Talk:Big Five personality traits

From Wikipedia, the free encyclopedia

WikiProject on Psychology
Portal		 This article is within the scope of WikiProject Psychology, which collaborates on Psychology and related subjects on Wikipedia. To participate, help improve this article or visit the project page for details on the project.
B This article has been rated as B-Class on the quality scale.
Mid This article has been rated as Mid-importance on the importance scale.

Article Grading: The article has been rated for quality and/or importance but has no comments yet. If appropriate, please review the article and then leave comments to identify the strengths and weaknesses of the article and what work it needs.

Contents

[edit] Old Discussion

Citation needed : A person's ratings on the five factors has been found to change with time, with Agreeableness and Conscientiousness increasing, while Extraversion, Neuroticism, and Openness generally decrease as a person ages.

I don't think its Costa et al. - so who is it? Cached 07:30, 7 October 2006 (UTC)


I am removing the following section for two reasons:

"Some anthropologists and sociologists are critical of the method because it measures personality traits on a basis of its synonymous adjectives in the English language, thus overlooking possible cultural construction of language. That English alone was used in factor analysis also arguably demonstrates an example of ethnocentrism. Furthermore, sociologists are critical that the theory only reinforces stereotypes and ignores social factors that might have lead to traits being present in certain groups of individuals. For instance, while women have a demonstrated tendency towards agreeableness, this is likely true of most subordinated individuals. Tying this trait to women at a psychological level then only reasserts inequalities by implying that differences exist as a matter of human nature rather than as a result of the social construction of gender."


1. The information given there is incorrect. I've checked the scientific research to be sure, and English alone was *not* used in factor analysis, and the existence and nature of the Big Five factors has been validated in multiple continents. Here's an excerpt from McCrae & Hofstede's fascinating paper, Personality and Culture Revisited: Linking Traits and Dimensions of Culture (2004):


"Research in the English-speaking world using the NEO-PI-R established that individual differences in the factors are stable throughout most of the adult life span; that self-reports generally agree with observer ratings; and that the five factors, as well as the more specific traits that define them, are strongly heritable. In the 1990s, researchers around the world began to develop translations 2 of the instrument, making cross-cultural research possible. A series of studies showed that much the same factor structure was found in a wide variety of cultures (McCrae & Costa, 1997), that developmental trends in the mean levels of personality traits between adolescence and later adulthood appeared to be universal (McCrae et al., 1999), and that similar gender differences were found among cultures (Costa, Terracciano, & McCrae, 2001)."


Here are the relevant citations from the above; if you can find them, I suggest taking a look:

McCrae, R. R., & Costa, P. T., Jr. (1997). Personality trait structure as a human universal. American Psychologist, 52, 509-516.

McCrae, R. R., Costa, P. T., Jr., Lima, M. P., Simões, A., Ostendorf, F., Angleitner, A., et al. (1999). Age differences in personality across the adult life span: Parallels in five cultures. Developmental Psychology, 35, 466-477.

Costa, P. T., Jr., Terracciano, A., &McCrae, R. R. (2001). Gender differences in personality traits across cultures: Robust and surprising findings. Journal of Personality and Social Psychology, 81, 322-331.


As a further note in this regard, one group of subordinated individuals - prisoners - show personality profiles with low Agreeableness scores. Teenagers also show lower Agreeableness than adults. So I don't think it's "likely true of most subordinated individuals" that they have high Agreeableness. In point of fact, it appears that Extroversion actually has more to do with social dominance than Agreeableness does. (And men score higher in the Assertiveness sub-scale of Extroversion, incidentally, as given in the last study cited above.)


2. Beyond being incorrect, I don't like the above statement because it demonstrates bias. The fact that psychometric research "reinforces stereotypes" does not show that the research is bad or wrong; instead, it supports the truth to these stereotypes. Whether women are Agreeable because of genetic or environmental reasons is irrelevant to the question of whether Agreeableness is a valid and universal psychomentric construct. And of course while sociologists and anthropologists may be unhappy about these research results, that doesn't change the facts.

Hopefully this explanation is thorough enough to obviate the necessity of future arguments.


Harkenbane 22:26, 25 Apr 2005 (UTC)

Clearly the point is biased in some direction; it is, after all, called criticism. However, it is appreciatable to rework claims rather than remove them entirely, as for example my move to redirect back the bias of your word "discovered", implying a factual truth of the matter, back to the more neutral term identified, and my return of your loaded attempt of beginning the criticism section with "While the general consensus favoring a five-factor model is strong, there is still dissention."
And how was the Big Five originally constructed if not through factor analysis of personality traits from the English dictionary? That's what I had read. If later attempts were made to see if other cultures and languages "fit" the pattern, that is not enough to convince me. That all the studies you listed came from the original proponents of this model also suggests to me some degree of bias in their perspective.
My use of the term "subordinated individuals" was admittedly very loose. I meant individuals of any subordinated master status. For instance, any minority groups, including women, and lower class individuals. I would imagine the scores of these groups would prove quite similar, but by this theory's limitation of analytical claims to biological groups there is the suggestion that this is because of some biopsychological component, and not because of some common social characteristic present in minorities and the less privileged in general.
As for your remark that it does represent stereotypes, I will not say it doesn't. Women are in all societies subordinate to men. But there is also sociological research to show this is because of a need for women in huntering and gathering societies to nurse, which means having to stay near the children, which eventually leads to a divide in terms of labor by sex, with men working exterior to the home and women within, which further precipitates the notion of women as domestic and men as worldly, and so on and on. That women are generally subordinate to men is without question, but to say that this is because of some localized psychology in women is to suggest that women are by nature different, and then any attempt at change would therefore be fallible.
That the critical addendum of the sciences of sociology and anthropology needs to be reworked is certain, but the attitude that outward criticism from other disciplines is somehow null is at the very least off-putting, and your remark that your explanation is "thorough enough to obviate the necessity of future arguments" seems to suggest a general dismay at the idea of said criticism. That you approach the subject as a psychologist is clear, but as other sciences also examine the structure of personality, disregarding the views of those sciences consitutes a clear POV. When I get a chance, I will post another criticism section, and if you would like you can revise it, and we can attempt to reach some consensus through mututal cooperation rather than through the throrough deletion of large chunks of text that you are at odds with. Sarge Baldy 23:59, Apr 25, 2005 (UTC)


> However, it is appreciatable to rework claims rather than remove them entirely,

I agree, and this is what I did to the section on traits "beyond" the FFM. Unfortunately, the entirety of the text I deleted was simply unusable, for the reasons detailed above.


