, and they’re important to you, you might want to go back and find the OB sequences about them (here’s Many Worlds, and here’s some cryonics). There were no similar effects on things like belief in God or belief in aliens.

Older people were less likely to spend a lot of time on the site, less likely to believe in Many Worlds, less likely to believe in global warming, and more likely to believe in aliens.

Everything in the God/​revealed religion/​supernatural cluster correlated pretty well with each other. Belief in cryonics correlated pretty well with belief in anti-agathics.

Here is an anomalous finding I didn’t expect: the higher a probability you assign to the truth of revealed religion, the less confident you are that your IQ is above average (even though no correlation between this religious belief and IQ was actually found). Significance is at the .025 level. I have two theories on this: first, that we’ve been telling religious people they’re stupid for so long that it’s finally starting to sink in :) Second, that most people here are not religious, and so the people who put a “high” probability for revealed religion may be assigning it 5% or 10%, not because they believe it but because they’re just underconfident people who maybe overadjust for their biases a little much. This same underconfidence leads them to underestimate the possibility that their IQ is above average.

The higher probability you assign to the existence of aliens in the universe, the more likely you are to think we’ll survive until 2100 (p=.002). There is no similar correlation for aliens in the galaxy. I credit the Great Filter article for this one too—if no other species exist, it could mean something killed them off.

And, uh, the higher probability you assign to the existence of aliens in the galaxy (but not in the universe) the more likely you are (at a .05 sig) to think global warming is man-made. I have no explanation for this one. Probably one of those coincidences.

Moving on—of the 102 people who cared about the ending to 3 Worlds Collide, 68 (66.6%) prefered to see the humans blow up Huygens, while 34 (33.3%) thought we’d be better off cooperating with the aliens and eating delicious babies.

Of the 114 people who had opinions about the Singularity, 85 (74.6%) go with Eliezer’s version, and 29 (25.4%) go with Robin’s.

If you’re playing another Less Wronger in the Prisoner’s Dilemma, you should know that of the 133 who provided valid information for this question, 96 (72.2%) would cooperate and 37 (27.8%) would defect. The numbers switch when one player becomes an evil paper-clip loving robot; out of 126 willing to play the “true” Prisoner’s Dilemma, only 42% cooperate and 58% defect.

Of the 124 of us willing to play the Counterfactual Mugging, 53 (42.7%) would give Omega the money, and 71 (57.3%) would laugh in his face.

Of the 146 of us who had an opinion on aid to Africa, 24 (16.4%) thought it was almost always a good thing, 42 (27.8%) thought it was almost always a bad thing, and 80 (54.8%) took a middle-of-the-road approach and said it could be good, but only in a few cases where it was done right.

Of 128 of us who wanted to talk about our moral theories, 94 (73.4%) were consequentialists, about evenly split between garden-variety or Eliezer-variety (many complained they didn’t know what Eliezer’s interpretation was, or what the generic interpretation was, or that all they knew was that they were consequentialists). 15 (9%) said with more or fewer disclaimers that they were basically deontologists, and 5 (3.9%) wrote-in virtue ethics, and objected to their beliefs being left out (sorry!). 14 people (10.9%) didn’t believe in morality.

Despite the seemingly overwhelming support for cryonics any time someone mentions it, only three of us are actually signed up! Of the 161 of us who admitted we weren’t, 11 (6.8%) just never thought about it, 99 (59.6%) are still considering it, and 51 (31.7%) have decided against it.

Scott Alexander12 May 2009 22:09 UTC
62 points
212 comments7 min readLW linkArchive