Goodman

Coal miner study

One part of my graduate education that I count as one of the most fortunate was the limited amount of experience I had interacting with Professor Paul Goodman. Paul Goodman unfortunately passed away shortly after I passed my qualifying exams. Paul Goodman was an extremely interesting and committed researcher that allowed his personal feelings of justice influence the direction of his work in a very real way without allowing them to cloud the scientific process. Paul was truely one of a kind.

After Paul passed away, I spent some time talking to his wife and children as they discussed his upbringing and what motivated some of this work. From what I recall, both of his parents were liberal social activists in New England. From an early age they instilled in Paul that organizations have a responsibility to treat their employees well. Though I'm sure many other things influenced his choice of career, Paul eventually began studying the ways that employees interact with management in organizations. Paul was an avid film-maker who did a series of videos about the current state and future of work. He typically interviewed average people in industries that were changing. Many of these films can be found at a permanent collection at Carnegie Mellon's library website: http://dli.library.cmu.edu/paulgoodman/

The last two projects that I know of Paul perusing was a long-term project on the science of science teams. Though I do not know his specific motivation, scientists often apply much less social science to their organization than what we actually know. After Paul died, this project dissolved due to the cohesive power of Paul's personality disappearing. The other was a more amorphous process that I think perfectly sums up Paul's outlook on the world. He and his assistants conducted hundreds of long-form interviews asking average people what they thought the American dream was, if they strove for it, and what kind of world they wanted for their children.

Though Paul completed a lot of interesting work, what I'd like to talk about today was some of the work that came out of his multi-year coal-mining project. In this work, Paul went to coal mines in the mid-Atlantic and interviewed miners in their place of work. By that I mean underground in the mine itself. Paul told me on multiple occasions that he thought that the ability to conduct the interviews and collect data in the mine itself gave him a much more accurate perception of what it was like to work in this environment. My father, who is from Pennsylvania described to me when I was very young that my great grandfather's worked in a coal mine. This profession and the work that Paul did therefore always seemed to touch me a bit closer as I always imagined my great grandfather in the place of the miners in the papers.

The paper I would like to describe of Paul's is one that he wrote with Dennis Leyden. This work was supported by the U.S. Bureau of Mines. I can only make a guess but I think that the Bureau were interested in how the relationships between the individual workers in the mine were related to mine outcomes. Mines vary in productivity enormously and one possible reason is the kinds of relationships the individual workers have with one another.  Goodman and Leyden proposed that the mines provided a good opportunity to look at the effects of familiarity on the small teams that work together within a coal mine. (In a prior study, the researchers had already identified that an individual with little familiarity with a mine was more likely to have an accident.)

Mining crews were sets of workers doing one of three unique roles. Those roles were: the miner operator, the bolter, and the car operator. Each crew typically had a pair of people performing each role. Though each role is unique and there are skills associated with the roles, the authors argue that the specific strategies the individuals use vary from crew to crew based on personal differences and the features of the part of the mine the group is in. Though the researchers do not mention it specifically in this paper, another factor that I imagine is important is the cognitive interdependence of the individuals on one another.

Without getting into too much analytical detail, the researchers used information about which groups individual were working on to create a measure of whether individuals had worked with one another before and to what extent a given group's members were familiar with one another. Overall, the researchers found that the levels of familiarity between the group members was predictive of the overall mine productivity. They found some evidence that different kinds of familiarity mattered more than others but they felt that overall familiarity mattered more.

In rereading this paper, I found myself reminded of some other interesting work by Karl Weick on aircrews. Like I am attempting to do in this blog, Weick preferred description over analytics and wrote extremely provoking papers based on his reading and observations of real events. Weick's observation of air crews found very similar effects of familiarity on the air crews ability to perform without errors. I'm sure I will discuss some of his other work later in this blog.

Coginitive Interdependence [Deep Dive] - Moreland series - Part 3

This post describes the studies commissioned by the Army that Levine et al. explored. In Part 2, the studies on productivity were explored. In this post, I focus on the experiments about innovation.

