Gender Genie

Discussion in 'The Lounge' started by peachalulu, May 4, 2013.

  1. GingerCoffee

    GingerCoffee Web Surfer Girl Contributor

    Joined:
    Mar 3, 2013
    Messages:
    18,385
    Likes Received:
    7,080
    Location:
    Ralph's side of the island.
    The research the algorithm is based on addressed the issue that the analysis did not depend on content.
    I think a few people in this thread are erroneously oversimplifying the analysis.

    Two things are important, one it only professes 80% specificity, meaning there is an 80% chance, not 100% chance gender can be identified by the algorithm. That suggests we overlap quite a bit but that there are some interesting differences in language utilization between men and women.

    The second thing is I'm not sure I saw anywhere in the study, (maybe someone else did), how long a piece one needs to submit for the algorithm to be applicable. I would think one needs at least a few hundred words but maybe it's more like 1,000.
     
  2. GingerCoffee

    GingerCoffee Web Surfer Girl Contributor

    Joined:
    Mar 3, 2013
    Messages:
    18,385
    Likes Received:
    7,080
    Location:
    Ralph's side of the island.
    You said,
    :confused: You can't dismiss the variable, male vs female, while agreeing there is variation, unless you have a good reason for suggesting the researchers were wrong about the variable.

    'Bland' was to make the algorithm content neutral.

    I agree the question is, why are there measurable differences between genders? But one does need to demonstrate there is a difference before one goes looking for the underlying cause.
     
  3. GingerCoffee

    GingerCoffee Web Surfer Girl Contributor

    Joined:
    Mar 3, 2013
    Messages:
    18,385
    Likes Received:
    7,080
    Location:
    Ralph's side of the island.
    Average. ;)

    Just kidding. There is evidence that androgen exposure in the womb, for whatever reason, increases the length of one's ring finger. Beyond that the differences are variable. It's probably not a whole lot more predictive of behavior or personality than one's horoscope (aka no effect). But in my case I can't help wondering if it's connected to my always having been such a tomboy.

    It wouldn't surprise me to find out my writing matched my personality.
     
  4. mbinks89

    mbinks89 Active Member

    Joined:
    Nov 14, 2012
    Messages:
    548
    Likes Received:
    27
    Location:
    Montreal
    I've used it. I'm a female too, supposedly, although the section I had it analyze was my description of a relationship, so maybe that contributed to it? I don't buy it though.

    Also, you're male? No offense, I just always had you pegged as a woman with your dp.
     
  5. funkybassmannick

    funkybassmannick New Member

    Joined:
    Mar 22, 2011
    Messages:
    828
    Likes Received:
    31
    Location:
    Chicago, IL
    I may not have put what I mean in the best words, but here is another attempt:

    It has external validity because it appears the findings are generalizable to the public.

    It has weak construct validity in that there is so much more to gender and writing than determiners, quantifiers, and pronouns. It's barely scratching the surface of what makes you a male writer or a female writer.
     
  6. GingerCoffee

    GingerCoffee Web Surfer Girl Contributor

    Joined:
    Mar 3, 2013
    Messages:
    18,385
    Likes Received:
    7,080
    Location:
    Ralph's side of the island.
    I'm very much female. I'm very much heterosexual, one son, single, with a couple long term relationships in the past.

    Besides that, the rest of your post confuses me. What don't you buy?
     
  7. GingerCoffee

    GingerCoffee Web Surfer Girl Contributor

    Joined:
    Mar 3, 2013
    Messages:
    18,385
    Likes Received:
    7,080
    Location:
    Ralph's side of the island.
    The data demonstrate an inherent difference in communication style. Sure there are a gazillion other variables one could look at, but that's not what the researchers were trying to get at.

    Now the place I might nitpik this would be to see how well the underlying premise was supported in past research, that is the claim, female writing exhibits "involved" while male writing exhibits "informational".

    I have merely said the research supports the conclusion men and women communicate differently. What that means is another story. I think the underlying premise, one is 'involved' and one is more 'informational' could be biased by established stereotypes about men and women. Might be true, but I've not looked into it and can't say.
     
  8. ProsonicLive

    ProsonicLive New Member

    Joined:
    Apr 17, 2013
    Messages:
    117
    Likes Received:
    6
    I use a form of writing that I am not sure has a name. I write in third person but depending on what perspective I want the reader to view the character determines the way I write. If I want a female character to view a male character in an unflattering light, I will focus on what and why he is unflattering based on if she knows why she finds him unflattering or not. IE:she knows the character is a blackmailer or rapist; or if she finds the way he looks at her as creepy or views his actions as underhanded and not justified. I use the third person to present perceived fact so that the reader understands the characters reasoning. In other cases it can be used to mislead. So having the information the test provides, maybe you can use that to enrich your own writing when using perspectives. Also, take into account men tend to use more forceful actions and certainties. Women often choose to be more vague. I read female writers so that I can learn to write "like a woman" so that my female characters are not a romanticized male idea. (at least I try) whereas being male, I do not require as much education. Perhaps this test can lead to insight rather than perceived insult or failure.
     
  9. Cogito

    Cogito Former Mod, Retired Supporter Contributor

    Joined:
    May 19, 2007
    Messages:
    36,161
    Likes Received:
    2,827
    Location:
    Massachusetts, USA
    Not true. It's the other way around. Unless you can demonstrate a basis for why the variable is meaningful, and propose a test to invalidate that assumption, there's no reason to accept the variable as related to the result.

    The variable may be influenced by the sampling method. Or it may be related in complex ways to other variables, identified or not. Or it may be random/ Or it may reflect bias by the experimenter.

    Have you ever worked as a scientist? On the other hand, even some professional scientists don't really grasp the scientific method.
     
  10. GingerCoffee

    GingerCoffee Web Surfer Girl Contributor

    Joined:
    Mar 3, 2013
    Messages:
    18,385
    Likes Received:
    7,080
    Location:
    Ralph's side of the island.
    You are confusing the correlation/causation issue (what you are calling 'related') with the fact a variable is consistently associated with what is being measured.

    The researchers found their algorithm could successfully predict if a piece was written by a male of a female. Regardless of why the algorithm predicts author gender, it does nonetheless predict author gender.

    Unless you can fault the methodology or if one does not find repeatable results, then male and female authors write in a predictably different way.

    As for proposing a mechanism explaining the difference, that's not hard, there are many possibilities. If, however, one failed to find the actual mechanism, that doesn't negate the fact a correlation was found.

    As for all the possible reasons the methodology could fail, I asked you to say which error in methodology you actually found problematic. Anyone can describe all the generic reasons a study might be a bad study. I don't see any of those obvious failings looking at this research.

    It's easy enough to validate or invalidate the algorithm, just have different researchers test it.
     
  11. GingerCoffee

    GingerCoffee Web Surfer Girl Contributor

    Joined:
    Mar 3, 2013
    Messages:
    18,385
    Likes Received:
    7,080
    Location:
    Ralph's side of the island.
    Additional research can contribute to this discussion:
    Gender Differences in Language Use: An Analysis of 14,000 Text Samples
    As a feminist, there are some differences I gladly embrace. :)

    I don't see studies looking at other cultures so whether this is nature or nurture or a combination is a question not yet resolved.
     

Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice