KONSEP DAN CARA ANALISIS REGRESI LOGISTIK SEDERHANA MENGGUNAKAN SPSS

Summary

This video explains the steps involved in conducting a logistic regression test, focusing on multivariate analysis. The process includes selecting relevant variables, assessing significance, and handling variables above or below a 0.25 significance value. It covers concepts like causal and predictor relationships, and explains how to manage confounding variables. Additionally, the video introduces methods like forward and backward entry for multivariate analysis, guiding viewers through testing, variable selection, and interpreting results. The presenter emphasizes the importance of understanding theory and adjusting models accordingly to draw valid conclusions.

Takeaways

• 😀 Logistic regression is a multivariate analysis technique used to identify relationships between variables, considering other interfering factors.
• 😀 Variable selection is the first step in logistic regression, where you conduct univariate analysis and assess the significance value (below 0.25).
• 😀 Variables with significance above 0.25 can still be included in the multivariate model if they are substantively important according to theory or prior research.
• 😀 The research concept determines the approach: if causal, focus on the relationship between X and Y; if predictive, consider the influence of multiple independent variables.
• 😀 After variable selection, you can enter variables with significance below 0.25 into the multivariate model, using either forward or backward entry methods.
• 😀 In causal research, if changes in the model's output exceed 10%, the variable should remain; if below 10%, it can be removed.
• 😀 Multivariate analysis in logistic regression involves entering and testing variables one by one, removing and adding based on their significance.
• 😀 When analyzing categorical data, ensure that the reference category represents non-risk cases, while the comparison group includes those at risk.
• 😀 After testing, the logistic regression model's results should be examined by removing variables sequentially, checking the changes in the output.
• 😀 A change greater than 10% in the model output indicates that the variable significantly impacts the model and should be retained.
• 😀 To understand logistic regression better, repeated study and practice are essential due to the complexity of the process and interpretation of results.

Q & A

What is the first step in conducting a logistic regression test according to the script?

The first step is variable selection. The analysis begins by performing a univariate analysis using Casper to identify variables that are significant (with a p-value below 0.25) for inclusion in the multivariate model.

What should be done if a variable has a p-value greater than 0.25?

If a variable has a p-value greater than 0.25, it can still be included in the multivariate model if it is considered important based on the substance of the variable, which means its relevance based on theory and previous research.

What is the difference between causal and predictor research concepts?

Causal research focuses on determining the direct relationship between variables (e.g., smoking and coronary heart disease), whereas predictor research aims to identify which independent variables have the most influence on a dependent outcome (e.g., identifying risk factors for coronary heart disease).

How does the script suggest handling variables with a significance value below 0.25?

Variables with a significance value below 0.25 are tested using either backward or forward selection methods to determine which variables should remain in the model. Backward selection involves testing all variables initially and removing them one by one, while forward selection starts by adding one variable at a time.

What happens if removing a variable from the model leads to a change in the results greater than 10%?

If removing a variable results in a change of more than 10% in the model’s output, that variable should remain in the model. If the change is below 10%, the variable is discarded from the analysis.

What is the importance of checking changes in the odds ratio during the logistic regression analysis?

Checking the changes in the odds ratio is essential to understand the impact of each variable on the model. If a variable causes a significant change in the odds ratio (above 10%), it indicates that the variable has an important influence on the model and should be retained.

What example is used in the script to explain logistic regression?

The script uses the example of obesity and various lifestyle factors, such as soda consumption, gender, and physical activity, to demonstrate how logistic regression is conducted and how different variables are tested for their relationship with obesity.

How is the concept of confounding variables explained in the script?

Confounding variables are those that might affect the relationship between the main independent variable and the dependent variable. In the example, factors like gender and physical activity are tested as potential confounders that may influence the relationship between soda consumption and obesity.

What statistical methods are mentioned in the script for evaluating the logistic regression model?

The script mentions using significance tests, odds ratios, and model fitting methods (such as backward and forward selection) to evaluate the logistic regression model. The changes in output values are also analyzed to determine whether variables should be retained or discarded.

What is the final step after conducting logistic regression analysis?

The final step involves calculating the odds ratios and checking for any significant changes in the results. If a variable causes a significant change (above 10%) in the odds ratio, it is kept in the model; otherwise, it is discarded.