From the linear model we find expected u

Mean: 28679.63452 ± 2*520.5117 => (27638.61112;29720.65792)

Individual: 28679.63452 ± 2*2382.337 => (23914.96052;33444.30852)

4. 
   The assumption of normality of the disturbance is tested by several tests: qnorm,
   

Qnorm test shows how normally our u_i is distributed by the discretionary illustration in the graph, which is normal:

The Jarque-Bera Test shows whether sample data is matching skewness and kurtosis with normal distribution:

The probability is 0.3338, which is higher than 0.5, meaning than we accept H₀, and say that there is normality.

00006550