> as for example my move to redirect back the bias of your > word "discovered", implying a factual truth of the matter, > back to the more neutral term identified, and my return of > your loaded attempt of beginning the criticism section with > "While the general consensus favoring a five-factor model > is strong, there is still dissention."

I didn't notice these changes, and I suppose I can agree with both.


> If later attempts were made to see if other cultures and languages "fit" the pattern, that is not enough to convince me.

Factor analysis in other languages found the same five constructs. (There were a few exceptions, of course; I recall that Openness didn't have the same character in a German sample, where it seemed more related to general "rebeliousness.") I'm sorry if this isn't enough to convince you, but studies were carried out specifically with the intention of seeing whether the Big Five traits existed as fundamental psychometric constructs or were merely an artifact of Western language and culture.


> For instance, any minority groups, including women, and lower class individuals. I would imagine the scores of these groups would prove quite similar

An interesting idea, to be sure! But there is simply no support for it; that doesn't make it false, but it renders it meaningless as an argument, because subordinate status could just as easily *depress* Agreeableness - in fact there is some evidence which is in line with this view; again, Agreeableness increases with age, and so does social and economic status.


> That women are generally subordinate to men is without question, but to say that this is because of some localized psychology in women is to suggest that women are by nature different, and then any attempt at change would therefore be fallible.

I don't think that Agreeableness has much to do with subordinate status; this is likely to have more to do with things like Extroversion, and with psychometric g (which lies outside the personality domain).


> That the critical addendum of the sciences of sociology and anthropology needs to be reworked is certain, but the attitude that outward criticism from other disciplines is somehow null is at the very least off-putting, and your remark that your explanation is "thorough enough to obviate the necessity of future arguments" seems to suggest a general dismay at the idea of said criticism.

This appears to be the central issue, so once more I'll try to be thorough in stating my position. Criticism from other disciplines is really of very limited importance in an encyclopedic article. After all, if we are to consult sociologists and anthropologists on psychometric matters, what about economists or biologists? What about politicians, philosophers, or theologians? These are not facetious questions - for instance, knowing that personality exists, and that it is to some degree heritable, and that it has important outcomes in religious behavior, raises interesting theological questions. We have been discussing Agreeableness, and it seems that those who are more Agreeable tend to be more religious, to be less prone to marital infidelity, and to have a more generous nature. Does God judge on a curve? And if not, is it religiously acceptible that virtue and sinfulness could be genetic, or should pious individuals of fundamentalist sects reject the findings of psychometrics as readily as they might the findings of physical anthropology? These are questions which personally interest me to a great degree, and perhaps they deserve to be discussed in a full length book dealing with these topics, but I really don't think they have a place in a brief encyclopedic article. The objectively relevant view here is the psychometric one.


> I will post another criticism section, and if you would like you can revise it, and we can attempt to reach some consensus through mututal cooperation rather than through the throrough deletion of large chunks of text that you are at odds with.

The trouble with this is that the FFM is approaching the level of "settled science." Criticisms of it are generally outdated and have been dealt with by research from dozens of psychologists. The most robust criticisms generally involve firstly the fact that the five factors aren't all-encompassing, though it isn't claimed that they are, merely that they exist, secondly that the five factors are not fully independent of one another but exhibit correlations amongst themselves (N & E are notorious here with correlations approaching -.50), though this also doesn't dispell their existence, and thirdly that self report bias remains difficult to deal with. The latter criticism seems more important to me personally than the others, because it calls the utility of the measure into question and leads to speculations about a possible "Flynn Effect" (spurious increase in scores over the generations). But even here, the various instruments testing the FFM have shown good validity and reliability, so the criticism is merely a quibble over the accuracy of the instruments used.

In conclusion, while some of Costa & McCrae's statements may be debatable (McCrae's interpretation of these traits as purely biological predispositions is indeed ludicrous) and while the instruments may be imperfect, it's extremely difficult to provide a meaningful and scientifically accurate criticism of the Big Five itself.


Harkenbane 08:07, 27 Apr 2005 (UTC)


Please, could anyone mention the name of O.P. John in this article. He IS a real father of Five-Factor Theory. However you can also correct the comment about the same factoring in other languages. There are different numbers of secondary level factors in different languages. (4 to 7(I think in japanese)) Im not too strong in English to write there, so please...

Martin Pírko, Czech Rep., 16 July 2005


Two comments. First, awful beginning. The reader has to wade through a bunch of stilted history before getting to the actual five factors. Few people looking for a Wikipedia entry will care about the history: please put the five factors earlier in the article and make the "origins" stuff later. Second, the five factors are often presented as contrast classes (e.g., extroversion/introversion, etc): why not include this (even if you have to include a caveat that introversion is just a negative value of the extroversion dimension this should still be included). Such breadth would be more helpful than the origins and its talk of new-fangled computing machines.


