WEEK 3: Logistic Regression

Frage	Antworten
What is the difference between Multiple Regression and Logistic Regression?	In Multiple regression the DV is continuous, in Logistic regression the DV is Categorical
When do you use Logistic Regression?	* Used to predict a categorical DV on the basis of one or more IV's IV's can be categorical or continuous DV MUST be categorical
MR or LG question? What are the predictors of frequency of cigarette usage per week?	Multiple Regression
MR of LG Question? What are the predictors of whether a student smokes or not?	Logistic Regression
In Logistic Regression does all data need to be normally distributed?	No LG Does not assume normal data Does not require a continuous underlying distribution.
Logistic Regression provides a number of useful functions such as?	• Measures the variance in a DV like MR • Gives coefficients indicating the strength of various IVs • Gives statistics (roughly) equivalent to R2 • Measures the odds of some outcome in onegroup relative to a reference group
What are the two forms of Logistic Regression?	(Binomial form) – Test is between the presence of some outcome, and its absence i.e smokes, doesn't smoke. (Multinomial form) – Test is between various levels of some factor. ie :0, 1, 2, 3, etc (HFA, AS, Neurotypical)
What is the Standard form of Logistic regression?	Where all variables are entered together
What is Sequential Logistic Regression?	Where variables are entered in blocks as specified by the researcher (based on theory)
What is statistical Logistic Regression?	Variables may enter or leave the equation as components of variance are consumed – Criteria to enter or leave the equation include – Tests of significance of the coefficients (Wald tests n- Z test) – Significance of the estimated R2 – Probabilities of likelihood of contribution
What is Logit transformation?	A transformations allows a line of best fit (curved, s shaped etc) to be applied to categorical data. It transforms the DV (Categorical) so that you can run the analysis because you can't predict a binary outcome, it needs to be continuous.
What is coding in Logistic Regression?	Coding is used for categorical Variables in LR (IV's and DV's) to understand the relationship between IV's and DV. Reference groups are used for meaningful comparisons
What is the Logistic function?	The mathematical way of explaining the line of best fit. Can capture straight, diagonal and asymptotic (bits bunched at ends) portions of the graph.
What do you do if your categorical IV has more than two levels?	You need to recode the variable into k-1 dichotomous variables, where k = number of levels of the IV.
What is the reference group in coding?	The group we are comparing against but don't really care about. Reference group is always represented as a 0.
What are the key assumptions in Logistic Regression?	* Assumes complete independence of error terms * Assumes linearity between continuous predictors or IV's and the Log of DV
What are some of the limitations of Logistic Regression?	* Incomplete information *Complete Seperation
What is complete separation in Logistic Regression?	When the outcome variable (DV) can be perfectly predicted by one variable or a combination of variables. (We need some error otherwise it's unmeaningful)
In Logistic Regression output how do interpret the model overall?	R2 for effect size and X2 Chi square for significance. R2 is calculated using -2LL at baseline compared to the -2LL of the new model.
What is classification accuracy in Logistic Regression?	LR predicts group membership and compares it against actual membership to see if the new model is more accurate in correctly classifying group membership.
What are the problems with R2 estimates provided in SPSS for Logistic regression?	Cox & Snell's R2 ranges from 0 - less than 1, where it should range to 1. Nagelkerke's R2 corrects Cox and Snells range problem but is excessively influenced by sample size. *For both tests- increases in sample sizes tend to inflate effect size.
What does the B weight represent in Logistic Regression? (What is it predicting)	When a logistic regression is calculated, the regression coefficient (B weight) is the estimated increase in the log odds of the outcome per unit increase in the value of the exposure
What is an odds ratio? How do we interpret it?	It predicts the odds of being in a particular group.
What is the Log Likelihood?	Tells us how much unexplained variance there is in the beginning and in the end and then the chi square is the difference between them.

Nächster

WEEK 3: Logistic Regression

Beschreibung

Zusammenfassung der Ressource

ähnlicher Inhalt

	Erstellt von seonapalmer vor etwa 9 Jahre