Clay Spinuzzi, University of Texas at Austin
At the meso level, actions are
So far, we've examined actions using two analytical models:
Think of your research site as a football game. In CEMs, the camera follows the ball, tracing through the series of handoffs and tosses that move it downfield. In GEMs, the camera follows the game, watching systemic dynamics and tactical changes as players and artifacts all over the field continuously reconfigure themselves.
Can we follow the ball and the game? Can we find a way to coordinate these two models? Sure: That's what we use sociotechnical graphs (STGs) to do. In the process, they allow us totriangulate data: to compare the stories that we get from different sets of data in order to make sure they agree.
Sociotechnical graphs (STGs) come from work done by sociologist Bruno Latour and collaborators. Latour and his collaborators compared participants' statements about technology, looking at these statements along two dimensions: syntagmatic and paradigmatic. (Think of these as "AND" and "OR" dimensions.) By graphing the statements along these dimensions, they could examine how participants chained together elements in their statements, and they could also see how different participants might substitute elements in these chains.
In our implementation, STGs are tables in which
For instance, here's an STG in which the researcher examines a single participant's work, using two different data sources (field notes from the observation and the post-observation interview). Notice that this STG helps us relate communicative events from the CEM to the genres from the GEM. And it helps us to see how our field notes differ from the participant's account. It helps us to triangulate the two data sources: we can see how closely they line up and where they disagree or are partial. See the italics in each cell: these are texts that are mentioned in just one account.
Table 1. An STG for a single participant.
Now suppose we do the same thing for multiple participants. We can collapse the list of texts from both data sources into one list for each participant, then compare the participants -- and we begin to turn up similarities and differences in how individual participants work. Triangulating at this level helps us to do the following:
For instance, in Table 2, Clara uses a text that the others don't use: a log of previous customer interactions. Does this log function as a substitute for some of the texts that others use, such as Arnold's spiral notebook? The STG helps us to spot differences and reexamine our data - including our copies of the spiral notebook and the log - to answer questions about how participants work differently. Through this triangulation, the STG helps us to catch innovations and workarounds, showing how these substitute for other texts.
Table 2. An STG for multiple participants.
If the organization is large enough, you may triangulate to spot differences in how groups do their work. Groups can be
For instance, if a company has two offices, it's common for the offices to develop different ways of doing things due to different technologies, training, backgrounds, expectations, or innovations. A group-level STG can help you spot those differences as well. Table 3 shows how two offices might handle the same communicative events differently - and how the second office has managed to use one text to substitute for many.
Table 3. An STG for different groups.
To build STGs, you'll determine what you want to compare or triangulate - data sources for one participant, accounts for individual participants, or accounts for groups - and use your GEMs and CEMs to build the table.
First, decide what you want to compare.
Your comparison categories will provide your row headings: the cells in the left column. See Tables 1-3.
Next, choose events to examine. The events will come directly from your CEMs, especially events that are often repeated. These events become the column headings: the table cells in the top row. See Tables 1-3.
Now you can methodically compare the texts used by each comparison point. Consulting your GEMs, fill in each of the remaining cells with genres that are involved in each communicative event for each comparison point. See Tables 1-3.
Make sure that each text you mention is backed up with at least one piece of evidence, preferably more. For instance, see the texts listed in Table 2. If challenged, you should be able to demonstrate from your field notes, interviews, and/or other research data that each of those texts was used by that participant to complete that communicative event. Don't guess, don't assume - prove.
Finally, we are ready to examine differences across the STG. The previous steps have helped you to join your GEMs and CEMs, yielding a triangulated, confirmable, systematically generated map of texts used to mediate events. Now you should be able to easily detect core associated texts that everyone uses as well as substitutions that allow individuals and groups to work differently. Such substitutions often include innovations and workarounds.
Highlight these differences, perhaps using italics (see Tables 1-3) or a highlighter.
At the end of the process, you should have one or more STGs that look something like Tables 1-3.
Finally, recall that you used CEMs and GEMs to detect discoordinations, points at which two or more texts just don't "fit."
Use STGs to map these discoordinations in the appropriate cells. Doing this may reveal a lot about where problems occur and what innovations have reduced them.
For instance, suppose you review Table 2, mapping discoordinations from the GEMs and CEMs of individual participants. Every time you notice a discoordination in your GEMs and CEMs, you put a red tally mark in the corresponding cell of the STG. When you're done, you notice that Arnold and Bill experience multiple discoordinations when preparing for a call - but Clara only encounters one. That suggests that Clara has found some sort of solution or workaround for the discoordination - an innovation that could provide a starting point for redesigning the work.
What could that innovation be? We can easily compare the texts used by the different participants, so we see at a glance that only Clara uses a log. Is that the difference? Go back to the primary data to confirm.