I removed "even a bit eccentric" as a discription of how introverts might be perceived for two reasons. One - eccentricity, to my mind at least, doesn't imply invtroversion, albeit introversion in the extreem does imply eccentricity. Two - the sentence was discribing negative perceptions of introverts, eccentricity isn't universally regarded as a bad thing. I recgonise that there may have been an element of irony in the phrasing, but it annoyed me anyway. --86.11.167.138 21:20, 15 November 2005 (UTC)


OK, I tried to address most of the concerns I saw here (although I've never heard of O.P. John and am really not very interested in or knowledgable about the origins of the Big Five).

I've reinstated and expanded the description of the five factors which originally stood at the start of the article, and tried to at least imply the contrasting nature of high and low scores on these scales (which I think is better than inventing a series of terms for the low poles of these dimensions which often don't exist in the literature).

I also cleaned up the main descriptions by quoting directly from an article by Dr. Johnson; whoever wrote the text I overwrote ought to find some sources for the various claims made there, many of which were misleading (E.Q. is positively predicted by high scores on E, A, C, O, and low scores on N, not just high scores in E and A) and others which I've never seen before (are Low-C's really better at multitasking? Where is the research for this?).

Finally, I fixed some unencyclopedic language in the future of the FFM section and tried to give it a better wrap-up. Hopefully others will find these changes to be acceptible. Harkenbane 20:18, 19 November 2005 (UTC)

Too much expansion in the introduction. I don't think the assertion "Each factor consists of a number of more specific traits" is correct: a factor is the product of a statistical analysis, each of the Big Five designates a trait artificially identified by such mathematical treatment. It does not "consist of" nor "include" anything, it is a single trait in itself. On the opposite, the basic point about the Big Five is that they are self-sufficient, that they can efficiently (though not perfectly) replace all other traits. To kow or describe someone, you can content yourself with evaluating her/him on these 5 aspects; only if you want to be more precise, you may evaluate her/him separately on each or several correlated traits. --Freb 15:00, 20 May 2007 (UTC)
Paragraph "Hiatus in research": except for last sub-paragraph, this paragraph does not pertain to the Big Five but to personality research in general. It should be put in [Personality_psychology|Personality psychology] and replaced by something like "This pioneering research has been forgotten for two decades because researchers failed to [Personality_psychology|demonstrate correlation between personality tests and observed behavior], until emergence of a more appropriate methodology around [1980]." And "This event was followed by..." could be turned into "This convergence of the Lexical Hypothesis with theoretical research lead to...", as previous versions said.
Paragraph "Consensus on the Big Five" partly clears up the nature of the Big Five and would be welcome at the beginning, partly details scientific findings and therefore should be included in the following paragraph. I suggest putting 2d sub-paragraph into the Overview and replacing the two last ones with a shorter statement at the beginning of paragraph "Selected scientific findings": "They have been proved significantly correlated with behavior as observed through professional performance (Barrick and Mount, 1991, 1998) or in personality disorders (Saulsman and Page, 2004)."
Last but not least, all the history should be put far further, just before "Criticisms", since a generally accepted piece of science (opposite to candidate theory) is just fact, how it has been established is an independent and secondary point. --Freb 16:27, 20 May 2007 (UTC)

Have a look: John, O. P. (1990). The "Big Five" factor taxonomy: Dimensions of personality in the natural language and in questionnaires. In L. A. Pervin (Ed.), Handbook of personality: Theory and research (pp. 66-100). New York: Guilford. He was first to put personality traits in OCEAN order - mnemonic...

Paragraphs "Biology of extroversion", "Causes of Openness", "Correlates of Openness" (except for the end) and "Biology of Openness" should appear under "Scientific findings", or? --Freb 17:01, 20 May 2007 (UTC)

[edit] Are authors personal links appropriate?

I am concerned with the Psychometrics link "A layman's overview of the field of psychometrics." and its promotion of the eugenics arguments on the pages of the website linked to from that page. Considering the contraversy of eugenics and Harkenbane's relationship to the site I don't believe that the link is appropriate -- or atleast that there should be some disclosure about the relationship between those writing / editing the article and the external resources being linked to Doctus 02:35, 19 December 2005 (UTC)

That's fine; I'll just remove the link. By the way, if you can find other good links, go ahead and add them. Harkenbane 23:37, 21 December 2005 (UTC)

[edit] SLOAN Notation

Surely this needs to be mentioned? matturn 05:16, 29 December 2005 (UTC)

Mmm... I'm not convinced that sloan is scientific. It seems largely an attempt to import MBTI approaches into the Five Factor model. But go ahead and add a mention - you may know more about it than anyone else here does. Harkenbane 19:22, 2 January 2006 (UTC)

You seem to know more about it than I do :-) I brought it up because most places I've seen the Big Five mentioned online, SLOAN is way invididual personalities are represented. I'm sure this is less accurate than individual scores for each continuum, but it still seems important. matturn 13:14, 24 January 2006 (UTC)

[edit] question

You say, some people say this model neglects Conservativeness... But isn't that what Openness to experience is about? I can't tell the difference. Note, this is a question regarding a specification of what you mean. Not whether or not what you mean is correct. _____________

Psychometric Openness is only peripherally about liberalism versus conservatism; the subscales of factor O which *best* correlate with Openness itself appear to be Openness to Fantasy, Openness to Ideas, and Openness to Aesthetics; the three other scales - Actions, Feelings, and Values - appear to be less central to Psychometric Openness. You can read about this in _The Big-Five Trait Taxonomy: History, Measurement, and Theoretical Perspectives_ by Oliver P. John and Sanjay Srivastava; I highly recommend reading through that file if you're interested in understanding the Big-Five:

http://darkwing.uoregon.edu/~sanjay/pubs/bigfive.pdf

In other words, Openness is "really" about imagination, intellectance, and aesthetics; liberalism is not a central characteristic to factor O (even though it is still related to factor O).

