Abstract
Quantitative research designs utilize standardized mathematical approaches to affirm or disprove theories or explanations. They are characterized by the collection and analysis of numerical data to draw conclusions on a phenomenon. The strengths and weaknesses of each design determine its suitability for a particular research. The correlational research design is recommended for the proposed study to determine the correlation between medical/social support and medical compliance in African American women infected with HIV.
Quantitative Research Designs
A quantitative research design applies mathematical and statistical approaches to test a hypothesis based on numerical data (Creswell, 2009). A key strength of this design lies in the use of unbiased statistical methods to test and verify hypotheses or explanations about a phenomenon. Quantitative research findings can be generalized to other subpopulations because data are obtained from random samples or replicates. In addition, the investigator can control for confounding variables, generating clear cause/effect relationships. The design relies on numerical data, which are more precise than qualitative data. A quantitative study procedure observes established reliability and validity standards, which enhances objectivity and eliminates researcher bias.
The major weakness of quantitative designs is confirmation bias. They lay emphasis on hypothesis testing or theory verification as opposed to generating new explanations for an observed phenomenon. In addition, quantitative models often produce “abstract and general findings”, which limits their external validity (Creswell, 2009, p. 41). The categories and variables measured in quantitative studies may not represent the actual characteristics of a natural phenomenon.
Cross-sectional Designs
A cross-sectional design “measures exposures and outcomes” in a study population at a particular point in time (Frankfort-Nachmias & Nachmias, 2008, p. 141). The investigator determines the level of association between a disease exposure and its outcome, hence, useful in hypothesis generation. Another major strength of cross-sectional designs lies in the simultaneous measurement of the exposures and ‘outcomes’ of the study population at one time (Frankfort-Nachmias & Nachmias, 2008). This approach helps reduce the confounding effects. The researcher can examine more than a single exposure/outcome using this design. The ‘outcome’ can be used to infer the prevalence of a particular condition or disease in a population of interest.
Cross-sectional designs have some limitations. First, it is not possible to determine cause/effect relationships using a cross-sectional design since both the exposure and outcome are measured concurrently. Second, the design is not suitable for studying rare infections with a short incubation period (Frankfort-Nachmias & Nachmias, 2008). In this case, finding a sufficient number of exposed cases from the target population may be difficult. Third, cross-sectional designs are also subject to “selection, measurement, and recall bias” stemming from subject forgetfulness (Creswell, 2009, p. 67).
Quasi-experimental Designs
A quasi-experiment design differs from a true experimental design in the sense that it is devoid of random subject assignment (Morgan, Gliner, & Harmon, 2011). The central aim is to identify a general trend associated with an intervention by comparing the pre- and post-exposure measures. Thus, quasi-experimental designs are useful when randomization of subjects is not possible. A second strength inherent in quasi-experimental designs is the ability to be incorporated into case studies. In addition, since quasi-experiments do not involve subject pre-selection and randomization, they are less resource intensive compared to ‘true experiments’.
Evidently, the lack of randomization weakens quasi-experimental designs. Without random assignment, statistical tests cannot yield meaningful conclusions. Quasi-experiments also cannot account for the effect of external factors on the experiment. Thus, unexpected factors can affect the validity of the findings.
Descriptive Designs
A descriptive design describes the nature of study variables without consideration of their causal relationships. Data collection may involve observation, interviews, and questionnaires, among others. The multifaceted data collection model is a critical strength of this design. It not only generates statistics about a phenomenon, but it also illustrates the subjects’ experiences. Descriptive designs are inexpensive to conduct and can yield clues about the probable cause/effect relationships. They also entail fewer ethical issues compared to experimental designs.
One limitation inherent in descriptive designs relates to confidentiality. According to Creswell (2009), the study participants may not entirely be truthful during interviews because they feel that they should act in a particular way. Others may refuse to respond to personal questions. Objectivity is also lacking in descriptive studies. The prescriptive nature of interview or survey questions may lead to subjectivity. Descriptive designs are also prone to researcher bias as the investigator may focus on what he/she considers pertinent to the study and ignore other information.
