Exercise 5d – Operationalizing Your Conceptual Definitions
In a previous exercise we developed the following conceptual definition for domestic violence;
Domestic violence – The actual or threatened infliction of physical injury or psychological trauma by a current or former household member onto another current or former household member.
How would you go about actually measuring whether this social phenomenon occurred or not? The answer really depends on the research methodology you choose to use while gathering your data. Here are two examples.
Secondary analysis
This research methodology gathers data from existing sources. For example, if you wanted to gather information on the crime rate over the past decade then you might use the Uniform Crime Reports published annually by the United States Federal Bureau of Investigation. Or, if you wanted to know something about the growth of the prison population over the past decade you might use the annual reports from the United States Bureau of Prisons or your state’s prison system.
The effectiveness of secondary analysis as a research methodology depends on two things. First, the manner in which the concept is measured within the existing data set must be similar to how you intend to measure it in your research project. For example, let’s say you’ve defined “youthful offenders” as “any convicted criminal offender between the ages of eight and thirteen years of age”. Not a bad conceptual definition. Unfortunately, the existing data set you intend to use defines “youthful offenders” as “any convicted criminal offender between the ages of nine and fourteen years of age”. Now you have a choice to make. You can change your conceptual definition to match that of the existing data set. You can collect a whole new data set that uses your own definition. Or, you can use the existing data set and report the difference between the definitions.
Second, the existing data set must have been collected in a manner similar to how you intend to collect data. If random selection is important to you then the existing data set should have been collected randomly. If your research requires a national survey then the existing data set should have been collected from a sample drawn from throughout the nation.
In our case let’s assume you choose to collect your sample of cases from official reports of domestic violence incidents made to the local police department. As long as the definition of domestic violence used by the police department (likely the legal definition) is similar to the conceptual definition you created then this would be an acceptable secondary data set for your research project. If not, then you’d have to keep looking for another data set or collect your own data.
The police data set might work fine for your domestic violence research, but remember that not all domestic violence incidents are reported to the police. In fact, a fair amount of them are not. If you choose to use the police data then you’d miss collecting data from research subjects who experienced domestic violence (per your definition) but did not report it to the police.
Survey or Interview
Surveys and interviews are common research methodologies among social scientists. Although very different in how they are administered, these research methodologies essentially require the researcher to ask research subjects to respond to a question, or more likely, a series of questions.
It would probably be a mistake to ask the survey respondents or interviewees, “Are you a victim of domestic violence?” You really don’t know how your respondents define the term domestic violence, but it is a safe bet that they don’t define it like you do. You could ask, “Have you ever been physically injured or psychology traumatized by someone you live with now or someone you lived with in the past?” A “yes” or “no” answer to that question would not be very useful. You may recognize this as a double barreled question. So, it might make sense to break up your conceptual definition into smaller parts and ask simpler questions.
Have you ever been physically injured by somebody you live now? (circle one - YES or NO)
Have you ever been physically injured by somebody you used to live with? (circle one - YES or NO)
Has anybody you live with now caused you psychological trauma? (circle one - YES or NO)
Has anybody you’ve lived with in the past caused you psychological trauma? (circle one - YES or NO)
Have you ever been threatened with physical injury by somebody you live with now? (circle one - YES or NO)
Have you ever been threatened with physical injury by somebody you used to live with? (circle one - YES or NO)
Has anybody you live with now threatened you with psychological trauma? (circle one - YES or NO)
Has anybody you’ve lived with in the past threatened you with psychological trauma? (circle one - YES or NO)
In general the manner in which you operationalize your concepts should insure that all, or most, of the key parts of the conceptual definition are measured. During the conceptualization process you identified the various dimensions or perspectives of your concept. During operationalization your task is to be sure you have a strategy for measuring each of these aspects.
It is normal for an operational measure, particularly in the social sciences, to be a bit lacking. Social concepts are often multi-faceted and difficult to measure. Just do the best you can but recognize where your operational definition might be lacking. It helps sometimes to seek guidance from a well informed peer just to be sure your operational definition covers all parts of your conceptual definition.