Binary Logistic Regression using SPSS :- by G N Satish Kumar

in this session we will be discussing about binary logistic regression logistic regression is used when dependent variable is categorical and independent variable can be either categorical ordinal or even scaled at up and when we say binary logistic regression that dependent variable is exactly two levels then only we use a binary logistic regression if the dependent variable is categorical and more than two levels then we must use multinomial logistic regression in this video we will see about binary logistic regression an example I am using is alcohol consumption in this example I have a dependent variable which is alcohol consumption which is binary the question asked is do you consume alcohol the respondent one means no to consumption of alcohol and they are for dependent independent variables are their gender which is nominal data age self-concept and anxiety score this for this three age self-concept score and anxiety score or scale datas agree now I want to create a model which adequately fits the data and differentiate alcohol consumption I will start the analysis and in this analysis we'll be seeing about the model the fitness of the model and choosing the right model and we will be discussing how these independent variables are varying the dependent variable I will start the analysis for doing binary logistic regression analyzed in this we go for regression in regression we go for binary logistic in this binary logistic the dependent variable is alcohol consumption I am taking as a dependent variable the gender age self-concept score and anxiety score are independent they are taken as covariates and the method i want to take is forward method step method this is what is forward method is the system will create different models it creates a first model without any independent variable and the next model it adds the independent variable which is having more impact on consumption of alcohol and it goes on adding step after step this method of adding variables in each model is curved forward the reverse is backward in backward method the system consider all independent first and it removes one variable after another variable so in this session we will be discussing about forward method I am taking forward lr e okay so the dependent variable is alcohol consumption the independent variables are gender age self-concept score and anxiety score and the method I am going to use is forward method in forward method the system takes each independent variable one after another and creates a different model and we are going to check the model which is going to have more fit in order to adequately fit the data okay I am for selecting categorical I have one data independent data that is gender I am taking gender as categorical covariate now gender has male and female male is 1 female is 2 sister will consider this male and female in the analysis okay I am going to go for save in save I am going to select classification plots and host mayor lemon Shaw goodness of fit this Hosmer lemon Shaw goodness of fit we'll see how far the model is fit classification plot explains about the hit ratio let me say continue and now I am saying OK now before I say ok I just take categorical and taken gender into categorical covariate and in the options I have selected classification plot and the goodness of fit now this is the output screen now here you can see total cases are 392 cases are there respondents are there and dependent variable which we are calling as alcohol consumption wave no means 1 0 and s is 1 and we have become for categorical variable gender male and female is there male is 1 and female is 0 so males are 248 and females are 144 members are there now this is a beginning block in the beginning block the system will not consider any independent variables now if we see this will understand that do you consume alcohol observed and predicted see this is not anyway good overall percentage of prediction is only fifty seven point nine percent okay so the classification table we understand that without any independent variable we are not able to fit the data let us go to block one with forward step method in this first let me see about the model fit the model fit is seen with the help of Hosmer lemon shortest we have a hypothesis to test the model the null hypothesis says the model adequately fits the data there is a null hypothesis the model adequately fits the data if the significant value is less than 0.05 we reject it in the sense step1 step2 we are rejecting the adequacy of the model but if you see step 3 it is more than 0.05 so we accept the null hypothesis that the model 3 the step 3 is adequately fitting the data so this is how we are going to come the first point of confirming the model we have a 0 model where we don't have any independent variable and in each step the model will add one variable after another variable that is what we are going to discuss now here lemon saw Hosmer and lemon shout test says that step 3 model 3 is fitting the data and if you come from model summary you can see that we have our square now in linear regression we talk about our square but when you talk about logistic regression we need to consider pseudo R square statistics the pseudo R square statistics measure the variability in the dependent variable that is explained by the logistic regression model more the R square pseudo a square better the variation is explained means these values should be more this values will lie between 0 to 1 and more the value that is more better if you see step 1 it is neglected and R square if you see it is 0.65 seven two it is zero point 7 8 2 and the 3 it is 0.8 means the pseudo r-square value is increasing so we can come to conclusion saying that model 3 is showing more variation in the dependent variable than other two models so we have seen the model fit and we have seen the variation in the models that is explained model 3 is explaining better than model one and model two now the third one is selecting the right model we'll see classification table how much percentage is correctly explained we have three models step one step two step three in each model the system is adding the variable independent variable in step one it is only giving 86.7% step two ninety point three percent and the step three 91.6% it mean that step three is having more percentage of correct explaining the data than the other two steps now let me see in which step what variable is added in step one self-concept is first added this is having more impact in step two with self-concept system added anxiety score also and in step three system added age so step three contains three independent variables age self-concept and anxiety but where is a gender yeah variables not in the equation if you see step three the gender is given in step one we do not have we do not have gender age and anxiety in each step the system is adding the remaining variables but when you come for step three system has a removed gender one now we can conclude that in forward method add a for independent variable system started adding one independent variable after and the independent variable and when independent variable added in first step it is significant if significant value is less than 0.05 it is significant step two it has added anxiety score it is also significant step three H is also added but system is not added gender means a gender is not having adequate impact on the data gender is not having anything to do with alcohol consumption it is with age self-concept and anxiety score and this step three that is model three is having better in predicting the correctness of percentage of correctness of alcohol consumption is 91.6% and the model fit is also is significant and it shows a pseudo asked you a statistic is also increasing the variation is also increasing it is from zero point six five seven it has slowly increases zero point eight zero to two so this is a model which we have formed the alcohol consumption is impacted by age impacted by anxiety and the self-concepts course and not impacted by gender

Channel: My Easy Statistics

Views: 74,947

Rating: 4.5911112 out of 5

Keywords: Binary Logistic Regression, Binary logistic regression using spss, logistics regression spss, logistics regression in r, binary logistic regression spss interpretation, binary logistic regression, binary logistic regression spss, binary logistics regession examples, how to do structural equation modeling in spss, amos software tutorial, structural equation modeling spss pdf, structural equation modeling spss youtube, introduction to structural equation modelling using spss and amos

Id: sejWXOLZwpk

Length: 16min 34sec (994 seconds)

Published: Fri Aug 26 2016