Quantitative positivist and qualitative interpretivist methodological approaches to design and causation

Social Epidemiology -- Methods

Describe quantitative/positivist and qualitative/interpretevist methodological approaches; include examples of their research methods of data collection.

In conducting social research, the design used is either quantitative or qualitative, or oftentimes, a combination of both. Quantitative or positivist research design aims to describe or analyze a specific phenomenon occurring in a specific population or group

Qualitative or interpretevist research design, meanwhile, provides an in-depth illustration of the phenomenon; rather than providing a general description of it to a specific population or group, the dimensions surrounding the phenomenon are uncovered, explored and studied intensively for greater understanding of the phenomenon.

In effect, quantitative or positivist research design aims to provide a general 'picture' of the phenomenon in a specific population, while qualitative or interpretive research design provides an in-depth understanding of meanings and concepts regarding the phenomenon.

Specific to social epidemiology, quantitative research design assumes the epistemology of positivism, which posits that there is a 'single fixed truth' that can be proven objectively and measured quantitatively. This truth is in the form of hypothesis that are quantitatively defined and measured, and tested statistically

. Qualitative research design is known for its interpretevist epistemology -- that is, unlike quantitative research design, there exist several truths and phenomena or events are 'socially constructed,' contributing to the variability of truth that is being established through qualitative methods.

Quantitative design has three (3) approaches to research: descriptive, relational, and experimental. Descriptive research provides a description of an event or phenomenon, while a more in-depth approach can be done through relational research, which measures the presence and kind of relationship in two or more variables. Descriptive studies, in effect, "make no attempt to analyse [sic] the links between exposure and effect" while relational studies only establish links between variables (Beaglehole and Bonita, 2006, p. 40). Experimental research, meanwhile, more than just describes and establishes relationships among variables; as a quantitative research approach, experimental studies are conducted to determine and measure the causal relationship among variables. It is in this third approach that the researcher can establish "causation," which is the most important and challenging aspect of a research study, as it provides an explanation and understanding of the event or phenomenon under study (Rosnow and Rosenthal, 1996, pp.15-16).

Methodologies used in quantitative research are also grouped according to the abovementioned approaches. Under descriptive and relational approaches (or collectively termed as observational studies), commonly used methods are ecological/correlational, cross-sectional/prevalence, case-control, and cohort or follow-up studies. Under the experimental approach, methods used are randomized controlled trials (or RCTs), field trials, and community trials (or community intervention studies) (Beaglehole and Bonita, 2006, p. 40).

The common factor among these methods is that variables and concepts are identified and determined quantitatively through measures. However, each method uses a different kind of technique, tools, and data analyses, which are described in the discussions that follow.

Qualitative research design is similar to quantitative design; both use systematic observation as one approach to understanding an even or phenomenon under study. However, as discussed earlier, qualitative design analyzes and interprets data in "non-numerical form," wherein words, phrases, or passages from interviews or discussions are documented, analyzed and interpreted using the event or phenomenon as context. Methods used in qualitative research design are participant observation, ethnography, and focus group discussions (FGDs) (Rosnow and Rosenthal, 1996, pp.78-80).

Participant observation puts the researcher as an 'active participant' in the process of observing an event or phenomenon as it occurs naturally in a specific group or population. The 'participation' of the researcher is in the form of interviews with members of the group under study while taking note of all aspects relevant to the occurrence of the event/phenomenon. Basically, this method is non-structured in its approach and subject to changes or modifications of the researcher, as s/he sees fit in the course of his/her observation and participation in the event/phenomenon. Ethnographic research or ethnography, meanwhile, has more structure than participant observation in that there is a specific approach and structure as to how the observation and documentation will be conducted by the researcher. Interpretation is usually conducted after observation and documentation, unlike participant observation wherein interpretation is done spontaneously, resulting to the modified approach during the participant observation process (Rosnow and Rosenthal, 1996, pp.75-77). Focus group discussions or FGDs, meanwhile, is the synergistic discussion of members of specific groups/population about an event or phenomenon under study. As the definition indicates, this qualitative research method generates insights from the 'synergy' in discussion that becomes inevitable when people with a common experience about an event/phenomenon are grouped together to provide this event/phenomenon in-depth analysis and interpretation from their perspective or worldview.

IB. Traditionally epidemiology (prevalence, distribution and determinants of disease) has focused on quantitative research for the types of research designs that are associated with IIA. Can social epidemiology include both quantitative and qualitative methodological approaches? Critically discuss.

Differences in approach between quantitative and qualitative methods provide a distinct illustration of how each method appropriately answers the research question and hypotheses and how data is collected, analyzed and interpreted. Quantitative methods provide descriptions, establish relationships and explain causation among variables through measurable (numerical data). Qualitative methods, meanwhile, explore concepts and socially constructed events and "document the customs, habits, actions" of a group or population of interest (Rosnow and Rosenthal, 1996, p.77). Looking at each design's limitations, quantitative methods can provide generalizations about a population of interest regarding an event/phenomenon, while qualitative methods cannot. Conversely, qualitative methods can provide in-depth background and insights about an event/phenomenon as it occurs in a group/population of interest, but since data generated is non-numerical and is not measurable, conclusions are case-specific at best and cannot be used to represent the population under study.

In social epidemiology, quantitative design and methods are commonly used because measurement is an integral part of the discipline as a body of knowledge and as applied in the medical sciences, particularly, public health. Social epidemiology is a critical component of public health planning, which brings with it important diagnostic measures such as the prevalence, distribution and determinants of diseases.