That being stated, I think it's silly to talk about the Big Five failing to describe a person's tendency towards certain political orientations - I only left that part in the text in the interests of politeness and collaboration. Conservatism and liberalism may not correlate perfectly with psychometric Openness, but that's largely because those are value systems rather than personality traits, not because the Big-Five is missing anything. Harkenbane 19:53, 2 January 2006 (UTC)

[edit] Number of Factors

From the article: Moreover, the methodology used to investigate these phenomena (factor analysis) does not have a well-supported, universally-recognized scientific or statistical basis for choosing among solutions with different numbers of factors.

The concept of explained variability can be brought into play here. What you find is that the first factor acounts for 70% of variability the second for 20%, and so on following a Zipfs law type of distribution (numbers not exact). This is a common patten amongst all multi factor models especially the related Principal components analysis. So with five factors you maybe explain about 95% of variability. Adding extra factors does not significantly add to explained variability. From the Gordon Allport and H. S. Odbert quote:

Those individual differences that are most salient and socially relevant in people’s lives will eventually become encoded into their language; the more important such a difference, the more likely is it to become expressed as a single word.

we can see that the five factors account for the most salient and socially relevant factors. The other factors, whilst important, are not as salient and socially relevant.

A pet theory of mine is to do with the fact that we can only hold 5-7 items in short term memory at one time. You find that virtually all of the various psycological models have less than 5 items. This maybe more to do with the our (the observers) ability to hold concepts in our heads than the mathematics. So five factors is a nice comprimise between mathematical explainability and being able to recall and comunicate this information. --Pfafrich 10:59, 5 January 2006 (UTC)


You're quite correct that if the solution accounts for 95% of the variability, then it would be silly to extract more factors. Do you know of any published work that accounts for 95% of the variance in R? Because my recollection is that 50% is more like what these studies will achieve in their final solutions (although this is rarely published).

The problem that I was addressing is that there are many criteria for how many factors to extract from a factor analysis. The most common (in my experience) is to retain all eigenvalues greater than 1. I think parallal analysis or examination of a scree plot are far superior to this. Specifically, the problem is that two competent factor analysts, using different rules, can extract very different solutions, particularly after adding in rotations.

Your pet theory is interesting but is't it 5-9 (7 plus-or-minus 2)? I think there really are five broad factors that, in turn, have more specific traits underlying (just like the facet scales of the NEO). And having helped 16PF users, I do absolutely agree that our cognitive limits shape our preference for simpler models. The 16PF is too complex for a lot of people, certainly the NEO's 30 facets are far too complex to be comprehended as a whole.

Amead

No backing for the numbers, probably wildly inacurate. I have done some post doctoral work using PCA recording shape variability, there I did find a very fast drop off in the contribution of factors, and also a long tail. I suspect this is common pattern.

For 16PF I'd guess there is quite a lot of corelations between the factors (the word abstract apears twice), so not really orthogonal. It would be really nice to get hold of some real data.

I note neither Parallel analysis or Scree plot have articles. I've not come across them before.

This is all rasing lots of questions. Quite what is the explained variability being measured? Is it variability in types of personality acording to some arbitary set of axis or is it more a textural analysis of the words we use to describe ourselves, which I think is closer to the original analysis? Theres all sort of issues with scaling of factors as well.

Time to head for the original papers! --Pfafrich 21:43, 5 January 2006 (UTC)


I agree that the number of factors to break personality down into is somewhat arbitrary, and in fact there are a variety of solutions which seem useful. While the psychoticism scale of Eysenck's three factor (P-E-N model) proved internally inconsistent, recent research has found that a two-factor solution works well, where the two factors are dubbed alpha (A, C, and the inverse of N) and beta (E and O). But there's a large amount of personality variation within any given level of alpha or beta, so the Big Five are a bit more accurate. Although it would probably be useful to break personality into more factors than 5, the 16 factors discovered by Catell weren't replicable, which is the main reason they were abandoned. Possibly the lexical hypothesis itself would have to be abandoned before future research could refine our understanding of personality - human language is useful, but far from perfect. It's imaginable that there is some variation in personality which we are instinctively unable to notice or even made not to notice, and any variation of this sort would be totaly invisible to language and even self report. Until we know more, however, it appears that the Big Five is the best we have. Harkenbane 07:38, 10 January 2006 (UTC)


Harkenbane: You should learn more about the 16PF. The 16PF is far from abandoned and it's actually a two-tiered system with a broad five-factor solution indicated by the 15 underlying personality factors (the so-called 16TH factor, B, is actually some sort of general intelligence factor, not a personality factor). I believe Goldberg credits Cattell as the first to find five broad factors but Cattell never thought they would be useful for anything which is why he pursued the 16-factor solution. And, in fact, Goldberg has an as-yet unpublished chapter (cited in the recent Buchanan, Johnson & Goldberg paper) in which he compares the cross-validated criterion correlations of a number of instruments... He concludes that the 16PF model predicts best (although his IPIP version of the 16PF outperformed the actual 16PF) beating the NEO at both the five- or 30-factor level, etc.

