Demographics

Demographics in Research

Demographics cannot be manipulated or randomly assigned; therefore, research efforts using these variables are considered “quasi-experimental” designs. The lack of control over these variables makes it impossible for research on them to be considered “experimental.” Establishing causation can be difficult in these studies, and eliminating alternative hypotheses may be problematic. There are many factors known to be related to demographic variables. Consider the following example: Coming from a single-parent family is associated with lower achievement because these families usually have lower incomes, more stress, or less parent interaction. All of these factors are related to lower achievement, but which ones are involved?

Research questions often focus on the demographic variables, such as, do boys differ from girls in reading or is SES related to empathy? However, because demographics tend to be so strongly related to achievement, they are often considered in studies of other factors. For example, if a study were to be conducted on an [Page 487]innovative reading program, the researchers would want to consider sex because they know boys perform differently than girls. If achievement in charter schools were being compared to public schools, all the demographic variables related to achievement should be considered among the students of the schools before making value judgments regarding the schools’ relative achievement.

Therefore, before the research question is addressed, demographics are crucial to any study. They define the population and therefore the sample. They impact validity of a study and limit the generalizability of the findings. Selection bias is one of the greatest threats to validity in educational research, and demographics are often at the heart of the problem. The nature of students when comparing teachers, grade levels, schools, districts, and states matters in both simple and complex ways. The interaction of demographic factors with each other and various outcome variables provide serious challenges that cannot be ignored.

Random assignment of demographic factors is impossible, and random selection of subjects in educational research is difficult. Often the sample in educational research is a convenience sample of available students, teachers, classrooms, or schools. If groups differ based on demographics, anything related to those demographics is likely to differ. This means the differences or relationships the researcher is looking for related to schools, teachers, or outcomes may simply be the demographics. To compensate for the lack of randomization, researchers sometimes match groups based on demographics or statistically control for the group demographics (e.g., covariants).

Demographics in Policy

In the United States, demographic categories came to the forefront in policy when the No Child Left Behind legislation required that schools meet an overall achievement criterion not only for their students but also for demographic subgroups. Each group based on race, poverty, special education, and English-language proficiency was required to meet the criterion. If one group did not pass, the school failed to pass. This disaggregation—breaking a large group into smaller subgroups—may have been well intended, so the subgroups would not be ignored, but it failed to account for interactions. In other words, some students were members of more than one subgroup, making comparisons unfair. For example, there are achievement gaps for students who are Black and students in poverty. If one school has poor Black students, but another school has poor students who are not Black, or Black students who are not poor, the school with the poor Black students is going to be depressed in both categories (making it more difficult to pass either category).

...