Variables in Sociological Research

Abstract

One of the key factors in developing and testing scientific theories is the identification and operational definition of the variables that describe phenomena observed in the real world. Variables are measured in a research study, and they can have more than one value. There are several types of variables of interest to the researcher. Independent variables are stimuli that are manipulated in order to determine their effect on the value of dependent variables. Extraneous variables are variables that affect the value of the dependent variable but that are not related to the question under investigation in the study. Intervening variables are variables that occur between the manipulation of the independent variable and the measurement of the dependent variable and that contaminate the relationship between the two. In order to be of use in scientific research, variables need to be operationally defined so that they can be measured and their effects analyzed.

Overview

Sociologists attempt to make sense out the world by observing the behavior of people within society, developing theories to explain this behavior, translating their theories into working hypotheses that can be tested, and conducting empirical research to test whether or not their theories are supported. Based on the results of the research, they then either accept or revise their theories in a continuing attempt to explain the world around them. One of the key factors in this process is the identification and operational definition of variables -- traits, characteristics, or other measurable factors that can have different values -- that impact the phenomenon of interest.

One is primarily interested in two types of variables: independent variables and dependent variables. The independent variable is the variable that is being manipulated by the researcher. For example, Dr. Harvey has a theory that the way that people dress affects how they are treated by others in the workplace. He believes that if people dress as if they are successful professionals (e.g., well-groomed, business attire), they will be treated that way and receive a disproportionately high percentage of raises, promotions, high performance appraisals, and other recognition. The independent variable in this theory is whether or not people dress like successful professionals. This is the variable that Dr. Harvey will manipulate in his research study to determine how it affects the way that people are treated in the workplace. The second major variable of interest to researchers is the dependent variable. The dependent variable (so called because its value depends on which level of the independent variable the subject received) is the response to the independent variable. In Dr. Harvey's research study, the dependent variable is the way that people are treated in the workplace. Dr. Harvey's theory is that the value of this variable (i.e., whether or not people receive recognition in the workplace) is dependent on how they dress.

Concepts such as "dressing as if one is a successful professional" and "how one is treated in the workplace," however, are rather nebulous and open to different interpretations. To one person, "professional attire" may be a power suit with white shirt and tie while to another it may be a clean polo shirt with Bermuda shorts rather than cutoffs. Therefore, to be of use to researchers, variables need to be operationally defined in such a way that they can be tested and statistically analyzed. An operational definition is a definition that is stated in terms that can be observed and measured. To turn his question into a hypothesis, Dr. Harvey needs to operationally define both the independent and dependent variables. For example, he may decide that "dressing professionally" means that the person wears a dark suit with a white shirt and tie for men and a dark suit with white blouse and pearls for women. Of course, this is not the only definition of "professional dress" possible. Business casual, blazers and slacks, or any number of other possibilities is also possible. However, since it is typically impossible to consider the entire range of possibilities in one research study, Dr. Harvey will have to restrict his study to include only those values of the independent variable that are of most interest. Similarly, Dr. Harvey will have to operationally define what it means not to dress professionally in the workplace (e.g., jeans and a t-shirt). These definitions, of course, restrict Dr. Harvey's hypothesis. His results will not really answer the question about "professional" versus "not professional" attire, but only about the difference in treatment that people wearing power suits receive from those who wear casual attire.

Based on this discussion, it would seem that Dr. Harvey would be well off to pick multiple operational definitions for the independent variable. Although in some ways this is true, operationally defining a variable can be a tricky proposition. The goal of operationally defining variables is not just so that they can be tested in research, but to adequately and accurately define them so that they completely represent the underlying concept as much as possible. For example, as discussed above, the concept of "dressing professionally" means different things to different people. These differences affect not only the persons who need to decide how to dress for success, but also the persons who judge them based on the clothing choices. For example, if one's boss is "old-fashioned" and dresses in a suit and tie, dressing in a suit and tie would be more likely to impress this person even if the standard for "business attire" for that company was jeans and a polo shirt.

In addition, Dr. Harvey will have to operationally define what he means by "how one is treated in the workplace." Operational definitions of this dependent variable could include the supervisor's performance appraisal ratings of the individual, the average time it takes before the person receives a raise or bonus, or whatever other factors Dr. Harvey thinks are indicative of success. Some statistical techniques allow researchers to design experiments where to test multiple conditions of both the independent and dependent variables (e.g., power suit, blazer and slacks, business casual, and casual clothing). However, given the infinite variety of human nature and behavior, it is unlikely that he will be able to include every possible condition in his operational definitions.