Correlational Designs
Frankfort-Nachmias and Nachmias (2008) define a correlational design as a study that examines the “association/correlation existing between variables not subject to experimental manipulation” under lab conditions (p. 184). Its strength lies in the ability to involve several subjects at a single point in time to achieve data saturation. In addition, researchers can measure many variables and determine existing interrelations between datasets. The design applies to measuring study variables that cannot be manipulated in the lab.
Correlational designs have three significant weaknesses. First, a correlation may not indicate causation because of the interaction effects associated with variables (Creswell, 2009). Second, correlation gives no clue about the directionality of the interaction. It only shows that the two variables are correlated. Third, measurement of variables using self-reports or questionnaires may be prone to bias.
Experimental Designs
A classical experimental design contains independent and dependent variables, pre- and post-treatment measures, and experimental and control groups determined through randomization of subjects (Frankfort-Nachmias & Nachmias, 2008). The ability to establish causality is a critical strength of ‘true’ experimental research. In addition, the design allows the researcher to control for the effect of extraneous factors. In this way, it is possible to establish causality with confidence. Experimental designs allow for replication because they involve standardized assessments. The variables can be measured over the duration of the study, which facilitates longitudinal research.
The main weaknesses of experimental designs relate to experimenter effect, confounding effects, and artificiality (Shavelson & Towne, 2012). The experimenter effect stems from the investigator’s “subtle cues that affect the subjects’ response” during treatment (Shavelson & Towne, 2012, p. 77). Laboratory experiments are also prone to artificiality, i.e., the lab procedures may not reflect natural conditions. In experimental research, extraneous variables are beyond the control of the researcher, raising the possibility of confounding effects.
The Recommended Quantitative Research Design
Quantitative study designs include descriptive studies, experimental/quasi-experimental studies, correlational designs, and cross-sectional designs (Creswell, 2009). The proposed study will examine the relationship between social and medical support and HIV treatment compliance among African American women with HIV. The researcher aims to establish if social and medical support (IV) affects medication compliance (DV) in a particular demographic. Hence, the correlational design would be most appropriate for determining the correlation between the two variables. According to Barnham (2015), correlational designs can utilize primary or secondary data to deduce the probable correlation or the effect of an event. The rationale for selecting this design for the proposed study is that it will allow the researcher to infer the association between social and medical support and medical adherence.
Reasons Why the Other Designs are not Appropriate for the Research
A descriptive design only describes the variables, but not their interrelationships. Thus, would be unsuitable for determining the effect of social/medical support on patient compliance. It does not attempt to determine the effect or relationship between variables. On the other hand, an experimental/quasi-experimental design can determine the effect of IV on DV. It entails controlling for one or more variables. Therefore, it is not appropriate for this study, which relies on latest researches and published statistics from HIV clinical programs to determine the correlation between the IV and DV. The study will not involve manipulation of variables; rather, the effect of medical/social interventions on patient compliance will be inferred from secondary data.
A cross-sectional design, though useful in determining the strength of the association between variables, may not be appropriate for the proposed study. It requires the simultaneous measurement of the exposure and outcome. In the proposed study, the data collection (published data) spans over an extended period, making a cross-sectional study inappropriate for testing the study’s hypothesis.
Conclusion
A quantitative study design tests a hypothesis using a standardized format involving statistical analysis. Various strengths and weaknesses are inherent in quantitative designs. Cross-sectional, experimental, quasi-experimental, and correlational designs give the researcher different abilities to manipulate study variables and measure the results. The recommended design for this study is the correlational design. The researcher can use this design to determine the correlation (negative or positive) between social/medical support and medical compliance in the target demographic using published data.
References
Barnham, C. (2015). Quantitative and Qualitative Research. International Journal of Market Research, 57(6), 837-854.
Creswell, J. W. (2009). Research Design: Qualitative, Quantitative, and Mixed Methods Approaches. Thousand Oaks, CA: Sage Publications.
Frankfort-Nachmias, C., & Nachmias, D. (2008). Research Methods in the Social Sciences. New York, NY: Worth Publishers.
Morgan, G. A., Gliner, J. A., & Harmon, R. J. (2011). Quasi-experimental Designs. Journal of the American Academy of Child & Adolescent Psychiatry, 39(6), 794–796.
Shavelson, R. J., & Towne, L. (2012). Scientific Research in Education. Washington, DC: National Academy Express.