The experimental farm is testing three different types of industrial fodder for cows/beef. It wants to know whether giving different types of fodder results in different fat content in the animal. You, as a professional statistician, were employed for this experiment. You collected a sample of randomly chosen 30 cows; randomly divided them into three groups, and assign type A fodder to group one; type B to group two and type C to group three. After having given the fodder for one year, you measured the average fat content for each group. The ANOVA test showed a significant difference in fat content across three groups. You, however, noticed that the animals are falling into different categories of breeds (Charolais, Danish, Devon, Holstein, and Kobe) in different proportions. How might this affect the study results? Suggest how the design for the study could have been improved.
Re: statistics problem
The key thing you should be asking is whether the nature of the dogs gives some kind of bias to the variables you are measuring and in particular the response you are using either as a single variable as some kind of regression model.
Can you construct an argument to say that the categories of breeds would bias the fat content in some way?