Chapter 2: How to Select a Research Topic for a Secondary Study

         While it is important to be interested in a certain topic, it is not easy to conduct primary research with the topic we have chosen mainly due to the feasibility that primary research requires time and cost for designing research, collecting data, analyzing data, and disseminating findings through making professional presentations and writing/publishing manuscripts. Therefore, conducting a secondary study by using an existing dataset that is available is practical, especially for students, to complete a graduate research paper/thesis in a timely manner. Furthermore, conducting a secondary study helps students experience and understand the entire process of conducting research that can prepare them to conduct primary research in the future.

        However, the secondary study has limitations in finding a dataset that fits our research topic and the existing dataset may not perfectly fit with our interest since the dataset has been collected by other researcher(s) reflecting their interests and purposes. As a result, we may not be able to find good variables from an existing secondary dataset for our own research and have no other option to adjust our research topics within the limit of the available data.

Identifying a Study Topic

Before we search and find a dataset in pursuit of our research interest, it is necessary to identify a research topic that reflects our research interest and will be formulated into a research problem in completing our graduate research paper/thesis. Research interests can be stem from our life experiences and everyday interactions with others surrounding us that stimulate our questions and thoughts. In particular, the discipline of social work literally covers all areas of research in all stages of human development such as pregnancy, infancy, childhood, adolescence, young, middle, and older adulthood. Topics can be health (i.e., dementia, cancer, diabetes, eating disorder, obesity, etc.), disability, health practices, health or mental health service utilization, mental health (i.e., depression, anxiety, trauma, attention deficit and hyperactivity disorder (ADHD), suicidal ideation, schizophrenia, etc.), substance use (i.e., illegal drugs, marijuana, cannabis, opioid, fentanyl overdose, smoking, drinking, underage drinking, binge drinking, etc.), social stigma, social/economic inequality, poverty, cognitive function, caregiving, child abuse, elder abuse, intimate partner violence/teen dating violence, social media use, physical activity, academic performance, and so on.  

For our research, one thing to remember at this step is that we need to think about and choose a pair of topics that are reasonably interrelated in terms of cause and effect, cause and outcome, risk and protective factor of our research topic. For example, when our chosen topic is trauma, then it is required to think about whether we are interested in the direction of factors that lead to trauma experiences (in other words, what has caused trauma experiences) or what might be the outcomes affected by trauma. The topic of how teen dating violence (TDV) affects academic performance is an example of an effect of TDV while why certain teens experience TDV more likely than other teens is an example of examining risk factors of TDV. Another example with substance use and TDV, a research topic can be why TDV increases the use of illicit drugs or how illicit drug use affects TDV.

It is noteworthy to mention that two interrelated topics can be both cause and effect, in other terms the independent variable (IV) and dependent variable (DV) depending on what direction the researcher examines. The independent variable (IV) is an active topic that actively affects the other topic/variable in a study while the dependent variable (DV) is a passive topic that is passively affected by the other variable (IV), which is an outcome of IV. For example, substance use (IV) can lead to TDV (DV) while TDV (IV) can lead to substance use (DV). However, we need to set one direction of the relationship between two topics and set which variables to be IV and DV for our research^[1]. Likewise, the literature review should focus on existing studies that examined one direction of our focus unless we decide to do bidirectional relationships between the two variables, which leads to more data analyses.

In short, empirical research is to examine to what extent a certain issue affects another issue, which is experienced by human subjects that are collected as data by researchers in a scientific manner. We call the first topic of concern the independent variable (IV) and the second topic the dependent variable (DV) in a study.

Table 2-1 presents examples of research titles that consist of IV and DV. As presented in Table 2-1, almost all study titles are made up of the study’s IV(s) and DV in general or specific terms to represent IV first and DV next. The first example in Table 2-1, the title presents adverse childhood experiences and digital media use without separating the two topics. But in this title, the first topic of adverse childhood experiences can be IV which affects digital media use (DV) in U.S. children. The second example of the title presents that the study examines how traditional gender myths (IV) affect beliefs about beating (DV). The third example indicates how social media experiences (IV) affect emotional responses (DV) and the fourth title presents how physical abuse experiences affect delinquency in adolescence. In most cases, IV can be multiple, but DV is one topic. In a few cases, DV can be multiple, which leads to more complex data analyses depending on how many dependent variables a study aims to examine. Once we identify broad topical areas of research interests with IV, DV, or both, we can begin to search and find a dataset that fits our interest for a secondary study.  

Finding a Dataset for a Study Topic

The U.S. Governments funded research projects $717 billion in 2020 and $792 billion in 2021 as a result of the U.S. efforts to increase U.S. competitiveness, property, and safety (Anderson et al., 2023). Thus, many research projects were funded by federal, state, and local governments. In many cases, research projects funded by governments are required to open datasets to the public since they use public funds. Data collected with public funds are publicly available without any restrictions or minimal restrictions such as sign-up.

The Centers for Disease Control and Prevention (CDC) is one of the leading federal-level agencies that conduct innovative public health research to understand and prevent health crises and risks across the U.S. population such as estimated national rates of adverse childhood experiences, drug use, domestic violence, and other health-related behavior or risk factors. All data collected by CDC are publicly available without any requirements of signing up. The Youth Risk Behavior Surveillance Study (YRBSS) and the Behavioral Risk Factor Surveillance System (BRFSS) are the most popular projects to collect data with nationally representative samples. The National Institute of Health (NIH) and the National Institute of Justice (NIJ) also conduct research projects and share datasets publicly available as presented in Table 2-2.

