Using random sampling methods alone does not reduce variability in a sample; they only reduce bias. The best way to reduce variability in an unbiased sample is to take a bigger sample—the bigger the population of the sample, the less widespread the results will be. It is important to remember that if there is bias present in a sample, a large sample population size (n value) will not be enough to overcome the bias.

Voluntary response bias usually results when those sampled are volunteers who select themselves.
Voluntary response bias can cause results to over represent those who have a strong opinion.

This is an example of undercoverage because the researcher did not speak to people who were not affiliated with a university. Therefore the researcher did not adequately sample all members of the overall population (people in the United States).

The correct answer (random sample of 300 students) includes a random sample which will reduce in bias and also is given to be done in person which reduces completion bias.  It is also taken from the whole population of students and is a large sample which makes it a good representation of the population.  Sending out a questionnaire has completion bias because the only people that will send it back will have strong and possibly biased beliefs already.  The hot line has the same effect as those that have strong and biased opinions will be the ones to do it.  Pulling students aside as they walk by is not a truly randomly constructed sample and using only 10 students is not a good representation of a whole college.  Selecting a sample of just athletes is very biased as they will not want their programs cut.

A simple random sample is obtained by randomly selecting individuals from a target population. Each individual in the target population (i.e. all students at the high school) should have an equal chance of being selected. Only one answer choice gives each high school student an equal chance of being selected: choosing  students at random from the high school. 

In order for there to be a stratified random sample, the target population must be split into different groups (i.e. grade levels). The sample population must be selected at random from each of these groups (i.e. choosing  students from each of four different grade levels or groups). The other examples, although random, are not specifically stratified in their sampling methodology. 

Stratified Sampling is the method of randomly selecting subjects from various stratum, or subsets of the population. In this case, there were 4 stratum from which participants were randomly selected.

Random sampling is a method in which every individual has an equal opportunity of being randomly chosen to participate in a study.

Cluster random sampling entails choosing from pre-formed "clusters"-- such as schools or hospitals-- and randomly selecting one of the clusters.

In order for cluster sampling to be present, the entire population must be divided into homogeneous groups (known as clusters). For sampling purposes in this problem, clusters must be chosen as a simple random sample. There is only one answer choice that fits this pattern.