Seminar

Cognition & Tie Persistence in Collaborative Networks

Date
Wed November 20th 2013, 12:30pm
Event Sponsor
Stanford Institute for Research in the Social Sciences & Stanford Sociology Department
Cognition & Tie Persistence in Collaborative Networks

Amir GoldbergDan McFarland, & Sameer Srivastava

This week we are joined by Amir Goldberg, Dan McFarland, and Sameer Srivastava; we will touch on a number of frontiers in network studies: culture, cognition, and network dynamics over time.

Our faculty panel last week with Matt Jackson and Sharique Hasan began with a discussion of diffusion and influence mechanisms and ended with a broader discussion about the importance of creative data and sound design. A more comprehensive summary last week's discussion (November 13) can be found below. Last week a few of us also met to discuss the schedule for next quarter. Reflecting back on our discussions this quarter, it was proposed that we focus this coming winter's panels on research design instead of methods, as we had originally planned. We are thinking about organizing faculty panels around the following four topics: What Constitutes a Good Network Model? A Survey of Causal Epistemologies across the Disciplines Designing Network Experiments (actual & simulated) Network Ethnography Designing Observational/Correlation Analyses A blog entry will be created for winter planning and I invite you to visit there and share any thoughts you may have about this.

McFarland's article

Srivastava's article


Questions for Srivastava & Banaji, 2011

1. Not so surprisingly, implicit self-assessments predict that individual's behavior.  More surprisingly, implicit measures are also predictive of how others treat us, whereas explicit measures are not.

    a. Is this the "actions speak louder than words" principle?
    b. Precisely what actions (mechanisms) effect other's views of us as desirable collaborators?
    c. Do explicit self-assessments ever matter?
    d. Might targets' explicit self-assessments be more predictive in an environment with lower levels of inter-personal interaction?

 

2. In models predicting colleagues enlisted in collaboration, or in target characteristics of network models, do you view target ECS/ICS as proxies for how initiators perceive the targets?  I realize this would have been a lot more work, but let's suppose you measured every employee's view of every other employee in terms of collaborative-independent.
    a. Would this be more or less predictive?
    b. How correlated do you think initiator assessments of targets would be with target ECS/ICS?
    c. Coming back to the mechanism question, what effect might third-party assessments of the target have on initiator beliefs about the target.
    d. To what extent are dyadic collaborative relationships influenced by peer-effects, externalities, the work environment?

 

3. Turning our attention to the environment, you studied highly collaborative environments and only modelled collaboration across department (horizontal) and salary grade (vertical) boundaries.
    a. Why not also study within department, within salary grade collaboration and compare it to boundary-crossing collaboration?
    b. How might within-group cognitive processes differ from between-group processes?

 

4. In some sense, it makes sense to study collaboration in a highly collaborative environment, in order to guarantee that you will observe it.  On the other hand, in environments where collaborative norms or expectations are high, I have to think that collaborative relationships will form regardless of ECS/ICS.
    a. Would you expect effect sizes to be smaller or larger in less collaborative environments?
    b. In what kind of contexts or environments do you think your findings are most applicable?
    c. In what kind of contexts or environments will your methods be useful? Less useful?
    d. How might we theorize and model differential cognitive effects from more dense, interactive social contexts to less dense, less interactive contexts?  Do you have a mental model for this in your mind?

 

5. Methods.
    a. Why ERGMs?
    b. How do you feel about actor-driven network models, strategic network models?
    c. What are the strengths or short-comings of these for modeling cognition within a network?

 

6. At the end of your article, you listed a number of possible directions.  What are your thoughts about these, and which are you most excited to follow up on?
    a. status
    b. mechanisms - How do initiators detect target dispositions?
    c. hierarchy
    d. multiple identities, multiplex networks

 

7. Besides ITA, how else might we measure cognition? Directly? Indirectly?

 

8. How do you think about cognition and beliefs over time?
    a. . Perhaps initiators learn through repeated interaction with targets, or third parties.
    b. How robust are these networks over time (perhaps we should direct this question to Dan...)?
    c. How robust are individual collaborative-independent identities over time?
    d. How might we model cognition and perception over time? Learning models?

 

9. This paper focused a lot more on cognition than culture.

a. What does culture mean to you?

b. What are different ways of operationalizing culture?

Questions for Dahlander & McFarland, 2013

 

1. Your paper distinguishes between short-term information gathering, long-term strategies and assessments of a relationship's worth, akin to Dan wang's study of ning.com, Podolny and Baron's resources and relations, Uzzi's paradox of embeddedness.

a. Resource and information gathering requires spanning domains, occupying structural holes.  Identity maintenance is advantaged by centrality.  Might these two orientations be at odds with each other?

b. Might these orientations also change over the life-course of a scholar?  networking, conferences earlier in your career; program building, amassing resources for others and managing collaborations later

 

2. How might we measure or describe these long-term relationships more fully?

 

3. When making connections or maintaining long-term relationships, what works, what doesn't (mechanisms)?  You mention features abstractly and measure some explicitly (e.g. same center).  More thoughts on this?

a. For making connections? Conferences? Mixers? Coupa? Workshops?

b. For maintaining long-term relationships? Centers?

 

4. Why has prior research focused more on tie formation than tie maintenance?

a. Network research diffusing over from EE, CS, and the substantive tie to information and resource networks?

b. Easier to model one-off exchanges versus a series of exchanges over a long period of time?

 

5. What calculations go into tie persistence in a multiplex context?

a. We can imagine a whole host of positive incentives and negative constraints, first-order and third-order.  Do you have a sense of which of these matters more and when?

b. How might we incorporate organizational and cultural expectations or differences into the analysis of multiplex ties?

 

6. What kinds of ties were captured in this study, and what kinds of ties were not captured, but still may be important to facilitating collaborative relationships?

a. Third-party catalysts, institutional or individual?

b. What other kinds of biases might exist from unobserved parts of the network?

c. If you had been able to observe one additional measure or type type of information, what would it have been?

 

7. How might these findings inform a more qualitative, perhaps ethnographic study of these same phenomena?

a. Are there things you would focus on or ignore, early on?

b. What would be the added value, drawbacks, convenience or limitations to such a study?