There are also data repository sites for research datasets funded by U.S. governments. For example, there is a search site for data repository sites to access datasets funded by NIH. The most popular data repository site for social research projects funded by U.S. governments and other open research datasets is ICPSR (Inter-Consortium for Political and Social Research). Table 2-2 presents the most frequently utilized sites retrieving existing datasets. Some datasets are publicly available, and others are required permission to download the dataset by signing up. Some other datasets have additional requirements in order for us to access the datasets. In this type of dataset that requires permission, the Institutional Review Board’s approval^[2] from the student’s university is required when we request a dataset.

        Another way to find a dataset related to our research topic is to find the peer-reviewed journal articles in our research interests that conducted a secondary study and identify in the articles what datasets the studies have used in the Methods section or even in Abstract of the articles as presented in Table 2-3. Many published studies conducted a secondary data analysis by using an existing dataset to test their study hypotheses. We can find the dataset by searching the name of the dataset in Google. Many datasets we can find are studies funded by public funds and are available to us as described above. In a few cases, the data were collected by the authors who used sources other than public funds. In those cases, we may reach out to the primary researchers of the study or the authors of the published article to use their datasets in completing the student’s research project.

Choose a Study Topic with a Dataset

        Once we determine a general research topic and find a dataset, then we review the dataset that fits our topic the most. In this book, the Youth Risk Behavior Surveillance System (YRBSS)^[3] data will be used to explain the steps of completing a research project. The YRBSS dataset is collected by the Centers for Disease Control and Prevention biannually and is publicly available without any restriction. To choose a specific study topic for the graduate research project from the YRBSS dataset, first, we look at the YRBSS user’s guide to understand what information has been collected for the YRBSS. For other datasets, we may review a file of a data dictionary, codebook, or survey questionnaire that describes what information or survey questions the original primary study collected.

As presented in Table 2-4, YRBSS collects data with 99 survey questions including risky youth behaviors of driving, substance use, physical activity, sexual activity, dietary, and mental and physical health status.  As explained above, we need to choose two areas of research topics for IV and DV. The following can be a research topic with IV and DV from the information the YRBSS collected: 1) Dating violence (IV) and depression/suicidal ideation (DV), substance use (DV), or difficulty concentrating (DV); 2) Sexual activity (IV) and dating violence (DV), or birth control (DV), 3) Exposure to violence (IV) and sexual activity (DV), depression/suicide (DV), substance use (DV), or dating violence (DV); 4) Bullying (IV) and mental health (DV), substance use (DV), school safety (DV), or school closeness (DV); 5) Parental supervision (IV) and sexual activity (DV), substance use (DV), or exposure to violence (DV).

Variables with a Study Topic and Hypotheses Development

         When we choose a study topic with IV and DV, the study topic is generic names of independent and dependent variables from the dataset. Once we choose a study topic with IV and DV from a dataset, identifying the specific variables would be helpful in preparing for the next steps of the graduate research project. As presented in Table 2-5, we can list the actual variables with generic names of variables.

When we identify the variables, we should include basic demographic variables, e.g., age, gender, and race/ethnicity as independent variables. In this step, some students are confused about identifying demographic variables as the dependent variables. However, as we learned above, DV is the outcome or effect variable affected by IV. Demographic variables can’t be changed or affected by other variables, but those are characteristics we have been born with. In addition, demographic variables can affect the probability of certain experiences as minority youth are more vulnerable to bullying victimization, exposure to violence, and so forth. Women are more vulnerable to experiencing intimate partner violence and depression. Therefore, demographic variables are independent variables. However, in many studies, demographic variables are not key independent variables that affect the outcome variable of the study.

As presented in Table 2-5, most studies have key independent variables rather than demographic variables, but we include demographic variables in the study variable list. This is because demographic characteristics are very powerful variables that affect our dependent variable since our life experiences are affected by our race, gender, age, and family. Therefore, most studies try to control the effect of demographic variables on our dependent variable to estimate the effect of the key independent variable on our dependent variable, which is called the control variable. Therefore, in many studies, demographic variables are used as control variables unless we have a special concern about the effect of one of the demographic variables on our dependent variable such as gender differences or racial differences.

The first example of a variable list presented in Table 2-5 includes dating violence as IV and substance use as DV. The variable list presents three variables for dating violence and twelve variables for substance use including all types of substances. Table 2-5 can be reviewed with Table 2-4 to understand why those variables are selected.

Develop Tentative Hypotheses

A hypothesis is a tentative statement indicating the direction of a relationship between IV and DV that we test with data. Without reviewing existing study findings for the literature review, we can develop tentative hypotheses for a relationship between IV and DV.  This step guides us to what to write in the Introduction and Literature Review sections. We can develop two hypotheses from the variable list presented in Table 2-5. One hypothesis can be formulated with your primary IV with the DV and the other hypothesis is with one of the demographic variables with DV (race/ethnicity, gender, or age). For dating violence and substance use presented in Table 2-5, the first hypothesis can be dating violence (IV) increases substance use (DV) in youth while the second can be female youth (gender in demographic variables, IV) experience experiencing dating violence use substances (DV) more than their male counterparts. For exposure to violence and sexual activity, the first hypothesis can be youth who were exposed to violence (IV) are more sexually active (DV) than those who were not while the second can be minority youth (demographic variable, IV) were exposed to violence more than White youth leading to be sexually active (DV). For physical activity and difficulty of concentration, the first hypothesis can be youth who are physically active (IV) show a lower level of concentration difficulty (DV) while the second can be White youth (demographic variable, IV) are physically active more than minority youth showing a lower level of concentration issue (DV).

Once we develop tentative hypotheses, we begin to search existing studies that examined and reported findings that are related to our tentative hypotheses to understand their findings on the relationship between our IV and DV. Then we begin to write the Introduction and Literature Review sections.