6.2: Abstract Concepts

Many researchers have documented the detrimental health effects resulting from the recent wave of aggression and discrimination against Asian Americans during the recent COVID pandemic (Iwamoto 2022; Nicholson and Mei 2020; Yellow Horse 2021; Le et al. 2020; Wang et al. 2020; Wu et al. 2021; Chen et al. 2020). This type of research can provide important insights for health professionals when assessing and treating the mental and physical health of their patients.

One research study sought to determine if racial microaggressions were influencing the health of Asian Americans (Nicholson and Mei 2020). Microaggressions are statements, actions, or incidents of indirect or subtle discrimination often resulting in harmful consequences to members of marginalized groups (Sue, 2010). When considering undertaking such a study, an immediate problem becomes obvious. How do the researchers measure health and exposure to microaggressions?

A person's health is a rather complex and abstract concept. What contributes to a person being “healthy”? Certainly, there are many related components that contribute to the overall health of a person. Some numerical measures may be somewhat easy to determine if a member of the population can be examined by a medical professional. These include measures such as resting heart rate, weight, and cholesterol levels. There are literally hundreds of other measures that can be taken on an individual. But how does one compute a measure of a person's health based on these measurements? This is a daunting task particularly considering the many complex relationships that may exists between these measurements and the overall health of a person. For example, a healthy weight for one person may not be considered healthy for another. Beyond these measurements, mental health may also play an important role in determining the general health of an individual. How do we measure mental health?

Assuming we can collect enough measurements from everyone that we would consider to be important to measuring the overall health of a person, how do we use them to compare the health of individuals? This comparison is important because our research hypothesis states that microaggressions have a negative effect on health. Therefore, we need to be able to compare the health of two individuals and definitively state which individual has better health. Given the apparent large number of measurements that we have taken on everyone, how do we decide who is healthier? It seems that we should combine all these measurements into a single measure of overall health, but how should this be done? Is high blood pressure less healthy than a high cholesterol score? How does mental health compare to physical health?

It should become apparent that this has become an exceedingly difficulty concept to measure. Nonetheless, these questions highlight the major difficulties that researchers routinely encounter in statistical research studies.

If microaggressions are having a significant adverse effect one or more groups in a society, then that group is being denied not only their health but they are also potentially being penalized with the economic consequences of additional health care needs and the potential for the loss, or reduction, in employment opportunities. So, the question is, how can this important topic be studied in a way that is relatively simple, reliable, effective, and efficient?

Let us consider how the researchers in this study approached this problem (Nicholson and Mei 2020). There are two important things that must be measured for this study to be effective: exposure to microaggressions and overall health. The data used in the study was based on observations from the 2016 National Asian American Survey, a large and ethnically diverse source of nationally representative data on Asians living in the United States. In this study the health of the individuals was measured by their responses to the question, “How would you rate your overall health in the past-year?” The responses were reported using the categories excellent, very good, good, fair, and poor. This is qualitative categorical data measure using an ordinal measurement scale.

This method for measuring overall health represents an incredibly simple solution given the complexities discussed above. However, there is a very important question that this raises, namely do these responses measure the individual's overall health? Of course, this is not a perfect measure of overall health, particularly because individuals may suffer from conditions that they may not be aware of, or they may discount other relatively important conditions. Hence, someone who is in poor health may report that they are healthy, and correspondingly someone who is in good health might report that they are in poor health. Hence, the measure is not going to be a perfect measure of overall health. However, can this measure still give us an indication of health that may be useful for a study such as this one?

To address this issue, we need to do a little more research. Looking into past research it was found that these measures were strongly correlated with multi-item health measures (Ahmad et al. 2014), predicted risk of mortality with high reliability (Idler and Kasl 1991), and were also strongly correlated with standard well-being scales (McDowell 2009). So, what does all this mean?

Some of the terminology used above may be unfamiliar, and much more on these subjects will be covered later. For now, let us tackle these issues one-by-one. The first result states that these measures are strongly correlated with multi-item health measures; instead of asking everyone their overall health as a single question, the researcher could ask a series of questions that are all intended to assess some aspects of the individual’s health. The idea that the single item measurement strongly correlates with the multi-item measurements means that, basically, the researcher will get similar information from the single-item measure as a multi-item measure. It is very important to note that the exchange is not perfect, and some information will be lost. But what this result states is that not too much information will be lost. In terms of the study, the researchers will hope that enough information is retained about overall health so that some interesting conclusions can be obtained from the data.

So, the single item measure gives us similar information as the multi-item measure. But does this mean that we get useful information about the true health of the individual, or is an individual's self-reported health not informative for physical health? That is, is it enough that we get similar information from the single item measure and is the multi-item measure itself a good indicator of health? The next result states that the single-item measure can predict the risk of mortality for individuals. In the context of this type of research, this means that the responses from the single-item measure are closely associated with mortality rates. That is, those who reported poorer health tended to have a higher mortality rate within a specified time than those who reported better health. Furthermore, this association was strong enough to conclude that useful information is being obtained by the prediction, which is a pretty strong endorsement in that an individual's self-reported health score usefully predicts something that will happen in the future that is associated with general health. Of course, individuals may die from non-health related causes such as fatal accidents or violence, but mortality rates should provide a good indication of general health.

The last result states that the self-reported health measurements have a high degree of reliability and are also strongly correlated with standard well-being scales. There are two results to consider here. The first refers to the reliability of the measure, a concept we will discuss in more detail later in this chapter. For now, what this roughly means is that the measure provides consistent results over time (Van de Sande and Byvelds 2015). That is, the same individual with the same health profile will give the same measurement with repeated observations. The fact that the health measure strongly correlates with standard well-being scales provides further evidence of the validity of the measure. What is being stated here is that the measure seems to be measuring overall health in the same way as several other established health measures. By stating this conclusion, the researchers are borrowing from the strength of other research. The implication here is that these established health measures have been shown to be useful in other studies, and that if this measure is essentially giving very similar observations, then it should be useful as well.

In the material that follows we consider how researchers develop these measures along with how to develop your skills to critically analyze how abstract information is measured in research studies. Research studies begin with a process of conceptualization and operationalization where the researcher decides what and how items should be measured. As in the example above, measurements will be useful only after the measurement systems have been shown to be both valid and reliable. No studies are perfect, and all research studies must forge a compromise between practicality and usefulness. From a critical point of view, one must determine if the compromises are reasonable and if the conclusions of the study are warranted considering these compromises.