Actually, this idea of having a broad five-factor level and a "facet" level is the norm, rather than the exception, even for so-called five-factor instruments like the NEO. Off the top of my head, NEO, 16PF, Hogan HPI, MBTI Step II all have this two-tiered (i.e., second-order factor anlaysis) structure. What is also true, however, is that each of these instruments has a different represnetation of the "facet" level; the NEO has 30 facets, the 16PF has 15+1, .. I don't recall the Step II structure bug Hogan forms what he calles HIC's and he has dozens. Amead 03:20, 25 January 2006 (UTC)

Well, let me ask you a very simple question: Are the 16 factors replicable, or aren't they? My understanding was that modern research simply couldn't replicate all 16 factors. I suspect that this is a problem with the tools used - self report is probably too clumsy to render this level of detail - but this is an untested opinion; to date, the "Little 15" haven't proven to be as robust as the Big 5. Of course, I will concede that neither the 16PF model, nor PEN model, nor several other models are completely phased out, but that really doesn't mean much when taking into account the fact that even Freud's and Jung's ideas are still bandied about in college classes as though they're meaningful. Most modern research into differential psychology uses the Big Five (although the sixth factor, g, is often avoided by researchers who are all too aware of the politically unpopular findings surrounding that factor). Harkenbane 08:35, 26 January 2006 (UTC)


Harkenbane: Your question isn't that simple. On the one hand, if you do research using an instrument such as the 16PF, yes, you do "replicate" the factor structure. For example, when the researchers at IPAT wrote hundreds of brand new 16PF items during the development of the 16PF, administered them to hundreds of volunteers, and then performed exploratory and confirmatory factor analyses... Yes, the factors were confirmed. Both in factor-analyses of the new scales and in analyses of the old and new forms.

But the same is true of the NEO with it's 30 facets and Hogam's HPI with it's dozens of HIC's. So if you mean, "Do researchers agree on any breakdown of the five broad factors?" then no. Since the choice of factors is somewhat arbitrary, there is no reason to expect that they would agree. But my point are that: (1) the big five is far too broad, so (2) virtually all commercial personality tests have adopted this two-stage model and (3) I have seen evidence (like Lew Goldberg's unpublished "Comparitive validity" chapter) that convinces me that these facets have significantly higher validity than the broad traits, even after you account for the different numbers of predictors (by using empirical cross-validation, in the case of Lew's chapter). Goldberg's chapter actually found that the 16PF factors were most predictive of all factorial models (e.g., NEO-5, NEO-30, Hogan-6, Hogan-36, etc.)

And I guess your also asking whether anyone has done the same things Cattell did and gotten the same results. That's not as cut-and-dried as you suggest. AFAIK, only one replication has been attempted (by Norman?) and it's true, he didn't find the same results. But one contrary finding does not constitute a stunning refutation and I'd argue that Norman's different findings could easily reflect that exploratory factor analysis is a poor way to find replicable factors. There are far too many judgments required (like **how many factors to extract** which is absoluetly **not** a solved problem).

Amead 07:31, 8 April 2006 (UTC)

_____________________

> Your question isn't that simple.

Actually, it's pretty simple. Numerous attempts to discriminate between the validity of 5 or more factors have been carried out, such as:

Saucier, G. (1997). Effects of variable selection on the factor structure of person descriptors. Journal of Personality and Social Psychology, 73, 1296-1312.

Or consider this study, which specifically checked Catell's factor structure using Cattel's own questions, and still got five factors:

Tupes, E. C., & Christal, R. C. (1961). Recurrent personality factors based on trait ratings. Technical Report, USAF, Lackland Air Force Base, TX.


> if you do research using an instrument such as the 16PF, yes, you do "replicate" the factor structure.

But that's because the test was created specifically with questions that would load on the factors they were supposed to load on. That tells us nothing about how many factors really do describe the broad range of human variation.


> So if you mean, "Do researchers agree on any breakdown of the five broad factors?" then no.

True, and I already conceded this point. Yet at the same time:

"what is different about the personality research community today versus ten years ago is that there is a clear trend towards embracing a single model — the FFM — as the research paradigm to follow."

Howard, Pierce J, and Howard, Jane M (2003). "The Big Five Quickstart: An Introduction to the Five-Factor Model of Personality for Human Resource Professionals (Revised)" Center for Applied Cognitive Studies (CentACS) Charlotte, North Carolina.


> (1) the big five is far too broad

You haven't offered any reason to believe that five are "too broad." Too broad for what? For many purposes, five appear to be too many, even, especially in light of the findings regarding alpha and beta - for instance, vengefulness corresponds well to low alpha and latent inhibition to low beta. The question is whether the factor structure of personality is such that, at a given level of detail, a certain number of factors emerge. At the highest level, it seems that alpha and beta emerge, and lower down, the Big Five become apparent. But after this things get murky. The 30 subtraits tested for by Costa and McCrae are (I believe) chosen capriciously, while the 15 factors don't replicate properly. At the present time, it can't be stated confidently what (if anything at all) lies beneath the Big Five.


> I have seen evidence (like Lew Goldberg's unpublished "Comparitive validity" chapter) that convinces me that these facets have significantly higher validity than the broad traits, even after you account for the different numbers of predictors (by using empirical cross-validation, in the case of Lew's chapter).

While I'd be interested to see the research you're alluding to, it isn't surprising that a more narrow trait could better predict a specific outcome - the question is whether that narrow trait is itself a unitary factor.


> AFAIK, only one replication has been attempted (by Norman?) and it's true, he didn't find the same results.

If you have one study which finds X, and one study which finds not-X, then you have no means of discriminating between the two. Fortunately for the purposes of this investigation, a variety of other studies have been carried out which find that the 16 factor model doesn't hold up.


> There are far too many judgments required (like **how many factors to extract** which is absoluetly **not** a solved problem).

It is a fairly easily solved problem in this case: Five are replicable. More factors aren't. Maybe there are more factors to be found (and I do believe that there are); maybe the emerging consensus in the Big Five will later turn out to be unfounded; science is progressing all the time. But the state of the research as it exists now is unfriendly towards Cattell's efforts of 60 years ago, and science rarely backpedals to an old model.

Harkenbane 18:38, 8 April 2006 (UTC)

>> Your question isn't that simple.

>Actually, it's pretty simple.

Ah. Then you will have a simple way to express a hypothesis test... please let us know what that might be. How does one "replicate" exploratory factor analyses?

Amead 05:47, 9 April 2006 (UTC)

[edit] Genetic Links

There was a good article in new scientist on the issue and it seems that there is quite a strong correlation between expressed geans and some of the factors, especially extroversion. It would be nice to add something on the subject, but the only ref I've got [1] is subscription only. --Pfafrich 21:59, 5 January 2006 (UTC)

Although I don't know of any specific genes which relate to personality, Bouchard and McGue, in their 2003 meta-analysis, "Genetic and environmental influences on human psychological differences" published in the Journal of Neurobiology, issue 54, found the Big Five traits to have the following heritabilities:

O-57% E-54% C-49% N-48% A-42%

I've seen a veriety of individual heritability studies, and they've returned highly similar results, indicating that those heritabilities are very accurate. It's interesting to note that the beta factors of E and O have the highest heritabilities while the alpha factors of C, N, and A have the lowest, although all the factors have heritabilities near 50%. It's also interesting to note that political conservatives are outbreeding political liberals in modern America, and given the high heritability of factor O and its modest relationship to political beliefs, this may have the unforseen consequence of gradually reducing psychometric Openness throughout the population.

Harkenbane 07:48, 10 January 2006 (UTC)

[edit] "extrAversion", not extrOversion"!!!

No, both are acceptable. Harkenbane 04:14, 3 March 2006 (UTC)

Although both variants can be seen in use, extrAversion is the contemporary standard in the academic literature. Note that:
  • Major psychology journals (including APA journals) have predominantly used the "A" spelling in the last couple of decades (see, for example, recent issues of the Journal of Personality and Social Psychology).
  • Google Scholar brings about 36,000 hits for extrAversion, versus about 8,000 for extrOversion.
  • All of the external links in the article that use the word spell it with an "A."
Based on these observations, I'd suggest going with the extrAversion spelling and noting the "O" variant. Irbatic 22:43, 22 August 2006 (UTC)
www.dictionary.com entries list "extraversion" version as a variant of "extraversion." Remember that although these are psychometric constructs, the terms are also used by laymen. I think it should stay as Extroversion. Harkenbane 02:27, 24 August 2006 (UTC)
(Oops, I meant "extraversion" as a variant of "extroversion".) Harkenbane
To me, "the terms are also used by laymen" is an argument for going with the expert spelling. This is not an article on the concept of "extra/oversion" in everyday usage; this is an encyclopedia article on the Big Five personality traits. In that context, the "A" spelling has been used very consistently from Goldberg (who coined the term "Big Five" from which this article takes its title) onward. (If Harkenbane and I cannot convince one another, maybe someone else wants to jump in?) Irbatic 00:03, 30 August 2006 (UTC)
No, I think your argument 'This is not an article on the concept of "extra/oversion" in everyday usage; this is an encyclopedia article on the Big Five personality traits' holds force. Feel free to change the references of "o" to "a," leaving in "o" as a variant. Harkenbane 04:10, 30 August 2006 (UTC)
ExtrAversion is pretty much standard in psychology now, and it was the original spelling used by such theorists as Jung and Eysenck. ExtrOversion seems to parallel IntrOversion better, but it doesn't make sense linguistically. Extraverts go extra or "additional to themselves" to seek social contacts and stimulation. This is in the same sense as extrAsensory perception and extrAordinary. I would recommend this usage over extrOversion. Jcbutler 17:58, 30 November 2006 (UTC)

[edit] Added links

I thought the Overview section was more useful with some links.Fionah 13:06, 15 March 2006 (UTC)

The linked articles don't seem to correspond to the definitions of the Big Five with the exception to extraversion. There's already descriptions of each factor within the article, so I'm suggesting that we unlink them.--Janarius 14:30, 15 March 2006 (UTC)

I prefer fewer links througout the text, as they impair legibility, and I therefore support anyone who wants to de-link words. Harkenbane 21:37, 15 March 2006 (UTC)

I prefer more links throughout the text, as they improve comprehension, and I therefore support anyone who wants to link words... as long as the words being linked to describe uncommonly understood concepts. matturn 07:58, 18 March 2006 (UTC)

Given the fact that anyone can always type a word into the search bar (or double click and paste it in), I think it's better to be conservative about linking. Still, the phrase "as long as the words being linked to describe uncommonly understood concepts" is probably the key. Articles commonly link dates, world leaders, countries, and other words which are familiar to anyone with a 6th grade education, but words which really do confuse most readers benefit from a handy link. Harkenbane 23:50, 4 April 2006 (UTC)

I think the links are useful and have been working to improve the individual entries for neuroticism, agreeableness, etc. Feel free to visit these pages and comment on or improve them as you see fit. Jcbutler 18:01, 30 November 2006 (UTC)

[edit] Diaereses?

Um, as much as I personally love spelling words like "altruïsm" with the diaeresis, it's really non-standard & somewhat bizarre, so I'm going to take them out, barring further objections. Any thoughts? Scott! 19:30, 25 September 2006 (UTC)

[edit] Contradiction about "consensus"

This article contradicts itself. First, it has a section called Consensus on the Big Five which isn't really about consensus. Then, under the section called Selected scientific findings, the quote "Ever since the 1990s when the consensus of psychologists gradually came to support the Big Five,..." asserts that a consensus of psychologists support this model. Later in the article there's a Criticisms section, which contradicts the assertion that there's a consensus. This definitively needs a rewrite. Nathanm mn 20:47, 19 January 2007 (UTC)

