Survival Of The Slowest? Study Suggests That Less Intelligent Soldiers More Likely To Survive In War

220px-CaumontadvanceAmong the other costs of war, there may be a type of counter Darwinistic effect on a population according to a new study. A new British study has found that the most intelligent soldiers in World War II had a higher mortality rate in combat. In other words, the war favors the least intelligent soldiers in terms of survival.

The study by Ian Deary, a psychologist at the University of Edinburgh, and his colleagues used IQ scores for Scots who died and who survived the war. The study found that 491 Scots who died had an average IQ score of 100.8. Several thousand survivors who had taken the same test averaged 97.4.

There may be various reasons for this difference. One of the most obvious might be that more intelligent soldiers tend to be placed in greater leadership roles and are thus exposed to greater combat threats. The study also refuted a prior theory that less intelligent men were simply more likely to be rejected by the military. In reality, men who didn’t serve were more intelligent than surviving veterans.

Here is a link to the study: Scottish Study

26 thoughts on “Survival Of The Slowest? Study Suggests That Less Intelligent Soldiers More Likely To Survive In War”

  1. “Men who didn’t serve were more intelligent than surviving veterans”……so basically the middle class? The dirt poor and the elites went to war. The officers died. The vets who survived were dummer than the ppl at home who kept the country alive. Compare to usa today….the best and brightest are recruited and volunteer…somehow only ten percent qualify….but the dod takes fifty percent of its force from the fifteen percent rural america. As it culls the one percent for dod. Are they trying to off rural america? On purpose? Because urbanites are ignorant on property and gun rights? Let them have a higher iq….if they cant survive in the end who cares.

  2. @bigfatmike

    I like your reply and that concept is statistically accurate however there are considerable assumptions based on that reasoning when compared to the article being referenced.

    I remember the principle of Occam’s razor: “Among competing hypotheses, the one with the fewest assumptions should be selected.”

  3. “Sure you are, that’s the entire article comparing 2 IQ scores for a significance. The IQ scale itself states what is considered significant called standard deviations. Read the article!”

    That is true for the population. It is not generally true for samples drawn from the population – not unless the sample is drawn in an unbiased way.

    The question we are trying to answer is whether battle selects for death randomly, or are those selected for death different from the population of those who survive. Whether battle is unbiased can only be answered by analysis of the data of the sample – not by reference to the general population.

    It is really important to distinguish characteristics of the general population, on the one hand, and characteristics of a sample drawn from the population. The characteristics of the sample depend on the selection process! Not just the characteristics of the population.

    Consider a simple hypothetical. We know the general population has a mean of roughly 100 and a standard deviation (sd) of roughly 15. So any difference of less than 15 is not likely to be significant when we select in an unbiased way from the general population .

    Suppose we select two groups. If the IQ is 101 to 105 we put them in group 1. And if IQ is 95 to 99 we put them in group 2. Is there a real difference in the means of the two groups? Of course there is! We selected the two groups that way. It matters not at all whether the means of the two groups fall within the sd of the general population. We are not talking about the general population. We are talking about the two groups. The difference in the means of the two groups is a real difference (we choose it that way) and we can use means and standard deviations of the two groups to distinguish the two groups.

    The question before us is whether battle selects randomly for death or does battle select with a bias. Does battle select for death in a systematic way that we can distinguish with mean and standard deviation? That question has nothing at all to do with the general population. We have to test the groups, the samples – in this case those who died and those who lived.

    Intuitively the question is: if we kept drawing samples, would the means of the samples for those killed keep coming out close to 100.8 and rarely close to 97.4? Would the means of the samples for those who survived keep coming out close to 97.4 and rarely close to 100.8? If so, we might have reasonable belief the two groups are different.

    On the other hand, if we found many samples where the means of the two groups overlapped, then we might have reasonable belief that the difference of the two groups is due to chance variation – and there is no real difference in the groups.

    Roughly, we want to test the hypothesis:

    H0: the means of the two groups are the same: M1 – M2 = 0,
    Ha: the means are different: M1 – M2 != 0.
    where M1 and M2 are the means of the two groups.

    That test requires that we measure variability around M1 and variability around M2. Note: that we need to know the variability around both means – not just one. You are correct that standard deviation is one measure of variability. But the variability we have to measure is around both the mean for group 1, and variability around the mean for group 2. The standard deviation of the general population is not relevant for this test.

    Most any standard introductory statistical text book will have a section on comparing differences of means of two groups. In addition, the comparison of means of two groups can be generalized so that we can compare the means of many groups.

    For example we might want to compare those killed and those who survived in different theaters of the war. We might compare those killed and those who survived in Europe, Africa, and Asia – 6 different groups. Once again the characteristics of the general population (mean and standard deviation) is not relevant to this question. For our hypothetical comparing theaters of war, we need to know and compare the mean and standard deviation for each of the 6 groups. One technique to compare the means of many groups is analysis of variance – ANOVA.

