# What can astrology tell us about American politics?

I might have thought the answer to that question was “nothing,” but it seems I would have been wrong.

Tony Vallencourt has an interesting post on his econball blog. He tallies the astrological sign of members of the U.S. House of Representatives. This exercise is inspired by the work of Anders Ericsson and others; they find that month of birth has an important impact on later success in a variety of activities such as professional soccer and hockey. We recently wrote about this work in a New York Times column.

It turns out that the astrologoical signs of the politicians are quite skewed, with Gemini, Cancer, Libra, and Virgo over-represented.

Anyone have ideas why that might be?

(While you are at Tony’s webpage, check out some of his other interesting posts, including one on the Coke/Pepsi stolen secrets where he argues much the same point as I did here on this blog — but Tony posted first.)

## princesaud1

Each "astrological sign" domain month, day, clock time of person's "birth". Merely count nine months back for date of approximate acts of conception to begin gestation stages to "live" birth. It has been shown conceptions(intercourse) arises most often in Western Hemisphere during late fall month through winter months.

Therefore those signs in your article with have highest number of births. And higher potential numerical population to attempt to and eventally win an election to a "statutory-government office". For back group info go to http://www.ask.com type in by months number of persons born. Then do the analysis suggested. Domain of: by months/days intervals for each astrological sign. For each sign's intervals minus 9 months, equals probable date/month of conception acts(intercourse).

These should be late fall through winter months. If unclear email me.

## carltoncl

So if what princesaud1 says is true then the Australian, NZ, or South African parliaments should show a bias towards astrological signs that are six months before or after the ones that Mr. Levitt posted? Did anyone correct for immigrants from climates where there really is no fall/winter as we experience it, like Cuba or Central America or northern Africa?

## zbicyclist

Why are they futzing around with Astrological signs? Can nobody find time to do a real study? (well, I can't)

1. What are the school cutoff dates in the state? (these vary)

2. Was the legislator born in the state? (imperfect control for whether he/she went to grade school in the state and was subject to the cutoff date). Exclude those who were born elsewhere.

3. Correct for expected births (see note #1 above).

4. There should be enough variance in the school entry dates to separate out whether it's relative age at work or whether it's prenatal environment (e.g. spent summer in womb so used to environments with lots of hot air).

Could be a birth order effect, as well, IF subsequent births tend to have a different typical birth month than first births. That seems possible. (For example, if you carried your first child through an Alabama summer, would you want to have your second child at the end of winter instead?)

Birth order effects on IQ are well documented.

## kate q

Do women produce different hormones / amounts of hormones at different times of the year, and might this affect the development of the fetus?

Would a seasonal diet have any effect? Does a woman's exercise vary seasonally?

## Taed

There's also the possibility that the data is wrong. With something as odd-sounding as this, I wouldn't personally believe it unless I verified it for myself or read it in a more "reputable" source (such as the Freakonomics blog). One shouldn't believe everything one reads on the Internet!

And this isn't about trusting Tony Vallencourt, who I gather is trusted by Steven Levitt, but about trusting where he got his data. He seems to have gotten it from a Washington Post web site which includes the congress members' zodiac sign... I'm a bit skeptical... Hmmm, I wonder if there's a list of the IQs of recent Presidents?

For more similar-sounding information, see http://www.gullible.info/...

(Of course, it could all be true; I'm just reserving judgement rather than believing it outright...)

## dwillis

As one of the developers of the Post's congressional votes database, I'd like to assure Taed that the astrological sign data is accurate. The astrological signs are determined using the lawmaker's date of birth, which can be found using the Congressional Biographical Directory.

And no, we haven't done any studies about the relevance of lawmakers' astrological signs. We (Adrian Holovaty is the other developer) just wanted to have a feature that people might find interesting.

Derek Willis

## John S.

Table 15 in this publication from the Centers for Disease Contro provides the distribution of births by month in the US in 2000.

## John S.

Mystery solved. Using the data in the CDC report I link to above, I estimated the frequency of each zodiac sign in the population. Now, the CDC report gives birth frequency by month, and each zodiac signs straddles two months. So, to estimate the probability of each sign, I took a weighted average of the probabilities for the two months.

This is better explained with an example. Consider Libra, which goes from September 24 to October 23 (inclusive, I assume). That's 7 days in September and 23 days in October. The CDC data says that 8.56% of births are in September, and 8.47% are in October. So I estimate the probability of Libras as (7*8.56% + 23*8.47%)/(7+23) = 8.49%.

From these frequencies, I determined the expected number of representatives with each zodiac sign. With observed and expected, I did a chi-square test. Bottom line, the p-value is 0.2884. This means there is no evidence that the observed distribution of zodiac signs in the current U.S. Congress is different from that of the general population.

## John S.

My bad! The p-value is more like 0.004. I blame R for my mistake!

## byomtov

John S.,

What probability did you test? That's crucial here.

There is, I think, nothing dramatic about the distribution once you recognize that some signs will have more births than others.

You can't really ask, for example, "what is the probability that 12.7% will be Cancers?" You have to ask, "What is the probability that there will be a sign with 12.7%?'

Those are much different questions.