I think it is possible to have both consensus and criticism. There can be general agreement that five major traits are central to human personality, and yet people may disagree on certain aspects of the model, or criticize it for being overly descriptive and not causal, etc. --Jcbutler 21:00, 19 January 2007 (UTC)
In the literal sense, consensus means complete agreement. If it's used in the sense of "general agreement" then at best it's weasel wording. I think it's always a poor choice of wording in academic writing, especially in the sciences (both the "hard" sciences and social sciences). I'm not a psychologist, and this article is poorly cited, so there's no easy way for me to get even a rough idea of how much agreement there is among psychologists. However, there's an easy way to fix this article:
  • The first section heading I referenced just isn't very representative of the paragraphs in that section. Picking a more descriptive title would eliminate that part of the problem.
  • The second section could be reworded somehow to avoid using the word "consensus." Maybe talk about the increasing use of the Big Five model or something.
Hope that helps.Nathanm mn 22:51, 19 January 2007 (UTC)
According to Wikipedia... Consensus is "a general agreement among the members of a given group or community, each of which exercises some discretion in decision making and follow-up action."
Given this definition, and given that psychologists who work in this area often use the term consensus when they are talking about the Big Five, I see no problem with using the term here, and no contradiction in this article. --Jcbutler 23:12, 19 January 2007 (UTC)

I'm no psychologist, but I recently saw these traits listed in a huge UK coffee table book, "Human," and it convincingly manages to avoid the polemics - each trait is paired with it's polar opposite on a continuum, making for 10 trait titles, but it never sides with one trait over its opposite. Some people are just this way, and some people that.

For instance, why not list "Conscientiousness vs. Spontaneity"? It's pretty obvious, as such, that both are desirable characteristics to be shared among us all. And yet, you've got "Agreeableness" vs "suspicious and antagonistic towards others"? Come on. You can show a bit more "Openness" than that! Franklin Roosevelt put forward a more "agreeable" personality than Churchill's "antagonistic" type, but (whether I'm right or not on those two), both types are needed. We need, say, thin skinned people to say "bull" when everybody else is just going along, failing to point out the king (or president) has no clothes, for instance. We all say "bull" at times, but we really need those who are constitutionally that more than the rest.

How about "Agreeableness" vs. "Principled." I can't remember what "Human" listed, but the principle is obvious (at least it should be to those thinking themselves up to writing for an encyclopedia) - be open to all, reasonably so. Start with 10 positive definitions, implying all types are need in our lives, and then give their aberrations, or excesses, etc.

The problem is that we aren't starting with traits and then setting out to find evidence for their existence. The entire framework of the five factor model is empirical; empirically speaking, 5 factors emerge, and low scorers on factor II (Agreeableness) are more likely to describe themselves as "suspicious" and "antagonistic." I do agree that "disagreeable" is better and more natural, but I think don't think this is a very important issue. The reality is that many traits are more socially desirable than others. Harkenbane (talk) 03:22, 19 January 2008 (UTC)

[edit] Proposal to re-name this page "The Five-Factor model"

Can I please propose that this is re-named "The Five-Factor" model rather than "The Big Five personality traits"? My reason is that Costa and McCrae, in one of their many publications on this issue, I think it was in 1992, clarified that they preferred this term. I know that if you a do search on either Google or Google Scholar, one gets about three times as many searches for "the Big Five" and "personality" as one does for "five-factor model" and "personality" but should it not be the case that Wikipedia should be based on what academic journals say, not simply a slave to internet records? ACEOREVIVED 19:34, 6 June 2007 (UTC)

The "Big Five" and "Five Factor Model" are actually subtly different. If you check the academic journals, you'll note that Goldberg discovered Big Five, Costa & McCrae originated the FFM. Since the FFM is essentially a theoretical interpretation of the Big Five, "Big Five" is broader, and therefore probably the better term. Harkenbane (talk) 03:25, 19 January 2008 (UTC)

[edit] Big Five system as "based on language"?

The article says: "What separates the five-factor model of personality from all others is that it is not based on the theory of any one particular psychologist, but rather on language, the natural system that people use to understand one another." But what does it mean to say that a model is based "on language"? This sentence has the strange implication that psychologists generally come up with bizarre, arbitrary ideas, but the Big Five idea somehow managed to escape this dilemma. If this so so, surely an explanation is in order. It's very tempting to delete this sentence altogether. Maybe I'll do that later on.


[edit] Neuroticism, Extro/introversion

"Freedom from negative feelings does not mean that low scorers experience a lot of positive feelings; frequency of positive emotions is a component of the Extraversion domain." Stop me if I'm out of place, but what do extroversion and introversion have to do with positive and negative emotions?

Yes, it's well known that E relates to positive emotionality, N to negative emotionality. E- N- individuals experience few highs or lows, while E+ N+ individuals experience soaring highs and crushing lows. Google is your friend here.
"High extraversion indicates proneness to experience positive emotions" from http://www.frokk.com/users/moon168/pictures/
Extraversion refers to a person’s tendency to be venturesome, energetic, assertive, and sociable and to experience positive emotions (e.g., joy) http://ajp.psychiatryonline.org/cgi/content/full/164/11/1714 Harkenbane (talk) 03:09, 19 January 2008 (UTC)

I was not taught that positive and negative emotions were a part of introversion and extroversion, rather that they were on the neurotic-stable spectrum. I would agree with the first part of the sentence, that if a neurotic person is free from negative feelings that they don't necessarily have positive feelings in their place. But I don't see what the second part of the sentence has to do with neuroticism. The second half of this sentence, and also a sentence in the extrovert/introvert part of the article, claim that extroverts or introverts, one, experience more positive emotions than the other. I have never read or heard this to be true. Again, I do agree that a lack of positive emotions doesn't mean the presence of negative emotions - and vice versa. But, I don't see what this has to do with extroversion and introversion.

