mario small discusses research method

Mario Small with the Hoosier Protection Service

Last Friday, orghead Mario Small gave a talk at the School of Public and Environmental Affairs, jointly sponsored by sociology. He gave a wonderful talk on his new research on social capital in organizational settings. I won’t go into that, you should read the book.

I will mention one valuable exchange. One of our post-doctoral visitors asked about his field site. Do the lessons from child care centers and tie formation carry over to other settings? Mario roughly said the following:

When thinking about field research, we often confuse the place of observation with the unit of analysis. In other words, we observed interactions, not child care centers. If we observe interactions in other places that also have similar characteristics, it is fair to ask if we observe the same outcomes. In other studies of similar interactions, there is some support for my main ideas.

Good point. Also check out Mario’s blog.

Written by fabiorojas

October 26, 2010 at 12:09 am

Posted in fabio, networks, sociology

6 Responses

Subscribe to comments with RSS.

  1. Interesting insight and heads off the follow-up question, “but what if the character of the interaction is a product of the place of observation?”

    Also, I think it may have been posted here before, but that turfing video on his blog is blowing my mind.



    October 26, 2010 at 4:18 pm

  2. Trey, why should one want to avoid such a follow-up question. Yes, the character of interactions at one place is non-trivial evidence about interactions in other places, especially if we have reason to believe the places are similar. But we should actively address how the place of observation might matter, not seek to distract people from thinking about it.


    Michael Bishop

    October 26, 2010 at 8:43 pm

  3. I think what Mario was trying to communicate was that place of observation is the wrong way to think about this sort of research. Yes, I think place will definitely matter. Mario brought up the example of medical clinics. People are so distressed by health problems that they probably wouldn’t form a tie no matter what.

    The point is that place can have a huge priming effect, but place isn’t the main variable. It’s interaction that’s being looked at. So we think of a range of priming effects. Medical clinics likely represent an extreme case. Child care centers probably in the middle. Grocery stores on the low end.

    If you start with place, then you get stuck. You need a theory of place, which is hard and distracting. By switching to interaction as the unit of analysis, then you can more easily think about the problem.

    And of course, if Mario is reading this and he thinks I go it wrong, do add your opinion!



    October 26, 2010 at 8:57 pm

  4. Fabio’s comments encapsulate what I meant. Often in single-site research, the gotcha question is “is this generalizable?” When, as pointed out, the unit of analysis is the interaction rather than the setting. Indeed, the burden is on the question-asker to demonstrate why it wouldn’t generalize than on the researcher to demonstrate that it does, in my view.



    October 26, 2010 at 9:14 pm

  5. Yes, I think this question is important, pervasive, and treacherous. Childcare centers are so unique that the question jumps to the fore, but the same would be true if studying churches, firms, or any other single type of organization: on what basis can we make inferences about what is happening in other organizations?

    I think all too often researchers make the mistake of seeking to specify the broad organizational types that the case purportedly represents (‘well, it should work in service organizations,’ or in ‘for-profits,’ or in ‘publicly-funded entities,’ etc.). That answer is never convincing because it relies on converting the case study into an n=1 sample of some larger population of organizations of the same type. It is always easy to find ways that other organizations in the purported type would differ substantially from the case (e.g., think of all the ways that other ‘service organizations,’ say, differ from childcare centers).

    We can avoid the tortured logic in which some authors find themselves by abandoning that particular quest all together. Yes, childcare centers are unique in several ways, including those that made the study possible. Instead, by specifying how institutional incentives structured social interaction—and how particular forms of interaction, in turn, shaped social ties—I directed attention toward the thing that this kind of case study is good for, understanding processes. On this basis, I would expect that particular kinds of social ties (e.g., compartmental intimates) should be evident in any ‘type’ of organization (for-profits, firms, churches, whatever), provided social interactions take the form I suggest (e.g, if they are strongly task-oriented), which, in turn, would depend on the presence of core institutional incentives. This approach is very consistent with Hedstrom and others’ call for a focus on mechanisms, which in this case ends up being a way to avoid the all-too-common and pointless pursuit of ‘representative’ case studies.




    October 26, 2010 at 11:27 pm

  6. Trey—on the video, I agree! Pretty sublime, no?



    October 26, 2010 at 11:29 pm

Comments are closed.

%d bloggers like this: