Mediator in SPSS

we're making another mediator video to try to help explain a concept it can be kind of hairy at times there's no question about it so a mediator in multiple regression we're going to show you how to do this in SPSS so the overall theory about a mediator and a multiple regression model is if you have a predictor more than one predictor which is an IV you have to decide is one of them acting as a mediator in other words is one of these predictors controlling the significance of the model so I think the easiest way is going to be able to show you just real quick what we're doing to do here so is one of these predictors we got IQ scores and study hours as the predictors we're trying to predict their final exam scores okay so is it does iq matter or study hours matter or do they both matter or neither matter that kind of thing so here's the typical model that you see in most books so with this one study hours would be considered a mediator this is always kind of confused me because I was always told that if the mediator was not in the model then you would have not you would not have a significant regression between your other predictor IQ scores and exam scores that's not a hundred percent true using SPSS everybody's best friends so we got our predictors IQ scores and study hours our DVR outcome is exam scores so in order for us to have a mediator the correlation between the first predictor and the DV has to be significant okay so the first thing you do is you run a multiple regression model strictly with one IV and one DV that's called a simple regression so we're going to go ahead and do that now hold on here's the data sheet in SPSS they're all scaled measures you could do ordinal I guess no use scale okay so we got plenty we got like 200 in a sample size so we're going to go to analyze regression linear or DV is our exam scores and we're just going to check IQ first okay that's the the only one we're going to do first see what kind of Statistics we need well we should probably click these just by habit don't need r-squared change Durbin Watson we should check plots I'm not going to do that now okay I'm not going to look for outliers either again I'm just going to stick to the mediator to keep this as short as possible so what we got here is a simple regression one IV one DB that has to be significant okay so let's look at our bucks there's the regression Bucks so there's the means there's the correlations and our r-squared is 0.4 for one which is pretty darn big we have a significant ANOVA so that means we do have a significant regression between our IQ scores and our exam scores and the last thing we're gonna look at is our beta weight okay so the beta weight is pretty big and it is significant let me scoot it over so you can see that significant part it over there you okay so the beta weight between IQ scores and exam scores is significant and large so we got the first step out of the way so we do have a significant beta way between the predictor and the divi the anova was significant so we do have the stage set so now we're going to add in the second predictor to see if it's a mediator or not so let's go back to the SPSS we're going to rerun the analysis regression linear we're going to add in the weekly study hours and again I'm going to skip all this stuff click OK so now let's take a look I should have shown you this first so we added in the suspected mediator we reran the regression with both predictors now so this is how you tell if the second predictor that you put into the model is a mediator or not so if your significant beta between your original predictor and your dvrs was IQ scores in exam scores if your significant beta changes to an N signal in beta and you're newbie right you're suspected mediator has a significant beta then your second IV that you put in there the one with the significant beta is in fact a mediator okay so this model is correct so let's jump back to the SPSS so here's the new regression model with both IDs okay the model summary your r-squared jumped way up to 0.7 94 which means we should probably be looking for suppressors but we're not going to go there so the ANOVA for the second model with two predictors is significant but now here's the money right here down on the bottom let me see if I can't move this up a little bit for you hmm so our original predictor of IQ scores was significant but now it's not got it I'll say that again the original predictor IQ scores is no longer significant the beta is very small the significant is not small enough but now look at the new one that we stuck in there the weekly study hours it went sky-high its beta is huge right negative point-eight that is huge and it is significant therefore we have proved that our mediator is weekly study hours weekly study hours is the mediator in this project I hope it helped mgz out

Channel: Math Guy Zero

Views: 120,440

Rating: 4.8866735 out of 5

Keywords: mediator in spss, mediation, spss, statistics, Stats

Id: pqyH2ik8kX4

Length: 5min 59sec (359 seconds)

Published: Tue Mar 10 2015