Introduction
Biostatistics is complex as it integrates statistical and biological concepts. Often, this course entails investigating the linear connection between exposure, confounders, and outcomes. According to the National Library of Medicine (n.d.), exposures are the independent variables whose alteration influence the results. Oppositely, regressands are the dependent variables whose fate relies on the determinants (National Library of Medicine, n.d.). The essay evaluates two linear regression models (causes of “diabetes” and “allergies”).
Model Interpretation
The first model has the following eight primary interpretations from a statistical perspective. First, people have a 130% autonomous chance of contracting diabetes at a 5% significant level, even without subjecting themselves to risk factors like income and age. The equation also holds that body mass index (BMI) increases the chances of contracting this disease by 240%. Family history with diabetes, gender, and age raise susceptibility to this infection by 230%, 170%, and 140%, respectively, at a 95% confidence level. Lastly, age, race, and height elevate people’s vulnerability to developing this complication by 170%, 260%, and 340%, respectively, at a 5% significant level.
Model two has six main interpretations from a statistical viewpoint. First, individuals have 450% autonomous possibilities of developing allergies at a 5% significant level, even without exposing themselves to risk factors like weight. Another impression is that folks from families with allergy histories have a 380% chance of contracting this infection at a 95% confidence level. Other factors like gender and age elevate vulnerability to developing allergies by 210% and 140%, respectively, at a 5% significant level. Lastly, race and weight increase the likelihood of experiencing this complication by 80% and 150%, respectively, at a 95% confidence level.
Potential Confounders
The two models never specified their confounders, but critical evaluation can help identify them. First, Wunsch (2007) defines confounder as a factor connected to the determinant and related to the dependent variable that can influence the exposure-outcome association. In this case, demographic factors like gender, age, and race, relate with all variables in both equations. As a result, these variables are the potential confounders in the two models. Specifically, gender, age, race, height, BMI, and income are the ideal externalities in the “diabetes” model. Gender, race, weight, and age are the perfect confounders in the “allergies” equation.
Matching Versus Including All Confounders (Randomization)
Data analysts have many approaches for handling confounders, including randomization and matching. First, randomization occurs when the researcher includes generic participants’ information in the regression model (Riaz et al., 2018). For example, the surveyor could disproportionately interview people from different genders and age groups. This approach would enable the investigator to incorporate all externalities in the model and reduce statistical biases. However, randomization is difficult and time-consuming, especially when gathering and including data into the linear model. As a result, applying this technique to the two equations could spark the two difficulties.
Matching is another potential strategy to control confounders’ impact on the regression model. Mainly, this tactic prevails in the controlled experiments where the researcher gives confounders equal opportunities to reduce biases (Mansournia et al., 2018). For instance, if age and gender are potential externalities, the surveyor could apply matching by picking a male and female participant in the 18-25 age group. The main advantage is that matching applies well to controlled studies like in these scenarios (establishing the causes of “diabetes” and “allergies”). However, using this technique could attract challenges like time-consuming and limiting the sample to achieve equality.
Conclusion
In summary, the “diabetes” model had eight main interpretations and six potential confounders. The externalities include income, BMI, gender, race, height, and age. Conversely, the “allergies” equation had six conclusions and four confounders. These externalities consist of someone’s gender, weight, age, and race. Including all confounders in the regression model minimize statistical biases but is challenging and time-consuming. Lastly, the matching technique is reliable in controlled experiments like the duo (causes of “allergies” and “diabetes”), but it is hard to use and limits samples.
References
Mansournia, M. A., Jewell, N. P., & Greenland, S. (2018). Case–control matching: Effects, misconceptions, and recommendations.European Journal of Epidemiology, 33(1), 5-14.
National Library of Medicine. (n.d). Dependent and Independent Variables. National Library of Medicine. Web.
Riaz, H., Khan, M. S., Siddiqi, T. J., Usman, M. S., Shah, N., Goyal, A.,… & Ahmed, H. (2018). Association between obesity and cardiovascular outcomes: A systematic review and meta-analysis of Mendelian randomization studies. JAMA Network Open, 1(7), 1-7.
Wunsch, G. (2007). Confounding and control.Demographic research, 16, 97-120.