Skip to main content

View Diary: Some thoughts on observational studies (22 comments)

Comment Preferences

  •  You can never be sure (8+ / 0-)

    but you can gather evidence.

    If you want a rigorous treatment of observational studies, read books by Paul R. Rosenbaum

    Democrats want us to share a dream. Republicans want us to fear a nightmare. I'm for dreams.

    by plf515 on Fri Jan 20, 2012 at 12:20:11 PM PST

    [ Parent ]

    •  Good stuff: keep doing this. (5+ / 0-)

      We need more diaries explaining the various elements of scientific methods.  

      Best if they're like this one: relatively short and stick to one topic, so people can read them in real time rather than "put on the stack to read later (after everyone has gone to bed)" and comment in real time.  

      "Minus two votes for the Democrat" equals "plus one vote for the Republican." Arithmetic doesn't care about your feelings.

      by G2geek on Fri Jan 20, 2012 at 02:59:28 PM PST

      [ Parent ]

    •  so I gotta' question: (4+ / 0-)

      What do you do about an apparent effect that is relatively small in size, but is persistent at a consistent level of statistical significance, and for which there is no clear mechanism?

      Do all of those get relegated to the status of "anecdotal" or worse, dismissed entirely?  

      "Minus two votes for the Democrat" equals "plus one vote for the Republican." Arithmetic doesn't care about your feelings.

      by G2geek on Fri Jan 20, 2012 at 03:00:51 PM PST

      [ Parent ]

      •  That's tricky (4+ / 0-)

        That would be some evidence, but not very strong evidence, that something is going on.

        Unfortunately, many effects in social science are like this

        Democrats want us to share a dream. Republicans want us to fear a nightmare. I'm for dreams.

        by plf515 on Fri Jan 20, 2012 at 03:40:23 PM PST

        [ Parent ]

        •  That's why computer models are so difficult to (4+ / 0-)
          Recommended by:
          Villanova Rhodes, G2geek, plf515, Orinoco

          program for sociology, psychology, etc. The complexity of the interrelationships is very difficult to commit to an algorithm. I was taking an upper level Soc course in the late  '90's, looking at Weber and Durkheim, mostly, but Weber really made a lot of sense to me. Thinking about him, it occurred to me that you needed something on the scale of a state-of-the-art weather modeling program to sort and handle all the variables. Something like the modeling that can predict weather behavior down to the microclimate level regionally  is something that should be able to handle the variables for psych and sociological models.

          Then all we have to do is figure out what factors are important and how they all relate to each other. /snark ;-)

          Information is abundant, wisdom is scarce. The Druid

          by FarWestGirl on Fri Jan 20, 2012 at 04:38:38 PM PST

          [ Parent ]

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site