It’s time for a little experiment.

Tomorrow I will simply ask people to choose one of three colours – RED, BLUE or GREEN.  Today I would like everyone to try to predict exactly what percentage of people will choose BLUE.

Please enter your guess as a comment (to 2 decimal places), and I will send a signed copy of one of my books to the person who is closest.


  1. I know what’s happening here: you’re actually conducting an experiment to find out the relative percentages of percentages people will blurt out when asked about what percentage of people will opt for the colour blue in a hypothetical experiement (which fact you don’t reveal to us). HA! Did I catch you there?


  2. 20.90%

    Reason being, I think the fact that you mentioned the color blue today, more people are going to shy away from it. I’d guess Red would be at like 50-60% as well.

  3. Normally, I’d say around half, as few people will vote green I think (as it’s not a primary colour). But now that you’ve made it a contest, that influences the equation — either people will vote for it more, or they’ll shy away from it (to influence their own chances of winning). There seem to be lots of guesses hovering around 20%, although those around 70% might compensate for that a bit by upvoting.

    So I’m going to go with something around 30… say, 32.84%.

    1. Green is a primary color, when dealing with light. Check the colors of the pixels on your monitor with a strong magnifying glass.
      Primary colors of light: red, green, blue. (monitor pixels)
      Primary colors of pigments: cyan, yellow, magenta. (laser toner colors)

  4. I predict 72.56 but won’t the outcome be heavily influenced by the fact that you have forewarned everyone about the experiment? Just trying to be scientific….

    1. That’s the point. How does the forewarning influence the outcome? I think it’s a great experiment. Of course, we don’t have a control group (unless Richard’s collecting that data elsewhere.)

  5. It all depends on whether the people asked to choose know about this experiment or not. If they do, then I guess 20% will go for blue, otherwise 33%.

  6. Although indeed a lot of people might simply choose BLUE now to skew the result … wouldn’t it have been better to first let people vote (without us being able to see the results yet) and then only guess?

  7. I think it depends on the order you’re saying it in. So my suggestion is O*33% (O = order which can either be 1, 2 or 3) which would result in 2*33%=66% for the order you’ve mentioned them in. So I’d say [66%]!

    Also it might depend on what colors you’re wearing and such things but that’s a different story. 😉

  8. I’m sceptical. Is this a trick question? Are we going to be asked to vote tomorrow or was that a fib?! I think I heard somewhere that blue is most people’s favourite colour, so one would expect the vote to be high. But the fact that we’ve been told in advance, I think will make people shy away from blue. So I’d say 22.4%. I’m sure the fact we’ve been told about it in advance will affect the outcome.

  9. I think by doing this first, you’ve turned it into a choice between “choosing blue” and “not choosing blue”. So my guess would be for it to be around the 50% mark.

  10. There are many influencing variables which could skew the results. Are people going to be asked face to face, or over the internet? If face to face, what colour shirt is the tester going to be wearing? What colour tie? What colour paper? Are participants who are wearing only certain-coloured tops going to be asked? Maybe the questions are going to be asked of those who are colour-blind, therefore when they say ‘green’ do they really mean ‘blue’? Other variables may include time of day, colour of light or room, hot or cold; but there are many others.


  11. Actually, I initially assumed that like me this would make people prefer to chose not-blue, however after a little thought I now suppose that because more people are picking higher figures it gives them more incentive to pick blue tomorrow, so I’ll change to:


  12. People guessing for more than 33.3% will select BLUE to support their own guessing. I counted that 87 of 126 did this. So my guess is 69.047619%.

  13. It would be interesting to see what the percentage was if we hadn’t been told about the question before-hand! I have a feeling the fact that blue is a middle option a high percentage might choose it, even though technically they should have 33.3% each!

    Sorry for waffling, I guess 66.6%.

  14. This prediction itself might affect how many people will choose blue.
    I saw a number 37.5 on my mind, so I’ll pick that one. 🙂

  15. 48.01%

    Little more than a random guess as I believe Blue is a colour more people would go for. Will be supprised if it is less than 35%.

    awaiting the results.

    1. Actually is this experiment flawed, will the people being asked not be aware of this experiment and try and influence the outcome with their answer?

  16. 67.5%

    Lots of folks are guessing over 33%, which means in order to make that happen they’d have to pick blue. Now I could find the amount mathematicially, take the people who are guessing over 33% and make that my percentage or I can just find a number in the vicinity of my guess that isn’t taken and take that.

    by the way, use search people. There are lots of duplicates.

  17. I went a very low estimate because I know a lot of people who prefer green or red to blue, and because I remember a similar experiment. In that one people were asked to name their favourite animal and predict what other people would say. People thought people would pick cats, but they didn’t really. Or something along those lines.

    But I wonder if all those people who guessed high numbers will come back tomorrow and say blue to increase their chances of getting a book? Hm.

    It’s a shame that you can’t also run the two blog posts in the other order (hiding the comments on the first one, obviously) and compare the results.

  18. If myself and Dave are both right (we picked the same number), do I get the book because I said it first, or do we both get one?