Developing an effective and responsive public health planning is merely the means to an end, which is for practitioners to have "knowledge of how to treat and/or to prevent the disease." However, this knowledge can only be generated and substantiated through quantitative methods, providing basic yet critical information on the prevalence, incidence, and severity of the disease in a population at a given point in time (Olsen and Christensen, 2010, pp. 3-4).

Take, as an example, the use of survey research method to determine the incidence of tuberculosis (TB) and knowledge and attitudes about it in a community. It can be assumed that the community has a low incidence of TB; however, this cannot be scientifically established until the public health centers will conduct a prevalence study to collect this critical information. Further, for public health planning, medical professionals and healthcare practitioners would not be able to identify the level of knowledge and misconceptions the community has about TB until the prevalence study/survey is conducted. Without the survey, the plan that will be developed may not specifically respond to possible misconceptions and potential lack of knowledge of the community about TB. This is just one of the basic yet essential uses of quantitative research in the practice of social epidemiology.

Indeed, in a discipline wherein numbers and statistics are critical and essential to prevent the spread of a disease, utilizing a qualitative research design could be the least preferred approach to understand and explain an event/phenomenon.

However, the discipline of social epidemiology, through its extant literature and research studies, shows a growing recognition for qualitative methods as a complementary approach to exploring and explaining an event/phenomenon. Two important concepts associated with qualitative methods could explain this recognition: validity and triangulation.

One of the important characteristics of a good quantitative research study is that data is reliable and results are consistent even if data is re-tested using the same kind of analysis. Qualitative research design's strength, meanwhile, is that data generated from its methods can be considered authentic or valid, mainly because sources of data and information are "experts" or highly experienced in the event/phenomenon being discussed. Qualitative methods and analysis and interpretation of information collected are considered authentic and based on real-life experiences of the participants or subjects of the study (Rosnow and Rosenthal, 1996, p.122). Thus, in cases wherein it is also important to establish strong data validity in a study, qualitative studies are recommended to be conducted as a complement to quantitative methods.

Another important use of qualitative methods as a complement to quantitative methods is to achieve "triangulation" in the study -- that is, when the researcher would like to "zero in' on the effect of interest" (Rosnow and Rosenthal, 1996, p.74). Recognizing that each method, whether quantitative or qualitative, has its own limitations, combining both designs in a study has become a common approach to provide the researcher substantial data and information about the study or event of interest. Indeed, as Muntaner (2003) posited in her research, qualitative methods can be included in a dominantly quantitative research design "in situations where qualitative research adds knowledge that would not be available via quantitative methods" (p. 55). Through a mixed-methods design, the researcher can provide better analyses and stronger interpretations and recommendations through balanced strengths of data reliability and validity -- that is, the achievement of "triangulation" in the research study.

IIA. Based on your reading of books such as Beaglehole (1993), describe what you know about observational epidemiology as a research approach and compare it to experimental studies. Describe some of the designs within each, e.g., RCT, case-control studies, etc.

Observational and analytical/relational studies provide different results and answer different research questions and hypotheses when compared to the experimental approach. In observation and analytical/relational studies, the highest kind of analysis that can be done is correlational and not causal.

This is reflective in the types of methods used under this approach, which include ecological, cross-sectional, case-control and cohort studies.

Ecological studies are conducted as preliminary research that could eventually lead to the development of hypotheses and further research about the phenomenon (i.e., higher level of analysis, such as analytical/relational or experimental). Ecological studies use the descriptive approach because it only shows the behavior of the variables under study; data and information are mainly represented as facts, and if the researcher would like to determine the existence of a relationship among the variables identified, then analytical studies would be the next research approach to use.

Cross-sectional studies are analytical studies that help "measure the prevalence of the disease." It is also known as prevalence studies. This kind of study is useful for analyzing relationship among variables at a given point in time, that is, when getting a "snapshot" of the phenomenon in a specific group/population is needed for that time period. Case-control studies are a longitudinal type of study that looks at specific groups of people within the same population. Commonly used to investigate the causes of diseases, it involves studying case and control groups within the same population over time, allowing the researcher to determine the "estimated relative risk of the disease" based on analyses conducted in these groups. Lastly, cohort is another type of longitudinal study that tracks the outcome of a phenomenon over time over specific groups of people within the population of interest. It is similar to case-control studies only, the path of inquiry is in sync with the progression of time, unlike case-control studies wherein the method of inquiry goes backwards (since the objective is to determine the cause(s) of the disease) (Beaglehole and Bonita, 2006, pp. 41-45).

Under the experimental approach, quantitative methods commonly conducted are randomized controlled trials (or RCTs) and its varieties, field trials and community trials (or community intervention studies) (Beaglehole and Bonita, 2006, p. 40). In RCTs, participants are randomly selected and assigned to either the experimental or control group. The strength of this type of study is that it ensures the randomness of subject's/participant's assignment in the study, "unaffected by the conscious or unconscious biases of the investigators (ibid, p. 50). However, this strength of RTCs is countered by the fact that there are also potentially inherent and common weaknesses within the researcher's study: random error, either through sampling or measurement error. Random error resulting from sampling recognizes that the researcher could have chosen an inappropriate sampling technique, therefore resulting to an "inaccurate measure of association" in the study. Another potential cause for random error is measurement error, which happens when constructs or variables are poorly determined and measured in the study. Both reliability and validity of data become susceptible to random error as a result of poor construction of the variables' measures (ibid, pp. 51-52).

IIB. Discuss why "causation" is important in the field of epidemiology?