Operationally defining dependent variables in human research can be a complicated process. For example, the construct underlying the variable "success in the workplace" is a nebulous and complex concept. Unless one is willing to wait for the end of the subject's career and look back to determine the value of the ultimate criterion of how successful that person was in the end, one can only estimate the ultimate criterion of success in the workplace using one or more predictor measures that one can operationally define. The underlying criterion is a dependent or predicted measure that is used to judge the effectiveness of persons, organizations, treatments, or predictors. However, one does not truly know whether or not a person is successful until s/he retires and can look back on the entire career. Practically, however, this is typically not possible in social science research. Rather than choosing an ultimate criterion of success such as success at the point of retirement, it is typically necessary instead to pick an intermediate criterion of success such as how many promotions one receives within a given period of time, how many (or how large) the raises are that the person receives during that same time period, or the performance appraisal ratings the person receives from his or her supervisor.

When operationally defining predictors to estimate the underlying criterion (in this case, success in the workplace), one strives to define measures that will collect data on as much of the underlying criterion as possible without measuring other extraneous variables that are not related to the criterion of "success." As shown in Figure 1, a condition known as criterion deficiency occurs when the predictor measures that are used as operational definitions of the criterion do not adequate define it. When this happens, the variable as operationally defined is not completely measuring the underlying criterion or hypothesis independent variable, and the research results will less than perfectly reflect the real relationship between the independent and dependent variables. Similarly, to the extent that the predictor measure is actually measuring something other than the criterion (i.e., is contaminated), the results will be imperfect reflections of the actual relationship between the variables. One way to help minimize the problem of criterion deficiency is to use multiple predictor measures, each of which measures a different aspect of the underlying criterion (e.g., use supervisor ratings, number of promotions, and amount of raises rather than just one of these measures). However, when doing this, one runs the risk not only of further contamination, but also of achieving spuriously high results because the predictors are related to each other (e.g., supervisor ratings are typically related to both promotions and raises).

Independent and dependent variables are not the only types of variables about which researchers need to be concerned, however. As shown in Figure 2, a third type of variable called extraneous variables can also unintentionally affect the outcome of a research study. These are variables that affect the outcome of the experiment (i.e., how a person is treated at work) that have nothing to do with the independent variable itself. For example, whether or not the person has visible tattoos or piercings may affect how s/he is perceived by others in the workplace no matter how s/he dresses. Similarly, the type of organization in which the experiment is conducted may affect the outcome of the study: a high-end consulting firm may have different expectations for "professional attire" than does a start-up software company. Other possible extraneous variables might include the expectations of the person giving the rating, how that person dresses, or any number of other factors that are not directly related to the relationship between the independent and dependent variable. As much as possible, of course, these extraneous variables need to be controlled. For example, one could restrict the hypothesis to only deal with consulting firms or business executives. However, the more a hypothesis is restricted, the less it reflects the real world. In addition, no matter how many extraneous variables are taken into account in an experimental design, it is virtually impossible to control literally every possible extraneous variable that may affect the outcome of a study. However, the more of these that are accounted for and controlled in the experimental design, the more meaningful the results will be.

There is a fourth type of variable that may affect the outcome of the research. Intervening variables are variables that occur between the manipulation of the independent variable (e.g., how one dresses at work) and the measurement of the dependent variable (e.g., how long it takes to receive a promotion). For example, if during the time intervening between the change in the person's dressing habits and the time that the person's success is rated s/he receives further training, the resultant rating of the person's professionalism may be related to the training rather than to the way that the person dresses. Like extraneous variables, intervening variables need to be controlled as much as possible so that the effect of manipulation of the independent variable on the dependent variable can be determined.

Applications

The articulation and operational definition of variables is typically not done in isolation, but as part of the process of theory development and hypothesis testing through empirical research. Designing a good research study depends in part on two factors. First, one must try to control the research situation so that the variables measure only what they are supposed to measure. Second, one must try to include as many of the relevant factors as possible so that the research fairly emulates the real world experience. Both of these aspects require the development of good operational definitions for the variables in the study.

