Hierarchical Linear Regression (SPSS STEPS)

greetings it is I the disembodied voice of SPSS and I will be your host for this video today so in today's video we'll be looking into how to run a hierarchical multiple linear regression using SPSS and at the end of the video I'll also show you how to run some of the plots for you to check the assumptions of multiple regression so without further ado let's take a look at the data we have today so we have a few variables here advertising budget album sales number of plays on radio and attractiveness of band so the main goal here is to see how album sales can be predicted from the other from the predictor variables so you can take a look at the data view to see some numbers if you want now to run a regression you would go to analyze and then click on transfer is a regression then click on linear linear regression so the first thing you want to do is to slot in our variables in a rightful place the first of which is album sales which is your outcome variable so slap that into your dependent box next we want to slug it our predictor variables but remember in hierarchical regression we want to enter our variables sequentially starting with the most important variable so you first predictor that we want to put in is advertising budget because money makes the Guru Guru so yeah so put up into your independent box then you want to start in your second parable but before that click on next so that SPSS will know that you are entering it you're entering your new variable in the second model so the next most important variable here is number of plays on radio so slot that in finally we want to slot in our third variable so click on next and then put attractiveness of band into the third model okay so that's done now we want to select a few statistics that would be useful for us so click on statistics now click on confidence intervals because all of this could use of confidence every now add a next we want to click on r-squared change descriptives part and partial correlations which will give me yours from which you can get your semi partial R square which is a measure of the effect size then you can click on coding narrative Diagnostics to check for multicollinearity as well as Durbin Watson's because anything with the fancy name must be useful but Emma Watson's will let you test for the independence of error assumption with that button you click on continue and then just click on OK and then expenses will immediately vomit out all of these tables for you to examine so we won't be interpreting any of these tables today because the main goal of this video is to show you multi branch of regression but just a quick note now because hierarchical decryption can be a bit confusing we want to make sure that we've entered a predictors correctly so one way to do so is to look at your model summary now your model summary here shows you that you have three models which is which is correct but we want to know what are in these models in the first place and you can do so by checking these I don't really call these captions but anyways so you consider the first model contains advertising budget which is the first predictor the second model will contain advertising budget and number of plays on radio and finally the third model will contain all three of our predictor variables so we've entered a predict that's correct okie-dokie we've run aggression successfully but now we want to produce the plots which will allow us to check for the assumption of homoscedasticity linearity and normally distributed residuals to do so we'll need to go back to our DW and rerun regression with a few more options so you can go to analyze regression and then linear and then you get this window again but it is 20/20 ladies and gentlemen we have no time to click on both of the buttons so let me give you a shortcut go to this button over here click it and you'll see that it will show you a list of the latest statistical test that you have run so click on linear regression and you get this button get this window immediately so all of the variables are already in their place what you want to do is to click on plots now do not be too confused by these confusing dreams what I want to guys to do is to put this Z pred into the x box over here and I'll put this Z seed into the white box over here so these parables are actually just your standardized predicted variable and your standardized residuals so for the purpose of this video you just want to make sure that you sort them into their place or you can produce the plots after that you want to click on histogram and you can click on normal probability plot if you want but I want you guys to also click on this produce all partial plots because they will show you scatter plots are showing you the relationship between your outcome variable and each of the predictor variables separately once you're done click continue and then click OK then you'll get the tables as per before but below here you'll see several of these charts so this histogram will give you a visual inspection of whether or not you have met you some giorno melody and then we have this scatter plot over here which is your standardized particular value against your standardized residuals this will show you whether or not the assumption of homoscedasticity has been met and you just need to look at the plot and see if there's some kind of funneling in the shape in the trend of the plots you can see here that the data points are mostly spread about randomly there's no clear funnel shape here so we can say that the assumption of homoscedasticity has been that as you scroll down below you see partial plots which are basically plots of the dependent variable against your predictor variables separately of course and you can check the assumption of linearity here you can see how the relationships are mostly linear in the first two plots this plot is a big beard kind of like me but yeah so that's how you would check for linearity so yes so that's how you check the assumptions from the plots overall and this marks the end of our video so yes I hope you've benefited from my education 30 well

Channel: Lord Pusheen

Views: 2,037

Rating: 5 out of 5

Keywords:

Id: 9uo-Ug47jys

Length: 6min 25sec (385 seconds)

Published: Thu Jan 02 2020