    What if two people are equally close having picked different numbers (I realise how unlikely that is, but still)?

  19. i think 46.1%

    It’s very difficult to predict this though, because the people who predict high will vote blue, and those who predict low will vote low, and then there’s 3 choices, and red is more appealing.

    With all this info it is still a very wild guess.

  20. it’s a question of how people answer today, and what they’ll do tomorrow when choosing a color, would they on purpose choose blue to increase the odds or one of the other to decrease.
    it’s way to complicated so I go with 46.3%

  21. Ok let’s do this a bit more scientifically. All we really know is that everyone here is going to try to win by voting tomorrow. So I’ve worked out the average of all the guesses above…

    First a bit of jQuery:
    jQuery(‘.comment p’).map(function(){return (jQuery(this).text())})

    Then some Python:
    percent_regex = ‘(\d{1,2}(?:[.,]\d{1,5})?)%?’
    average([percent for comment in comments for percent in re.findall(percent_regex, comment)])

    Final answer:

  22. It’s hilarious to read the comments from a few people complaining that the result of the experiment will be contaminated or affected by the fact that we have been told in advance what’s going to be asked. That’s the whole point of the experiment, obviously.


  23. If you ask people on this blog: 37.28%.
    If you ask other people (e.g. a group of students): 48.22%.

    If you will do the second poll among today’s participants, then, unlike tomviner, I don’t think that you should calculate the average of the responses. Following the idea of CH23, you should rather calculate the percentage of people of who have guessed a number above 50% (assuming that only they will pick blue tomorrow). Strangely enough, this is about 1/3 in the 100 first votes…

    Richard, I hope that you have software to delve through all the responses, but it seems unlikely that you can automate it since some people have mentioned other percentages than their final guess.

    Also, it would be nice if you are actually testing something else, like which last digit is used most.

  24. 7.25% choose blue – because its presented between the two other colours and because of the strong effect of red 😉

  25. Wisdom of crowds 🙂
    I have averaged all the comments above and got 43.02%
    However Everyone will have an agenda, since everyone will vote blue for their own percentage if high, and wouldn’t vote blue if low, I would expect a stronger skew of the results than just that of the averaging.
    Also there will be a proportion of people who will vote tomorrow having not guessed a percentage today. Of these I would expect more to choose blue than the other colours* so this will increase the percentage that choose blue even further.
    My guess will be

    * (though as other have said the order and format the colours are presented will affect this)

  26. My guess is 44.17%
    So far mean= 43.38%, median = 42%.
    But my guess is based on the value x% above which x% of the guesses lie. Those who guessed above X are have an incentive to pick blue, those below it have an incentive to not pick blue. If everyone picks the way their incentives tell them, we should arrive at the value x as the answer. x=43.31%
    Unfortunately, somebody already guessed this. and the I want my guess to slide into a relatively large gap between other guesses. The closest decently sized gap was at 44.17%
    Perhaps I overthink things…

  27. There’s not going to be a winner. You have already skewed the results by asking someone to predict a specific color, and therefore have caused people to either pick that color more in order to raise the percentage (if they think it’s a high percentage) or choose another color in order to push the percentage lower (if they think it’s going to be a low percentage) There is no point in placing a guess.

  28. 43.79

    the whole point of the Wisdom of crowds is that the extremes cancel each other out. No reason to think the same won’t be true tomorrow. Those who picked high will vote blue, those who picked low won’t and those who picked in the middle might or might not pick blue. Really interested to see how this one turns out!!

  29. .63
    I predict that 63% of the respondents will choose blue. I am guessing that the majority of the readers of this blog are male and that blue is the preference of males.

  30. %62.00
    I think I should of waited until the end of the night to make my prediction so I could averge up all the guesses and then I would have a fairly accurate prediction, especially if they’re the same people answering the question tomorrow.

  31. Hmmm… it’s fun actually. Originally, I would’ve expected blue to get a little above average – say 40%, leaving Green for the second place with 35% and Red with humbling 25%. But that’s just pure guess work.

    However, at this point this survey is conducted. Now the people who will choose between those colors are divided – those who posted a comment (thus have a vested interest to make their prediction correct) and those who haven’t made a comment and probably haven’t seen this post at all (no interest whatsoever).
    For the first group the choice now is not between three colors, but actually between two – blue and whichever other one. At this point I have two things to say.
    1. I do not know how much people visit this site to compare first group to the second. It is possible to look through comments and make some fancy math, yet it would be entirely pointless.
    2. I would love to see correlation for the first group between the person’s prediction and actual vote.

    All things considered, I’d say that this ends up being way to complex. My guess is 33.33%, apply standard deviation if necessary.

  32. If enough people enter a pact they could all be closest, unless a dissenter guesses the size of the pact better than the pact-makers. Though that better guess would have to fit in no more than two decimal places. A difference in the third place wouldn’t matter.

  33. It’ll be 49 percent, trust me. Everybody loves blue. I voted for green but I like all of them. Who doesn’t like colors? I feel sorry for the color-blind. And by the way, who spells “color” with a “u”–oh, that’s right you people over the pond.

  34. Pingback: money

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s