## John S.

I calculated the chi-square statistic as defined in equation (1) on this page. The n_i are the observed number of congressmen with each zodiac sign, p_i are the expected probabilities for each zodiac sign, and n = 432.

The value of chi-squared is 27.16, which is way out in the tail -- a very unlikely value.

By the way, here is the distribution of zodiac signs (p_i), calculated as I explained above:

January 8.13%

February 7.82%

March 8.39%

April 7.81%

May 8.41%

June 8.41%

July 8.60%

August 8.87%

September 8.56%

October 8.47%

November 8.22%

December 8.30%

## John S.

Sorry, I gave the distribution by birth month. The distribution by zodiac sign is:

Aquarius: 0.0793

Pisces: 0.0821

Aries: 0.0801

Taurus: 0.0821

Gemini: 0.0841

Cancer: 0.0854

Leo: 0.0879

Virgo: 0.0865

Libra: 0.0849

Scorpio: 0.0829

Sagittarius: 0.0828

Capricorn: 0.0819

## John S.

Okay, I've already posted far too much on this thread, but here is my final word. Here is a file containing the number of babies born on each day in the U.S. in 1978. If you use that file to determine the distribution of zodiac signs, you get slightly different frequencies than those in comment #12 above. When you run the chi-square test using the 1978 frequencies, you get a p-value of 0.018 for the distribution of zodiac signs in the current Congress.

That may seem small, but consider that this is the 109th Congress: even assuming random congressional birthdates, we could expect to see a distribution of zodiac signs like the current one at least once -- maybe twice -- in 109 draws (yes, I realize there is always some overlap of congressmen, and that has some effect).

The only thing that would settle this for me is to see the birthdates of all individuals who have ever served in the U.S. Congress. Wikipedia?

## Tony Vallencourt

I was thinking the same thing, John. But I don't have the energy to track that all down, since I couldn't find it easily in one source. Though I'd probably want to stick to Congressional members in the past 50 years or so, in order to have a sample for which school cut-off dates actually existed.

Hmmm, maybe it would be better to take all members of Congress born into a state-year cohort for which there is a school cut-off date. But now it's getting harder and harder.

Either way, I'm starting to believe that it probably is random chance....

## byomtov

John,

That's what I suspected, and I think it's the wrong calculation. To see what I mean note that the discussion is about the fact that Gemini, Cancer, Libra, and Virgo seem to be overrepresented.

But suppose the percentage breakdowns into signs were the same, but with different high-frequency signs. Then we could wonder why, say, Pisces, Aquarius, Leo, and Capricorn were overrepresented. Or some other set of four.

The relevant question is "What is the chance that some set of four signs will be overrepresented to this degree?"

## John S.

byomtov: The chi-square test does exactly what you think it should do. See the link I gave above to Planet Math. The formulas are there.

## PiledHighandDeep

Something to consider with this type of data, going forward: it is becoming a trend, and has been for several years in my school district, to keep your late-for-the-school-cutoff-birthday child back a year in school. Parents of boys, especially, who are born in May-August in our school district, are often kept back a year.

Many of the parents do it so their child will be more competitive in sports, but some have concerns about social development as well. This throws a bit of a wrench in the works as far as your data go, as you'd have to figure in actual graduation dates vs. age to determine if a person was held back as a child.

## byomtov

John,

Either I'm being unclear or I don't understand what you did.

I am suggesting that there are two different questions:

1. What is the probability that Gemini, Cancer, Libra, and Virgo are overrepresented to the degree described?

2. What is the probability that

there exists a set of four signsthat are overrepresented as described.My understanding is that you addressed #1, but I think you should address #2.

This is the same distinction as:

1. What is the chance that the top five cards in a deck will all be spades?

2. What is the chance that the top five cards will all be of the same suit?

Clearly, the second probability is much higher.

Now, if you draw the top five cards, and find them to be all spades, you may remark that this was unlikely. But you would make a similar remark if they wereall hearts, diamonds, or clubs. So the second probability is, I think, the relevant one.

As I said, I may have misunderstood your calculation.

## John S.

Yes, you misunderstood my calculation, because I addressed neither of those questions. I did a chi-square test, pure and simple. I could explain in words what this test does, but I already posted a link to a good description of it. Not much more I can do if that doesn't clear it up for you.

## Valerie

Sorry I am a bit late for this discussion. I have been web trolling to find a place discussing astrology and US politics in order to post something rather intriguing.

In the mid 80s I did gradute work on the history of critical realignments in US presidential elections. The results in my data range showed three rather predicatable critical realignments in US history from 1828 to 1984(the range of my data): Lincoln, Roosevelt and Reagan. I immediately noticed that these are three Aquarian presidents since I am Aquarian. I thought it was kind of cute, Aquarius being the sign for dramatic change and all.

Recently I was looking around for data on changing patterns of electorial alignment and found an article in Wikipedia. This article lists six presidential realignments. The first two are Jefferson and Jackson (both Pisces). The additional four include the three that my analysis found as well as the election of William McKinley. Just for fun, I checked McKinley's birthdate and, sure enough, he was also Aquarius.

As a footnote, the only other Aquarius US president was William Henry Harrison who died after only 30 days in office, the shortest tenure of any US president.