Question: About linear regression mentioned in the reference, "The magnitude and direction of the relationship between the independent and dependent variables are denoted by coefficients ". How do you interpret those coefficients for audience with no statistical background?

Reference;

Reflection from the results- Logistic Regression and Multiple (Linear) Regression

Comparison of findings from the logistic regression conducted in Assignment 1 and the linear regression conducted in Assignment 2.

An examination of the results obtained from the logistic regression in comparison to the linear regression revealed variations in both interpretation and analysis. More precisely, a logistic regression analysis was utilized in Assignment 1 to investigate the correlation between sex, exercise frequency within the previous 30 days, BMI category. Sex and exercise did not exhibit a statistically significant impact on the BMI category, as indicated by the findings. Notwithstanding the absence of statistical significance, the analysis yielded valuable insights regarding the determinants of BMI and emphasized the necessity for additional investigation into these variables. Conversely, a multiple linear regression model was employed in Assignment 2 to examine the identical relationship. In a comparable way, the results indicated that neither sex nor exercise were substantial predictors of BMI category. Nevertheless, the linear regression analysis accounted for only a small percentage of the variance in the BMI category, suggesting that the model may have certain limitations in its ability to predict.

On the other hand, logistic regression yielded odds ratios, whereas linear regression presented coefficientseach of which provided distinct perspectives on the interrelation among variables. By utilizing the odds ratios, it was possible to deduce how the probability of an individual falling within a specific BMI category varies in relation to the independent variables. Instead of likelihood ratios, coefficients were generated by linear regression. The coefficients in question symbolize the effect of a one-unit modification in the independent variables on the dependent variable (BMI category).

How are the findings from a logistic regression analysis best reported/presented. Justify your position?

When reporting and presenting the results of logistic regression, it is most effective to emphasize odds ratios, significance levels, and model fit statistics (LaValley, 2008). Probability ratios offer significant insights into the direction and extent of the correlation between the dependent and independent variables. It is advisable to include confidence intervals alongside these in order to evaluate the accuracy of the estimates. Furthermore, the inclusion of significance levels for the predictors aids in comprehending the statistical significance of the individual variable's impact on the model. Model fit statistics, including the omnibus test and classification table, furnish insights into the model's predictive accuracy and its ability to account for the variance in the data (Lund Research Ltd, 2018).

How, in your opinion, findings from a linear regression analysis are best reported/presented. Justify your position?

When reporting and presenting the results of linear regression analysis, it is most effective to emphasize the coefficients, significance levels, and goodness-of-fit measures. The magnitude and direction of the relationship between the independent and dependent variables are denoted by coefficients. The inclusion of standard errors in conjunction with coefficients is crucial for evaluating the accuracy of the estimates. The significance levels assist in ascertaining whether the coefficients predict the dependent variable in a statistically significant manner. Goodness-of-fit metrics, including adjusted R-squared and R-squared, furnish insights into the degree of correspondence between the model and the data (Daniel & Cross, 2019).

How could the findings from each of these analyses be used in public health research and in public health practice?

The implications of logistic regression results for public health practice and research are diverse. An example of this is how the identification of factors linked to high BMI categories can provide valuable information for the development of targeted interventions designed to decrease obesity rates. The results of logistic regression can also be utilized to identify populations at high risk, which could be advantageous candidates for customized preventive or therapeutic approaches. Moreover, gaining insight into the correlation between diverse risk factors and health outcomes can provide valuable guidance for formulating public health policies and allocating resources.

In contrast, the outcomes derived from linear regression can make diverse contributions to the fields of public health research and practice. To begin with, gaining insight into the influence of demographic factors, such as gender, on body mass index (BMI), can contribute to the formulation of targeted public health interventions for particular populations. Furthermore, the identification of modifiable risk factors, such as exercise patterns, can provide valuable insights for health promotion initiatives that seek to enhance the overall health and well-being of individuals. The findings of linear regression can provide valuable insights for evidence-based interventions and policies that seek to enhance public health outcomes by identifying factors that are associated with BMI (Redman & Thackway, 2021).

References

Daniel, W. W. & Cross, C. L. (2019). Multiple regression and correlation. InBiostatistics:

A foundation for analysis in the health sciences(11^thed., pp. 416-454). Wiley.

LaValley,M.P. (2008). Logistic regression.Circulation,117(18), 2395-2399.

https://doi.org/10.1161/circulationaha.106.682658Links to an external site.

Lund Research Ltd. (2018).Binomial logistic regression using SPSS statistics.Links to an external site.

Laerd Statistics.

https://statistics.laerd.com/spss-tutorials/binomial-logistic-regression-using-spss-statistics.phpLinks to an external site.

Redman,S., & Thackway,S. (2021). Regression-Public Health Research & Practice.

Public Health Research & Practice,25(1).