The concept of sampling distribution is not easily understood by university students of agricultural and environmental sciences. Although students can define correctly the errors involved in hypothesis testing, they show lack of a deep understanding of the subject when they have to draw conclusions. To strengthen the learning of these concepts, a practical assignment was planned in the second course of Statistics. Starting from an experimental situation, students had to design their own experiments by randomizing four varieties of canola in twenty plots. In the following class, they received the results of the “virtual sowing”, with data provided by the teaching assistant. The data were simulated from normal probability distributions. Students analyzed their own data and the variability of results was commented. It was also discussed that a type II error does not imply a mistake. This study discusses the results of the teaching and learning experience.