A very important concept to understand is that the Big Five traits are the result of empirical research, not theoretical speculation. While I agree that there is no a priori reason to assume that E would be related to positive emotions, the observed fact is that E does indeed relate to the way people report their emotional states. Harkenbane (talk) 03:09, 19 January 2008 (UTC)

Also, I think statements like "low scorers" and "high scorers" could be replaced by saying something more accurately descriptive like "more stable" or "more neurotic". The big five are not measured by "high and low", they are measured by stability and neuroticism, extroversion and introversion et c. I don't quite understand why that language was used or how the writer decided which end of the spectrum was high or low. —Preceding unsigned comment added by 70.238.154.170 (talk) 06:41, 12 November 2007 (UTC)

The consensus of psychologists determines which end of the spectrum is high and which is low. Some of the reason the traits are named and used as they are is historical; for instance, Eysenck used Neuroticism rather than Emotional Stability, so Neuroticism is still more common. In other instances it appears more practical, with "disagreableness" being less intuitive than "agreeableness," so the latter is used. But which direction a given scale is set up is essentially irrelevant. Harkenbane (talk) 03:09, 19 January 2008 (UTC)

[edit] Openness versus intellect

It is somewhat ironic that Goldberg (1993) is cited at the beginning of the article but there is mention at all regarding the discussion of "Openness" versus "Intellect", see e.g. http://dx.doi.org/10.1002/per.2410080405 . The Costa and McCrae questionnaire methodology usually calls that factor "openness to experience", but for Goldberg and others of the lexical school the factor was usually labeled "Intellect" or "Culture" or something like that. Maybe this should be added to the "criticism" section. Wolf rauch (talk) 10:51, 12 December 2007 (UTC)

I keep meaning to get around to this; if someone else knows about it, please give it a shot! (The controversy regarding the correct name for Openness actually goes much deeper than merely an argument between Goldberg and C&M, and my personal opinion is that not one of the big names in the field really knows what to call factor V.) Harkenbane (talk) 21:35, 31 March 2008 (UTC)

[edit] Sex differences

Citation needed - Sex differences : Men and women show differences in Big Five scores across cultures, with women scoring higher in both the Agreeableness and Neuroticism domains. These findings may indicate innate differences in personality between the sexes, but by themselves the results are not conclusive. Jesse McNulty (talk) 18:41, 29 March 2008 (UTC)Jesse McNulty

If you are going to call woman more neurotic please cite the source. —Preceding unsigned comment added by 66.66.218.26 (talk) 18:39, 29 March 2008 (UTC)

Maybe more neurotic, but more aggreeable and conscientious too. In any event, I updated this section and supplied a solid, recent source. --Jcbutler (talk) 19:46, 31 March 2008 (UTC)

Rats, you beat me to it! My source was older and used fewer cultures as well, but I added it as well. Note that the finding of higher C and E in females is not nearly as universal as higher A and N; after years of reading Big 5 studies, I can't remember ever seeing the A and N difference failing to appear, but there were many times when I can recall E and C showing no difference or a difference favoring males. (In fact, Eysenck's older tests typically found higher E in males - how different groups will fare is sometimes determined by the way the test is structured.) Harkenbane (talk) 21:31, 31 March 2008 (UTC)

[edit] Repetition?

In recent edits, user 122.3.195.209 has placed the quick descriptions of each trait which appear at the start of the article into the start of each trait's subsection. Is such repetition really helpful? Harkenbane (talk) 16:46, 13 April 2008 (UTC)

It's slightly repetitive, but it's also nice to have quick definitions at the beginning. Maybe these sections could be rewritten to avoid redundancy. --Jcbutler (talk) 21:18, 14 April 2008 (UTC)


[edit] Article seems a little politically bias

The article seems to be a little politically bias, only showing positive personality traits associated with liberalism, such as higher openness (which corelates with higher creativity), while failing to report positive traits associated with conservativism, such as higher conscientiousness (which corelates with a stronger work ethic). Conservatism (or at least people that voted for Bush in this study), are also more likely to be extroverted (although whether this is positive or not is debatable) I doubt the bias is intentional, but if you are going to show the corelation of positive traits and political leanings in one area, then you should show them in all areas. --Jtd00123 (talk) 07:19, 25 April 2008 (UTC)

http://www.eharmony.com/labs/blog/2008/01/personality-predicts-voting-intention/

I agree with you - you should add the link between C and rightism to the article. Be aware that it's much weaker than the relationship between O and leftism, of course, but I think it has been substantiated on more than one occasion, and I might be able to dig up more references on the C-rightism link to add myself. (Honestly, I feel badly about leaving this article alone for so long, but then again I think it's in much more capable hands than some of the other articles I'm working on!) Harkenbane (talk) 18:14, 25 April 2008 (UTC)

[edit] Frank Sulloway's Born to Rebel "discredited"?

I changed "discredited" to "called into question", and generally softened the tone with some rewording of the section on birth order. A brief survey of abstracts using Google Scholar showed that some attempts at replication seem to weaken Sulloway's case, and others supported it.

I'm not a pyschologist, much less a specialist in birth order theories. I don't know the current status of the theory. However, I think it's important to be careful here, given that Sulloway has been accused in the past of manipulating and falsifying data, but challenged this characterization legally and won. [2]

If Sulloway's birth order theories are still a subject of serious debate and attempts to verify, they shouldn't be characterized as "discredited"; at worst, they are still under challenge. A study as recent as 2006 [3] seems to lend support to Sulloway's approach.

The book cited in this section as "discrediting" Born to Rebel seems to be a good read, but ... well, I'm not interested in the fact that Sulloway's eyes are "blue" ("he's naive!") or "wideset" ("and did I mention dumb?"), at least from what the author can tell "from photographs" (as if to underscore how little desire she had to meet him?) The book is a generally snotty treatment, but ... well, what else would you expect from a firstborn? ;-) If it's right, I suppose that's OK. But I don't know if she's right. Yakushima (talk) 12:13, 13 June 2008 (UTC)