As shown in Figure 3, research design starts with a theory based on real world observation. For example, from personal experience Dr. Harvey may know that he is taken more seriously in professional situations when he dresses in a suit and tie. From this observation, he may develop a preliminary theory that if someone who wears "professional attire" is more likely to be perceived by others as being a competent professional than if one does not wear professional attire. Based on his observations and this preliminary theory, he next forms an empirically-testable hypothesis (e.g., "People who wear professional attire are more likely to be successful in the workplace"). To find out if this hypothesis is true, Dr. Harvey next operationally defines the various terms (i.e., constructs) in the hypothesis. As discussed above, he needs to determine how he is going to measure both professional attire and success in the workplace. To do this, he might conduct a study using research confederates who wear specific clothing that he has chosen for them. He might also define success using not only readily available measures such as raises, promotions, or annual performance evaluations, but also might develop a series of rating scales that measure the various components of success in the workplace. He would then run the experiment, using confederates dressed in different ways, collect the measures of the dependent variable, statistically analyze the resulting data using inferential statistics, and -- based on the statistical significance of the answer -- determine whether or not his hypothesis was correct.

Despite the problems with operationally defining the various measures associated with the dependent and independent variables, however, conducting such an experiment in a laboratory setting is a relatively easy task. However, in behavioral research in general and in sociology research specifically, the phenomena are sometimes too big to be controlled in a laboratory setting or the mere fact that the research is conducted in a laboratory changes the results. Because of this fact, many sociology research studies are not conducted in laboratory settings. One approach to research that overcomes some of these limitations is the use of a simulation that approximates the real world setting. Simulations allow the researcher to bring in more real world variables but still control many of the extraneous variables. For example, a laboratory experiment about the relationship between business attire and perceptions of professionalism could be done by having people sit in an empty room and rate pictures of people who are dressed in different ways. Although this would yield some interesting data, it has little to do with the way the supervisors, customers, and employees interact and judge each other in the real world. A possible simulation would be to set up a workplace-like setting and have experimental subjects try to lead people (e.g., teach them a task) and rate how well they did. Another approach would be to conduct a field experiment in which real supervisors would rate the professionalism of real employees in the workplace who -- at the behest of the researcher -- wear specific types of attire. Although this approach has the advantage of being more realistic than laboratory research or simulations, it has the concomitant disadvantage of giving the researcher less control over extraneous variables that may taint the results of the study.

Sometimes, of course, the researcher does not even have enough control over the situation to manipulate the variables at all. For example, letting real employees in the workplace know that one is interested in the effects of different kinds of clothes on the perceptions of others might be enough to taint the experiment. A field study could be used in such a situation. This is an examination of how people behave in the real world. In a field study, the experimenter would just look at the way that people dress and collect data on the operationally defined dependent variables. Frequently, this approach is combined with another research technique called survey research in which subjects are interviewed by a member of the research team or asked to fill out a questionnaire regarding their preferences, reactions, habits, or other questions of interest to the researcher. This could be used to gather information about various extraneous or intervening variables that might taint the results. Unfortunately, although a very thorough interview or survey instrument can be written that would hypothetically gather all the data needed for the researcher to make decisions about the impact of work attire, such instruments are often more lengthy than the potential research subject's attention span. Further, as opposed to the other research techniques, surveys and interviews are not based on empirical data. Therefore, there is no way to know whether or not the subject is telling the truth.

Although the underlying theory is the same in all these research paradigms, the operational definitions of the independent and dependent variables may change depending on the degree of control that the researcher has over the situation. In addition, the possibility of contamination of the research results by extraneous variables becomes greater the less control the researcher has over the experimental situation.

Conclusion

Sociologists attempt to make sense out the world by applying the scientific method to their theories about the way that people act in society. One of the key factors in this process is the identification and operational definition of various variables that account for the observed phenomenon. The two primary variables of interest are the independent variable, which is manipulated, and the dependent variable whose value changes depending on the value of the independent variable. These concepts must be operationally defined in such a way that they can be tested and statistically analyzed. In addition, extraneous variables can unintentionally affect the outcome of a research study while having nothing to do with the independent variable itself.

Intervening variables are variables that occur between the manipulation of the independent variable and the measurement of the dependent variable. They can contaminate the relationship between the independent and depending variables. The articulation and operational definition of variables is part of the process of theory development and hypothesis testing through empirical research and is essential for the conduct of research that yields meaningful results.

Terms & Concepts

Confederate: A person who assists a researcher by pretending to be part of the experimental situation while actually only playing a rehearsed part meant to stimulate a response from the research subject.

