Understanding "Unit of Analysis"

I had quite a hard time understanding the entire concept of "unit of analysis" when it was first being introduced by Kerry, but gradually began to grasp the concept when Kerry kept prompting me to change my unit of analysis for my research project on self-efficacy beliefs. Kerry's intention of wanting me to change my unit of analysis wasn't because he felt that the individual level of analysis was inappropriate for the purpose of my research, but he wanted me to understand what the unit of analysis is, and how I could go about using it to analyse a journal article, research, or theory.

Units of analysis are issues we examine to account for differences among separate entities.  The biggest misconception that most people have regarding the unit of analysis is that it refers to the level of observation in a study. I fell into this trap as well. When Kerry mentioned that the unit of analysis could be at the individual, group, organisational, institutional, or societal level, I immediately jumped to the conclusion that the unit of analysis refers to what the researcher is studying. It is not the sample that you're studying which determines your unit of analysis, it is the analysis that you do in your study that determines what the unit is. If one were to liken the unit of analysis to the unit of observation/sampling, then there would only be one "unit of analysis" throughout the study. However, in qualitative research, it is possible for a study to have different units of analysis. In fact, researchers can position their research in different ways based on different units of analysis. For example, I might decide to position my analysis based on individual self-efficacy beliefs. In this case, the unit of analysis is at the individual level. However, when analysing my collected data, I might decide to compare the average level of self-efficacy. In this case, the unit of analysis is at the group level.

In varying the unit of analysis, Kerry urged us to be consistent and anticipate what conclusions we wish to make with regard to each unit of analysis. He has seen many examples of research where statements of a particular unit of analysis are actually based on another unit of analysis. Essentially, in exploring the different units of analysis, we might risk drawing invalid conclusions where the units of analysis do not match.