Multiple Regression Interpretation in Excel

hi everyone its Kevin just going to show you a quick example of interpreting a multiple regression output from Excel right here you see a data set the variable I'm trying to explain would be sales I'm trying to understand how well these three variables will explain sales and those three variables my X variables my explanatory or independent variables are airplay how often is the song played per week the attractiveness of the band that's on a scale of one to ten and then finally how much money is spent advertising for this band again it's always important to pay attention to the units we see advertising dollars in sales are in units of 1000 so in cell a2 they're when we see three hundred and thirty that's three hundred and thirty thousand dollars worth of sales and airplay again that's just a basic unit one two three four five and our attractiveness is on a scale of one to ten now let's take a look at the output now first and foremost we want to say something about the model so if we look at the R square we see that it's sixty six percent and so what that means is these three variables airplay attractiveness and advertising money explains about two-thirds of the variation in the dependent variable and again that would be sales is the model significant yes we see our F statistic has a p-value that's well below 0.05 or 0.01 for the 95% and 99% confidence levels so we can say that the model is statistically significant but let's look at our actual variables and their p values we see that the p-values for airplay attractiveness and advertising dollars are all statistically significant at the 99% level and so 99% confidence level and that's very good that allows us then to actually look at these coefficients and try to say something about the model from a predictive standpoint so we see that our intercept is a negative twenty six point six one of course that's where the line of best fit is intercepting with the y-axis and we see airplay is three point three seven a positive coefficient positive coefficient on attractiveness and a positive coefficient on advertising dollars now let's think carefully about what each of these mean individually first and foremost for airplay if we have a positive coefficient of three point three seven that means for each one unit increase in airplay which would be plays per week our sales will go up by three point three seven so again we need to think about sales in its actual unit that's in thousands of dollars so that means that if our airplay goes up just one unit in a week we can expect that to increase our sales by three thousand three hundred and seventy dollars pretty significant even more significant is the attractiveness as we can imagine or as we would expect this would be a positive coefficient so for each one unit increase on this attractiveness scale in other words the better looking a band or an artist is we can expect an $11,000 increase in their sales album sales last but not least when we spend advertising money so for each one unit increase in our advertising dollars which is a thousand dollars we get an additional 80 dollars in sales so not the not the impact that we maybe would have imagined so if a band were to come and say hey I'm not getting the kind of sales for my record in my album that I would like you need to advertise more you might say well actually you're willing to get more plays per week or you're too ugly so might be a crude way of putting it but advertising dollars do make a difference they do have an impact and again that is a statistically significant impact it doesn't just doesn't have the coefficient strength that we would have imagined especially if you're spending $1,000 okay so let's imagine that I want to answer this question below what sales what kind of sales number would we expect if we have 50 a band that had 50 air plays a week an attractiveness level of six modern moderately attractive we might say and they were spending about 1 million dollars worth of advertising so if that were the case what would we expect and so writing this out our equation is going to be y is equal to negative two six point six one plus three point three seven times whoops 50 plus eleven point zero nine times six plus point zero eight times one thousand again because if I'm spending a million dollars that would be expressed in units of a thousand so it'd be one thousand dollars here would be or one thousand would be the equivalent of 1 million if you were to do the math on that you will get Y is equal to two eighty eight point four three and that's basically like saying for a band that would have 50 airplay as a week is moderately attractive and is spending somewhere around a million dollars we would expect based upon our mod model sales of two hundred and eighty eight thousand four hundred and thirty dollars for this particular band

Channel: TheWoundedDoctor

Views: 396,446

Rating: 4.8821111 out of 5

Keywords: Doceri statistics, excel, multiple regression, statistical interpretation

Id: tlbdkgYz7FM

Length: 6min 32sec (392 seconds)

Published: Mon May 06 2013