91Ó°ÊÓ

Selection bias happens when the sample chosen for a study does not accurately represent the whole population. In the exercise, the health teacher only takes into account the weights of the students attending the 9 A.M. class. This specific group may have different attributes compared to the overall student population. For example, a 9 A.M. class might predominantly consist of early-risers, who could have different health and lifestyle habits than students who prefer later classes.
Such characteristics make this group unrepresentative. Selection bias thus causes incorrect conclusions because the results are skewed towards a specific subgroup instead of the entire college student population.

A representative sample accurately reflects the entire population’s attributes. If the exercise's goal is to understand the weight of all college students, the sample must include a diverse array of students with various backgrounds and schedules.
Therefore, it should consist of students from multiple classes, majors, and time slots. By ensuring such diversity, the teacher minimizes the risk of sampling bias and obtains more reliable, generalizable results.
This makes the findings applicable to all college students, not just those in a specific 9 A.M. class.

Stratified sampling is a method aimed at improving the representativeness of a sample. Here, the entire population is divided into different 'strata', or groups, based on relevant criteria such as class timings, majors, or year of study. Then, samples are drawn from each of these strata proportionally. In the context of the exercise, this means the health teacher could categorize students based on these different segments and randomly select a proportional number of students from each segment to form a balanced sample.
This process ensures that every significant group within the population is represented fairly. As a result, it addresses the problem of sampling bias and offers a more accurate reflection of the entire student body.
In essence, stratified sampling provides a structured approach to achieving a broad and comprehensive view of the population, eliminating the risk of overrepresentation or underrepresentation of any group.