Homoscedasticity

our next assumption is the assumption of homoscedasticity okay that's a that's a fancy word that statisticians love statisticians love to throw around uh and here's what it means it means that for each set of values of the iv the variance of errors is constant right so so what's that mean again for any given that marker if we consider x and y for any given value of x so we take any given value of x say this one right here there is some spread of error around our expected value of y right so for that value of x there is only one expected value of y and then there is some spread of errors around that expected value for each people right because because different people at this level of x may have different values of y the idea here is that this that spread that variance of uh of y minus y hat the variance of the residuals for any particular value of x should be constant across all values of x and that is homoscedasticity when we have heteroskedasticity what that really means is that we have a moderator problem so we would typically see that we may have this scatter plot going on but it turns out that this is one value and this is some other value of x right so that's a case of of missing a moderator variable and what we call a heteroskedasticity which means that the variance of x is not constant for all or the variance of the residuals is not and the prediction errors is not constant for all values of x and again our estimate is still kind of the same but it's definitely a sign that we're missing a moderator variable

Channel: David Wallace

Views: 255

Rating: 5 out of 5

Keywords:

Id: zOQxMG_HHSE

Length: 1min 58sec (118 seconds)

Published: Sat Jul 18 2020