Policy, Research, Outreach: Environment, Food & Biosecurity
Home » Research » Actor-Network Theory » Using Actor-Network Theory in the Field
Using Actor-Network Theory in the Field

Getting Started

This research consisted of three phases.

  1. Semi-structured interviews were conducted with artisan bread, cheese, and jam producers and food safety inspectors from the Michigan Department of Agriculture and Rural Development (MDARD). In semi-structured interviews, the researcher uses a set of questions to organize the discussion, but not in a strict format or order. Unanticipated topics can be explored as they arise. The interview is more like a conversation.
  2. Field observations were conducted of bread and cheese production and food safety inspection. Issues that arose during interviews were explored in more detail.
  3. Two focus groups were conducted in order to discuss preliminary findings and provide a check of internal validity of findings made during the first two phases. Both focus groups involved artisans together with inspectors.

ANT was especially useful during field observations, and this section focuses on the observations.

“A researcher walks into a bakery.” Now what? What do you look for? How do you know what to look for?

These are questions of methodology. A methodology is a set of beliefs and priorities that determine what you look for in the field. This study’s methodology was based on ANT. Thus, as described in the previous section, it sought to suspend broad assumptions about artisanship, regulation, and the relationship between them and, instead, to closely observe what happened as people went about their routines.

Twelve cases were selected for the field observation phase of the study. Cases were selected based on three criteria:

  1. Artisans met the study’s working definition of artisanship by emphasizing handcraft production, engaging in each step of the production process, and producing at a small scale.
  2. The facility had an inspection scheduled within the period of the study.
  3. Both the artisan and his or her inspector were willing to participate in the study. MDARD provided me lists of facilities, facility contact information, and inspection due dates.

Data Collection

I first observed processing operations in order to familiarize myself with the various practices of the artisans who participated. Observations of production lasted between 2 and 4 hours and averaged 3 hours. The artisans explained their operations and their particular approaches. Next, I returned to each facility with its food safety inspector in order to observe the inspection. Inspections lasted between 1 and 3 hours, averaging 2 hours. Inspectors explained their practices to me as they worked, and in one-on-one conversations before or after inspections.

To return to the question of what I looked for during observations: I focused on the cast of humans and objects involved in artisanship and inspection. Below are the forms that I filled out during observation of production. Data are fictitious but representative. (As any cheesemaker will tell you, do not try to make cheese based on this information.)

Forms Used During Observations of Artisan Processing

I listed objects that artisans used. Objects are numbered as a reference for the other forms below. The numbers reflect the order that I jotted the objects down in. The following forms show cheesemaking:

prod objects

I listed production activities:

prod activities

I drew floor plans of production facilities:

prod floor plan


This may seem like a lot of minutiae to write down, and it was. But there were both methodological and practical reasons for doing so.

Methodologically, from an ANT perspective, all objects “count,” and we cannot assume that some objects are more important than others. ANT has a flattening effect; phenomena are ensemble productions, and all actors are equally important. If the cheesemaker picks up a whisk, the whisk is an actor, part of the enactment of cheesemaking, and you write it down.

Practically, it is very easy to forget what you’ve seen. So much of this looks obvious when you’re in the room. But once I got home I would forget. Of course there was a cheese vat; of course there were shelves and a sink. But what were they made of? How big were they? ANT helped structure a plan for recording all of this. As is discussed on the next page, these minutiae paid off by leading to unexpected results.


Observations of Food Safety Inspections

When I observed food safety inspections, I was interested in how the objects and activities of artisanship entered into discussion between the inspector and artisan. I aimed to organize my notes according to the issues that arose between artisans and inspectors, and how these issues concerned the objects and activities.

This is the form that I developed for observations of inspection:

inspection notes

In practice, however, it was difficult to organize notes using this form, and I stopped using it about halfway through the inspections. The column format didn’t allow for detailed and on-the-fly observations.

More important, this view of interactions wasn’t capturing what appeared to be going on during inspections. I began to get a sense for the rapports that artisans and inspectors built (or didn’t build) and the give-and-take that characterized many of their interactions. My notes became messier but more comprehensive and helped me explore these dynamics in a more open-ended way.

I continued to take the form along to remind myself what details to get. For example, it helped me realize how few objects inspectors work with: flashlights, thermometers, clipboards, hairnets, lab coats, sanitizing strips, and sometimes a laptop computer.

Comments on Using the Methodology

This methodological focus on objects and activities proved useful by generating unexpected insights into artisanship and inspection. These insights are detailed under How ANT Led to Unexpected Results.

But it’s equally important to note that a methodology can be useful when it does not work as expected (Law 2004). I ran into two main challenges in observing inspections.

First, the study’s focus on objects paradoxically foregrounded the importance of interpersonal rapport. Although my notetaking focused on physical objects, I found that many interactions had nothing to do with physical objects. The trust or distrust that inspectors and artisans had developed over time, the ways in which each of them handled disagreements, their conversations about non-regulatory matters—all of these figured in the interaction between artisanship and regulation. How inspectors interpreted what they saw was influenced by their interaction with an artisan and by their previous experience of him or her. Whether an artisan adopted or challenged the inspector’s directives was influenced by a sense of fairness and by his or her trust of the inspector. Interactions were shaped by the interpersonal relationship formed between artisans and inspectors as well as by the objects (such as equipment and ingredients) that I noted.

Second, as mentioned above, the data collection instrument developed to characterize interpersonal interaction proved inadequate. This problem was actually helpful because it dramatized how nuanced the interactions were. Based on other studies of regulatory implementation, I had planned to characterize interactions between artisans and inspectors as corrective, instructive, clarifying, or conversational. Yet interactions could rarely be characterized according to only one type; in fact, many of them fell into all four categories. When correcting an artisan’s practices, inspectors also usually instructed. Artisans and inspectors conversed about the correction, and they clarified details as needed. The tone was almost always pleasant and relaxed.

Research results continue in the next section.