The analysis shown here uses a balanced experiment, with equal numbers of observations for each pairing of the factors. What happens to the estimates and their standard errors if that's not the case? Does your software still work? Try it! Delete a few rows of the data and see what happens.
A consumer products company surveys people who recently bought new cars. Each customer is asked to express satisfaction with the purchased car by stating the likelihood (from 0 to 10) of buying another of the same brand in the future. A rating of 0 means no chance of buying another, and 10 means the likelihood is almost certain. The data in this example give the ratings of 220 purchasers of economy, standard, and premium vehicles from Asian, European, and domestic manufacturers.