Creativity Experiment 1 -Assigned and/or maligned (published as Choi & Levine, 2004)

The researchers then shifted away from performance as the primary variable of interest and into the effect of turnover on group innovation. These studies used a air-surveillance task that John Levine and his students have used in several papers that I know about. Groups work together to monitor the radar at a base and assign threat levels to the different radar contacts. In each of the three member groups, 2 individuals were specialists and 1 acted as the commander. The specialists essentially collect information about the radar contacts and the commander receives that information and is tasked with making a decision. There were two different strategies that could be used in the strategies to collect information that varied on whether the importance of the information the specialists collected was the same for both or whether the difficulty of getting the information was the same for both.

In the first experiment, the researchers manipulated whether the group was able to choose their strategy and how well their feedback suggested that they had performed. In the experimental setup, the group was either assigned one of the two strategies above or they were allowed to choose one. Then the group performed the task. Half the groups were told that they had performed well and the other half were told that they had performed below a passing rate. One of the specialists was then chosen and replaced with a confederate. In social psych research, a confederate is someone who pretends to be a normal participant but has been coached to act in a particular way. The newcomer then proposed that the group switch to the opposite strategy of whichever they had chosen in the first trial.

The researchers used whether the group accepted or rejected the strategy the newcomer proposed as the variable of interest. Because this could be affected by a multitude of factors, the researchers measured how committed the members were to the previous strategy, how much they liked the team, performance in the first trial, etc. The researchers found results that were inline with what they anticipated. If groups were told they failed to perform well in the first trial, they were more willing to accept the newcomers idea. The groups were also more likely to accept the idea of the newcomer if the group had not been allowed to choose their own strategy.

The researchers then did some additional analyses and proposed what led to the group's receptivity to the newcomer's proposal. The two variables the researchers proposed mediate the effect of team choice on the acceptance of the newcomer: commitment and perceived performance. If the group had a choice in their strategy, they were more committed to their strategy and they perceived their performance as better.  The researchers were fairly satisfied in these findings but they also thought that the way the newcomer proposed their innovative idea likely would have an effect on whether the group accepted it. This led to the second experiment.

Creativity Experiment 2 -An Assertive Story

This study was run very similarly to the first creativity study except that the kind of language the newcomer used was varied. As before, groups are more likely to accept the newcomer's innovation when the group was told that they had failed in the first trial. There was also what is called a statistically significant interaction. An interaction just means that whether one variable has an influence depends on another variable. When the groups were told that they had succeeded, it did not matter whether the newcomer was assertive or not, the acceptance rate was always about 45%. If the group had been told they failed, however, they were more likely to accept the ideas of the newcomer if the newcomer was assertive (~85%) versus if the newcomer was not assertive (~60%). [Note: this effect is only 'marginally significant' meaning that our confidence in the effect is not overly high.] The researchers had hoped for stronger effects but still thought this study was valuable.

Computational simulations, the shallowest dive

The last part of the technical report provides some information about a series of computational simulations that were included in this project. Very briefly, a computational simulation puts a bunch of agents into a box. Each agent represents a person, organization, etc. The agents are given some rules to live by, some of which may vary systematically (share information with another agent if they are within 2 spaces vs. share information with another agent if they occupy the same space). There is also a level of randomness that is added to the agents decisions to help simulate the real world. Simulations are becoming more and more accepted within management-type research though I am not sure how accepted they are within general social psychology.

In the series of simulations presented in the report, the authors focus on the effect of transactive memory and changes in the environment. In the first simulation, the researchers find some evidence that suggsts that the value of a transactive memory is curvilinear with the size of the group.  They found that if the group is fairly small, the difference in speed to completion of a task by the agents was about the same regardless of whether the group had a transactive memory or not. There was a definite benefit of TMS when groups were larger (between 15 and 27), but the benefit reduced for larger groups (35). I personally think that this is an artifact of how the agent's task is structured, but it does seem fairly reasonable. The last simulation suggested that transactive memory is particularly useful if the group completes multiple different kinds of tasks that are completed in alternating order. A transactive memory allows the groups to more quickly shift tasks, leading to a consistency in time to completion.

