When I became an Assistant professor, it was important to me to begin a somewhat independent steam of research. My graduate school projects were typically large, time consuming, and long-term. My hope was that I could design and collect data on research ideas I had been kicking around by myself. This was successful to a certain extent as I conducted several projects that ended up getting re-incorporated into one of my thesis papers. For that project, I wanted to extend some of my own work and that of a friend’s (Jerry Guo) by investigating perceptions of networks. At UMass Dartmouth, I’ve collected a variety of survey and group-based lab data. Some of these data collection efforts were restricted by my own time pressures and a few were limited by the COVID-19 pandemic.
In this Summer blog series, I hope to write up a few of these projects that either aren’t substantial enough for a paper or that I doubt I’ll return to soon. In addition to that, I’m going to use these blog posts as a way to stretch my R and Shiny chops, I’ll include some this accompaniment when it makes sense. The first project I’ll describe was from before I became a professor but it kicked off a few other things (and is one of the smallest).
Fall 2015 – Discussions about Network Preferences
In the Fall of 2015, I was in a bit of a difficult spot. The task modality I used for 2 of the studies that would end up in my dissertation was Yahoo! Pipes. Though an interesting piece of software intended to allow for non-programmers to create widgets, it was aging when we chose to use it and was announced to be shutting down shortly before or after I began data collection. That meant that I knew I had a deadline to hit that would dictate how many groups I could collect. I spent a lot of time that Spring / Summer recruiting for the study and running as many sessions as I could.
Around this time, our research group had gotten interested in taking the ideas from some of my dissertation work (see Argote, Aven, Kush, 2018 or my dissertation). In those studies, we put groups in a communication network that dictated who could talk to who. In ‘real’ groups, however, people ‘can’ generally talk to anyone, but they choose to talk to only some people. We call these emergent networks. We had already seen emergent networks in our 2018 study since, in one condition, we let everyone talk to everyone else. One of our key findings was that sometimes, even if everyone can talk to everyone else, sometimes they don’t. We wanted to try and take our assigned networks and make them a bit more ‘realistic.’ So we decided to start small, what if we keep the network as fixed (only certain people can talk to each other) but we give the group some choices about what network they got. Thus we could compare groups who worked in a really centralized network and got it randomly vs. chose it. There’s a lot of complexity here and alternatives (issues of agency and even the limited endogeneity we introduced), but we thought it seemed like a good challenge to go after.
As part of this, I carved out a few of the final sessions of my group study with Yahoo! Pipes and ran a modified protocol where I gave participants some choices. In the standard study, groups communicated in one of 4 structures. In this version, we started with everyone being able to talk, and then asked them which of the 4 structures they wanted to work in. I didn’t have time to actually have them work in those structures, but it gave us some ideas about how people would think about and respond to this situation.
Four of the groups chose what we called the star structure, where one person is connected to everyone else. They typically stated that there reasons for choosing this structure were related to coordination efficiency. That makes some sense as it looks more like an organizational chart that we are all familiar with, a “leader and three subordinates” as one of the participants stated. Interestingly, in the regular studies, this group is far from the best structure.
One group chose the best performing structure, the circle (each member has 2 connections with the person on their left and right, forming a ring). This group, however, also discussed everything in person whereas the other 4 groups only spoke over instant messenger. There could have been some societal pressures of course not to lead to someone being ‘left out,’ but I did find this intriguing, and it sparked a further investigation at UMass.