Criterion: A dependent or predicted measure that is used to judge the effectiveness of persons, organizations, treatments, or predictors.

Data: (sing. datum) In statistics, data are quantifiable observations or measurements that are used as the basis of scientific research.

Dependent Variable: The outcome variable or resulting behavior that changes depending on whether the subject receives the control or experimental condition(e.g., how long it takes to receive a promotion).

Empirical: Theories or evidence that are derived from or based on observation or experiment.

Extraneous Variable: A variable that affects the outcome of the experiment that has nothing to do with the independent variable itself.

Hypothesis: An empirically-testable declaration that certain variables and their corresponding measures are related in a specific way proposed by a theory.

Independent Variable: The variable in an experiment or research study that is intentionally manipulated in order to determine its effect on the dependent variable (e.g., how one dresses at work).

Inferential Statistics: A subset of mathematical statistics used in the analysis and interpretation of data. Inferential statistics are used to make inferences such as drawing conclusions about a population from a sample and in decision making.

Intervening Variable: A variable that occurs between the manipulation of the independent variable and the measurement of the dependent variable).

Operational Definition: A definition that is stated in terms that can be observed and measured.

Statistical Significance: The degree to which an observed outcome is unlikely to have occurred due to chance.

Variable: An object in a research study that can have more than one value. Independent variables are stimuli that are manipulated in order to determine their effect on the dependent variables (response). Extraneous variables are variables that affect the response but that are not related to the question under investigation in the study.

Bibliography

Black, K. (2006). Business statistics for contemporary decision making (4th ed.). New York: John Wiley & Sons.

Bollen, K. A. (2012). Instrumental variables in sociology and the social sciences. Annual Review of Sociology, 38(1), 37-72. Retrieved November 6, 2013 from EBSCO Online Database SocINDEX with Full Text. http://search.ebscohost.com/login.aspx?direct=true&db=sih&AN=77755951

Karlson, K., Holm, A., & Breen, R. (2012). Comparing regression coefficients between same-sample nested models using logit and probit: A new method. Sociological Methodology, 42(1), 286-313. Retrieved November 6, 2013 from EBSCO Online Database SocINDEX with Full Text. http://search.ebscohost.com/login.aspx?direct=true&db=sih&AN=83576838

Magill, M. (2011). Moderators and mediators in social work research: Toward a more ecologically valid evidence base for practice. Journal Of Social Work, 11(4), 387-401. Retrieved November 6, 2013 from EBSCO Online Database SocINDEX with Full Text. http://search.ebscohost.com/login.aspx?direct=true&db=sih&AN=66698630

Witte, R. S. (1980). Statistics. New York: Holt, Rinehart and Winston.

York, R. (2018). Control variables and causal inference: a question of balance. International Journal of Social Research Methodology, 21(6). Retrieved from http://search.ebscohost.com/login.aspx?direct=true&db=sxi&AN=132083077&site=ehost-live&scope=site

Suggested Reading

Anderson, M. L. & Taylor, H. F. (2002). Sociology: Understanding a diverse society (2nd ed.). Belmont, CA: Wadsworth/Thomson Learning.

Kalmijn, M., & Liefbroer, A. C. (2011). Nonresponse of secondary respondents in multi-actor surveys: Determinants, consequences, and possible remedies. Journal of Family Issues, 32(6), 735-766. Retrieved November 6, 2013 from EBSCO Online Database SocINDEX with Full Text. http://search.ebscohost.com/login.aspx?direct=true&db=sih&AN=60221243

Schaefer, R. T. (2002). Sociology: A brief introduction (4th ed.). Boston: McGraw-Hill.

Stockard, J. (2000). Sociology: Discovering society (2nd ed.). Belmont, CA: Wadsworth/Thomson Learning.

Tilden, T., Hoffart, A., Sexton, H., Finset, A., & Gude, T. (2011). The role of specific and common process variables in residential couple therapy. Journal Of Couple & Relationship Therapy, 10(3), 262-278. Retrieved November 6, 2013 from EBSCO Online Database SocINDEX with Full Text. http://search.ebscohost.com/login.aspx?direct=true&db=sih&AN=62823250

Wilson, S. (2018). Haunting and the knowing and showing of qualitative research. Sociological Review, 66(6). Retrieved from http://search.ebscohost.com/login.aspx?direct=true&db=sxi&AN=132681156&site=ehost-live&scope=site