Please enable JavaScript.

Coggle requires JavaScript to display documents.

Logistic Regression (Simple (Data exploratiiion - 2x2 table (IV cat) (SLR,…

- - - - Checking assumption
        
        Interpretation/Conclusion/Presentation
        
        Formula regression: Log (odds) = B (regression coeefcient - constant) + B (regression coeeficient IV)
        
        Odds ratio = Exp (B) = Exponentiation of the regression coeeficient
        
        Exp (B) if by %, <1 = less likely (have to 1- Exp (B))
        
        Exp (B) >1 = more likely (no need to use %)
        
        "Interpretation: Subject with IV has ...more/less like@has odds .... times higher or lower compared to another IV". (95% CI, ... , ...., pvalue ...)
        
        Presentation table: include Variables, coeffiicient (B), SE, Odds Ratio, P value, CI 95% + R2 value cox and nagelkerke
        
        R2 Cox and Nagelkerke (assesing model fit)
        
        Classification table
        
        Omnibus test (P<0.05) indicate significant model
        
        Hosmer and Lameshow (P>0.05)
  - - - Checking assumption
        
        Interpretation/Conclusion/Presentation
        
        Formula regression: Log (odds) = B (regression coeefcient - constant) + B (regression coeeficient IV)
        
        "Interpretation: Respondent with 1 unit increase in (IV cont) has .... (%) higher/lower odds to have diabetes
        
        Presentation table: include Variables, coeffiicient (B), SE, Odds Ratio, P value, CI 95% + R2 value cox and nagelkerke
        
        Omnibus test (P<0.05) indicate significant model
        
        Hosmer and Lameshow (P>0.05)
        
        R2 Cox and Nagelkerke (assesing model fit)
        
        Classification table
- - - - SLR. buat each IV and DV simple linear regression first. look for significant or not first. Iv npt sig np need to go into mlr
        
        MLR
        
        Checking Interaction
        
        Checking Multicollinearity (correlatoion matrix, R < 0.8) between all IV
        
        Checking Assumption
        
        Interpretation
        
        5 more items...
        
        ROC
        
        1 more item...
        
        LInearity Log (y) and x
        
        Hosmer Lameshor
        
        1 more item...
        
        Omnibus test
        
        Random sample
        
        Independent sample
        
        Linearity log (y) with x
        
        R2 residuals Cox and Nagelkerke
        
        Classification table
        
        1 more item...
        
        To remove interaction formula it any of the primary IV become no significant. Primary IV is superior than the secondary iteraction IVs
        
        Between all iv, if interaction significant and does not change the level of significant in the individual (primary) IV, then will put in the interaction in the report
        
        Variable selection method
        
        Enter
        
        All iv into model
        
        Forward Selection
        
        sequentially entered (most significant/largest correlation IV first)
        
        Conditional (stepwise selection method with entry based on Wald score and remove based on LR score
        
        Likelihood Ratio (LR) (stepwise selection method with entry testing on the significane od the score statistic)
        
        Wald (stepwise selection method with entry testing on the significane od the score statistic)
        
        Backward Elimination
        
        all IV into model first, then sequentialy removed from the least significant first
        
        Conditional (stepwise selection method with entry based on Wald score and remove based on LR score
        
        Wald (stepwise selection method with entry testing on the significane od the score statistic)
        
        Likelihood Ratio (LR) (stepwise selection method with entry testing on the significane od the score statistic)
        
        The aim of variable selection:
        
        Model with most model fit
        
        Biological parsimanious model
        
        Include biological/clinical plausibility