Statistical modelling/Milk offtake

From the parameter estimates the fitted equation can be written as:

TOTALM_i = 2.136(±0.112) - 0.8218(±0.0971)L₂ - 0.0511(±0.0169)AGEC_i

where the coefficient for L₂ refers to the difference in intercepts for Assa from Bilisa. It is customary to include standard errors in parentheses. The results show that the intercept is 0.8218 litres lower for Assa than for Bilisa.

Thus, separate regression lines for the two locations can be written:

For Bilisa: y_i = 2.136 (±0.112) - 0.0511(±0.0169) x_i
For Assa: y_i = 1.314 (±0.125) - 0.0511(±0.0169) x_i

The constant term for Assa is determined by subtracting the value 0.8218 from the constant, namely 2.136 - 0.8218 = 1.314. However, the standard error for the constant term for Assa is not so easily obtained. One way is to run GenStat again but first to reorder the LOCATION code so that level 2 is recognised as the first level, i.e. making Assa the reference location.