    You can download OpenIntro Statistics 3rd at

    Chapter 5 is has a pretty good discussion of testing the difference of two means, and testing many means with ANOVA.

    Happy reading!

  4. Sorry Paul, I missed what you said…there is this pocket watch swinging back and forth as my wife picked my pocket….at least I think that’s what just happened. Excuse me as I go peck around for the seed on the garage floor.

  5. Uh Steve…so that leaves liberals as what? Hmmm…let’s see. Draw a line in the sand and run away? How’s that working out? Also…who is that “JV team”? BTW let’s tell the enemy what we won’t do…see Truman’s Doctrine and Korea and Johnson’s non-bombing of Haiphong and Clinton with the Serbian holocaust. Obama’s …no troops. Oops almost forgot Kennedy’s withdrawal of support for the Bay of Pigs…but didn’t bother to let the folks know that we reneged.

    1. Veterans Ministry:

      Truman was a conservative Democrat who may have killed one hell of a lot of people unnecessarily to take possession of Japan before Russia did.

      Johnson was a Texas crook who started a war with a lie and is a likely as anyone to have had knowledge of Kennedy’s demise (his mistress said as much) before Kennedy’s plane landed in Dallas. He was no liberal.

      William Clinton, like his wife, is a neo-liberal, i.e., a fascist lite. I say lite because they’re smart enough to let others do the dirty work while they climb the political ladder.

      Combined with Johnson’s jealously/hatred, Kennedy may very well have been killed because he fired Allen Dulles the fascist. Curtis LeMay may have been more to your liking? We’d be glowing right now had Kennedy taken his advice to use nukes or torpedo a Russian sub during the Cuban blockade. Kennedy swallowed his pride and skillfully negotiated the settlement of the Cuban Missile Crisis in the midst of conservative hotheads telling him to do otherwise.

      Best regards.

  6. We do realize that the IQ test of old did not actually measure the brain’s processing power right? It was a cultural and biased measurement focused on one method of learning-comprehension. I doubt that our Sgt York would have done well with the test. (Sgt York?? Just in case….he’s a MOH winner in WW-I)

    Now how about S. L. A. Marshal’s study of US POWs in Korea? The college educated had a much higher rate of capitulating or breaking down to “brainwashing” than the simple mid-Western farm boy. (BGen. Marshall served in WW-I, WW-II and Korea.)

    1. Renegade – there is some thought that the more intelligent you are the easier you are to hypnotize.

  7. “We are not evaluating whether an IQ of 100.8 is significantly different from an IQ of 97.4.”

    Sure you are, that’s the entire article comparing 2 IQ scores for a significance. The IQ scale itself states what is considered significant called standard deviations. Read the article!

  8. Old saying at Ft. Bragg, NC. “We dare you, to dare your self”. IQ is not the complete picture. Test a 100, but only 3 get the green beret.

  9. Here is what you do with “psychologists”: insert a big fat Freud cigar in the rear end and push the creature off a cliff. They are like Sgt. Schultz– they know nothing.

  10. “Thus there is no significance in evaluating an IQ score of 97.4 as compared to 100.8.”

    True but irrelevant. We are not evaluating whether an IQ of 100.8 is significantly different from an IQ of 97.4.

    We are asked if a group with a mean for 100.8 can be distinguished from a group with a mean of 97.4 – a very different question.

    The answer to that question depends on the deviations about the means of the two groups. Those two statistics are not given in the article (although we can guess that Corely et al had that data when they wrote the journal article). So from the information we have we just don’t know if the Scots who survived war are different from the Scots who were killed in action.

    As a thought experiment, it is easy to imagine sample deviations for the two groups that would allow us to conclude that the two groups really are distinct. And we can also imagine different values for the two deviations that would suggest that the differences in means occurred merely by chance.

    There really is no way to know if the two groups are different with out more information.

  11. You’ve no doubt heard of ‘dumb luck’?

    No doubt being privy to THIS information could be hazardous to your health. Just kidding. It’s tin foil hat conspiracy theory stuff only nincompoops believe in.

    SHOCKING EXCLUSIVE! Zullo CIA informant reveals how Gov’t could destroy YOU!

  12. Maybe the slow ones survived because all they could manage to remember was one thing: keep your head down.

  13. Give me a break!

    In reality, the 3.4 point difference from what’s considered “normal” is an IQ between 85 and 115 a standard deviation of +/-1. Thus there is no significance in evaluating an IQ score of 97.4 as compared to 100.8. However, moving outside the +/-1 SD such as +/-2 SD clearly implies a significant difference.

Comments are closed.