What is the assumption of 'homoscedasticity' in regression analysis?
The residuals have constant variance across all levels of the predictor variable
The predictor variables are normally distributed
The model has no intercept
The relationship between variables is perfectly linear