Though the studies in this report were not all successful, I found it particularly interesting. The ability to try out new ideas that this study provided also certainly helped the researchers develop their later studies and directed other researchers toward these topics.

I think this post completes my sequence on cognitive interdependence for now, though I'm sure it will crop back up :P

Coginitive Interdependence [Deep Dive] - Moreland series - Part 2

The US Army funded work by four social scientists in Pittsburgh all centered around the influence of turnover on small groups. Work groups in the Army often experience member turnover for a variety of reasons (e.g. transfer, injury, death, etc.) which makes their interest in this area very understandable. In this post, I hope to walk through some of the studies that the Army funded. As far as I know, only one of the studies in this set has been published in an academic journal. There was, however, a technical report given to the Army that I will be basing the information from this post on. This report can be found here: http://oai.dtic.mil/oai/oai?verb=getRecord&metadataPrefix=html&identifier=ADA433897

On the project were 4 primary researchers: John Levine, Dick Moreland, Linda Argote, and Kathleen Carley. Dick and Linda were mentioned in prior posts on cognitive interdependence from their extremely important experiments. John Levine was a frequent collaborator with Dick and Linda who is also interested in group behavior. Kathleen Carley is a somewhat different kind of researcher, specializing in computational simulations. In computational simulations, researchers create a set of rules for a world and then see what the outcomes of the world are once the actors in the world interact for a while. The rules in the world can then be adjusted to see if the actors behave much differently or if the outcomes are different. From the abstract of the study, we can see that the researchers intended to gain insight into how personnel turnover impacted groups completing different kinds of tasks. The variety of the researchers also allowed the use of laboratory and simulation-based approaches. Due to two of the researcher's prior investment in the concept of transactive memory, this was included as a component in these studies. Indeed, the lab studies that these researchers competed were a direct extension of those studies.

Productivity Experiment 1 - Turnover and Rumors of Turnover

In the first study, groups of 3 were trained together on a construction task (it isn't made completely clear but I believe it was the radio assembly task used in Liang et al., 1995). There were two manipulations: the groups were warned that there would be turnover (or not) and groups experienced turnover (or not). The warning occurred before the group trained together and the turnover occurred at the beginning of the second performance session. The researchers measured transactive memory and two measured of performance: whether the group could recall the task without having access to the circuit and assembly errors. The results for this first study, in the words of the researchers "were difficult to interpret".

When groups didn't actually experience turnover, they recalled more of the task if they were told that they were going to experience turnover. This makes sense because the group members may have tried more to individually memorize how to do the task if they knew that they couldn't rely on each other. For groups that experienced turnover, however, groups that did not expect turnover recalled more of the task than those that did expect turnover. As for errors, if groups experienced turnover, they performed much better, regardless of whether they were warned that there could be turnover. The researchers made a guess that the newcomers may have just tried really hard, which could explain the effects with errors. In future studies, they made sure to limit the newcomers training harder than the other members.

Productivity Experiment 2 - Turnover and Expertise Information

In this study, all groups were trained together on the task. In the control condition, the group was not warned of turnover and there was no turnover. In the second condition, turnover occurred without warning. In the other three conditions, the groups were warned there would be turnover and then given information about the newcomer's skills. The conditions varied on who received the information, just oldtimers, just newcomers, or both. The researchers measured transactive memory and errors.

As expected, groups that didn't experience turnover made fewer errors than those that experienced unexpected turnover. Groups in the other three conditions where someone received information about the newcomer, all made fewer mean errors than the groups that unexpectedly experienced turnover. Groups where the oldtimers received information about the newcomer made the same number of errors as groups that didn't experience turnover. Interestingly, when the information only went to the newcomer or to both newcomers and oldtimers, groups made slightly more errors. The researchers found nearly mirror results for transactive memory. Groups that didn't experience turnover had the highest transactive memory and groups were oldtimers received information had similarly high levels of TMS.

The researchers then shifted into looking at the effects of turnover on innovation. These studies will be considered next.