Presentation on theme: "SADC Course in Statistics Revision of key regression ideas (Session 10)"— Presentation transcript:
SADC Course in Statistics Revision of key regression ideas (Session 10)
To put your footer here go to View > Header and Footer 2 Learning Objectives At the end of these sessions, you will be able to have a good understanding of reasons why a regression analysis may be done have greater confidence in fitting regression models using statistical software and assessing the appropriateness of the model for its intended purpose conduct and interpret a residual analysis to check model assumptions select a subset of explanatory variables from large number of potential ones
To put your footer here go to View > Header and Footer 3 Contents of Session 10 Using a checklist to highlight concepts that need to be fully understood. Revision of key regression ideas using example 2 of Practical 9. Practical work to ensure that ideas learnt can be put into practice. Participants will work in groups and produce a brief report of their key findings and conclusions.
To put your footer here go to View > Header and Footer 4 A checklist of underlying concepts Under what circumstances would you undertake a regression analysis? Can the methods so far discussed be undertaken if the y-response is a binary variable? Can the methods so far discussed be undertaken including x-variables which are nominal or un-ordered categorical variates? What is meant by a simple linear regression?
To put your footer here go to View > Header and Footer 5 A checklist – continued…(2 of 5) How may the parameters of such a model be interpreted? How may the corresponding t-probabilities be interpreted? What is meant by a correlation coefficient? What values can such a coefficient take? Can you interpret an R 2 value? How is it related to the correlation coefficient? What is meant by an analysis of variance? What hypotheses are tested by it?
To put your footer here go to View > Header and Footer 6 A checklist – continued…(3 of 5) How would you interpret the residual mean square in an anova? In a multiple linear regression model, how would you interpret the regression coefficients? After a multiple regression analysis, can you write down the model equation? How would you use this equation to make predictions? What are the two types of predictions you can make? How would their standard errors be calculated?
To put your footer here go to View > Header and Footer 7 A checklist – continued…(4 of 5) What do the t-probabilities of a multiple linear regression tell you? Would a t-test concerning the slope of a simple linear regression model give different results to the anova F-test in a multiple linear regression? If so, when would this happen, and why? What do the t-probabilities of a multiple linear regression tell you?
To put your footer here go to View > Header and Footer 8 A checklist – continued…(5 of 5) How would you select a subset of regressor variables from a set of potential ones that may affect the variability in y? What dangers are associated with using automatic selection procedures? How would you assess the model once the best subset of xs have been selected? What is the purpose of a residual analysis? How would you conduct and interpret results from such an analysis?
To put your footer here go to View > Header and Footer 9 Demonstration to discuss key concepts (with Example 2, Practical 9), followed by practical work …