Joseph Nathan Cohen

Department of Sociology, CUNY Queens College, New York, NY

Structure in Measurement Strategies

To see more about the “Teaching through YouTube” series.  For more, see this post.

Original Video Description

Please pardon the editing errors and uneven volume. It is fine for my 334 students in the interim, but I will repost a reedited version shortly. This video talks about the concept of structure, and about what you might consider when deciding how much structure to put in your research design.

Transcription (Auto-Generated)

when you conduct research, you face conflicting desires. One conflict is the desire to understand a phenomenon in depth and capture all of its complexity versus the desire to reduce things to make them simple, testable, and usable. Another conflicting desire is our yearning to discover new things that we haven’t thought of before. This conflicts with the drive to confirm what we already know and to test or falsify. Both of these tensions in research play out in measurement design when we ask questions about structure. In this video, I’ll describe the choice between structured and unstructured measurements. I’ll explain their strengths and weaknesses and encourage you to think about the amount of structure you’re going to use in your study. Your data collection is structured to the extent that you’ve predefined what you’re looking for and are merely quantifying its presence or absence. For instance, a structured survey question might ask the respondent to quantify their attitude on a one-to-five scale. Another example is when you give the respondent a set of choices reflecting their political party preference. Structure doesn’t mean people have to make hard commitments. For example, you can ask respondents to distribute their support among multiple political parties, but their choices are still predetermined. The article that first captivated me in sociology was a structured observational study. This research looked at non-verbal female courtship behaviors. The study observed different non-verbal behaviors in women in various settings, from singles bars to university snack bars, libraries, and women’s center meetings. The results suggest that females display a set of non-verbal behaviors when in a courtship situation. Unstructured data is different. When collecting unstructured data, you don’t pigeonhole subjects into categories or predetermined behaviors. You aim to limit your preconceptions and see what information emerges. Unstructured data is most valuable when trying to uncover unexpected information or when you have little prior information. Ethnographic research is often unstructured, as are focus groups used in marketing research. Unstructured questions can also be survey-based, like open-ended questions asking for class strengths and weaknesses. When you design a research project, you should consciously consider the amount of structure in your measurements. It’s possible to blend structured and unstructured elements; it’s not an either-or choice. Benefits of structured data collection include efficiency, replicability, and the control of observer bias. Structured questions are quicker to answer, and the results can be tallied by a computer. The strict observation rules in structured data collection ensure replicability, and observer bias is controlled since observers must follow specific protocols. However, structured data doesn’t eliminate all biases. The structure of questions might incorporate the biases of the person who wrote the survey. Benefits of unstructured data collection include producing deep understandings of your subjects and being more open to unexpected findings. The problem with structured data collection is…