Interpreting Odds Ratio with Two Independent Variables in Binary Logistic Regression using SPSS

hello this is dr. Grande welcome to my video on interpreting the ratio in logistic regression when you have multiple independent variables as always if you find this video helpful please like it and subscribe to my channel I certainly appreciate it I have here in the SPSS data editor fictitious data and I have four variables one is an ID variable for participants in the study the second is assessment this is a continuous variable and I have two categorical variables one named license and another named admission so when working with a binary logistic regression we're going to have one dependent variable and that dependent variable have two levels in this case the dependent variable is admission and let's assume this is referring to admission in a brief counseling skills training program we want to determine assessment and license these two variables how they contribute to membership in the admitted level of the independent variable admission and the not admitted level so are these two variables predictive of being a member in one of these categories over the other so assessment could be a counseling skills assessment with a higher score representing a better developed set of counseling skills and license would just be whatever jurisdiction that counselor is in it would be the license status so a counselor would either be not licensed or licensed so here we're only recognizing two levels we're not including any type of graduate or associate license either some of these license to practice independently or they're not that's all we're considering for this variable so we want to measure the contribution of these two independent variables on this dependent variable and interpret the output in SPSS which among other output is the odds ratio before we move to the analysis want to cover a couple of points one is a logistic regression weather and like in this case a binary logistic regression or a multinomial logistic regression which would be if the dependent variable had more than two levels we have assumptions that have to be met for these statistics in this case I'm not going to cover those assumptions because I'm focusing on the odds ratio but it is important to be aware that this statistic and the multinomial logistic regression have assumptions also the coding for each of these variables moving to the variable view you can see that license and admission are both nominal and here's the coding for them not licensed 0 equals not licensed 1 equals licensed for the license variable and for the admission variable 0 equals not admitted and 1 equals submitted to see this in the editor back the data view there's this button up here on the ribbon a 1 you see the zeroes and ones are displayed and press it again and you have these labels so now moving to the analysis to conduct a binary logistic regression and going to go to analyze regression binary logistic and the dependent variable in this case I mentioned admission the dependent variable and I have two independent variables and notice here the referred to as covariates so it'll be the assessment independent variable licensed independent variable under options I'm going to add classification plots and the confidence interval for exp beta let's continue and then click ok to conduct the analysis so again I'm not going to focus on the assumptions here but rather move to the odds ratio and before I look at the variables in the equation that's where the odds ratio is I'm just going to move down to the end here with this graph and note here that predicted probability is of membership for admitted so this is important admitted or that helps us to interpret the odds ratio so we have these two independent variables loaded in this table variables in the equation assessment this variable is continuous and licensed this variable is dichotomous only has two levels and if we look at the exp beta we can see for assessment the value is one point zero three four and for license it's three point one five four so what do these values mean so these values are not probability probabilities rather they are odds ratios so the way we interpret this I'm going to start with the license variable the categorical the dichotomous variable here this value is three point one five four this tells us that if a participant in this case a counselor is in the licensed category if they're licensed versus non licensed there are three point one five four times more likely to be admitted into this brief training program then if they were not licensed so it's in this case licensed versus not licensed that's what the odds ratio is referring to so they're over three times as likely to be admitted than non licensed counselors so what about the assessment variable remember assessment is not a dichotomous variable it's a continuous variable so how do we interpret the odds ratio for that it's the same way except we have to consider that this variable is not dichotomous we're not looking at just two levels it's continuous so we're looking at this incrementally so with this odds ratio here is saying is that for every one unit that we move up in assessment so is this variable increases incrementally by one point the odds of being admitted increase by one point zero three four times we could also look at this as it's three point four percent more likely one point zero three four three point four percent more likely similarly looking at the licensed and a pet variable we have here three point one five four this would be two hundred and fifteen percent more likely if it was one point one five would be fifteen percent two point one five one hundred and fifteen percent so at three point one five it's two hundred and fifteen percent more likely so a key odds ratio in these variables in this table would be one so because these odds ratios are both above one the being admitted level of the dependent variable is more likely as we see change here in the case of assessment incremental change in the case of license moving from not licensed to license if these odds ratios were below one we would interpret it as less likely so instead of one point zero three four if this value were point nine nine we would say this one percent less likely to be admitted as you moved up in the assessment variable is removed up by one point and if we had a ratio for license it was 0.5 we would say that a licensed counselor was 50 percent less likely to be admitted into the program than a non licensed counselor so it's important to keep in mind how the value 1 is interpreted here for exp beta if you had a value of exactly 1 that would mean that the odds are the same so to be no difference as you moved up in points an assessment and no difference as you looked at non licensed versus licensed I hope you found this video on interpreting the odds ratio with two independent variables to be helpful as always if you have any questions or concerns feel free to contact me up happy to assist you

Channel: Dr. Todd Grande

Views: 72,056

Rating: 4.9274611 out of 5

Keywords: SPSS, odds, odds ratio, ratio, expb, variables in the equation, output, binary logistic regression, multinomial logistic regression, logistic regression, regression, dependent variable, independent variable, two independent variables, independent variables, continuous, continuous variable, scale, dichotomous, exp beta, exponentiated, beta, confidence interval, outcome variable, dichotomous outcome variable, variable, group membership, predictor, covariate, statistics, counseling, Grande

Id: Y56BDHt0uXc

Length: 9min 33sec (573 seconds)

Published: Sat Nov 26 2016