Sampling Best Practices

(Adapted from Crafton Hills College)

Why should we Sample?

If your program has a large number of students or if the artifacts take a long time to review, sampling is more feasible than assessing the entire population.

Census vs. Sampling

Assessing the entire population is called a census, and it is best used by small programs (5 to 10 students). A sample is a portion of the population.

Determining Sample Size

The university sampling policy is 10 students or 10% of the students, whichever is greater. However, there may be circumstances when other sample sizes may be favorable.

The size of the sample depends on two factors, which must be kept in mind when making sampling decisions:

1. The length and complexity of the assignments

Programs that have a very long, complex artifact should use a smaller sample size. Programs that have a short, simple artifact should use a larger sample size.

2. The number of faculty/staff members serving on the panel

Assessment panels should be used to evaluate the artifacts independent of the instructor.

The number of raters on the panel affects the number of artifacts that can be evaluated each semester or year. Programs with a small panel (4 members or less) should evaluate a smaller number of artifacts. Programs with a large panel (5 or more members) have the resources to evaluate a larger number of artifacts. 

Common Types of Sampling

There are a variety of sampling methods. Simple random, stratified, systemic, and cluster sampling are examples of four common and appropriate sampling methods for institutional assessment activities.

Sampling should be representative of all modes of delivery and locations.

If you are not selecting your sample from all course sections, your sample should include both full and part time faculty sections. Selecting sections based on faculty volunteers, should be avoided.

A. Simple Random Sampling

Randomly select a certain number of students or artifacts. The students or artifacts should be randomly selected from the entire population in a way that each student or artifact has an equal chance of being selected.

Example: You have 100 students in your program who have completed the mandatory artifact. You want to sample 20% of your artifacts (20 artifacts). Therefore, you randomly select the 20 students or artifacts without any order or plan. Random sampling can be done with a random numbers table, by random number generators (computerized), or by selecting from a hat.

B. Stratified Sampling

Students are sorted into homogenous groups and then a random sample is selected from each group. This is useful when there are groups that may be underrepresented.

Example: In a program that has few female students, it may be desirable to ensure they are represented in the sample. Therefore, all students are sorted by gender and a sample is selected from each group.

C. Systematic Sampling

You select the nth (e.g. 7th, 9th, 20th) student or artifact from a list.

Example: You have an alphabetical listing of all 100 students who have just completed your program. You want to sample 20% of your student population (20 students). Therefore, you go through the list of100 students and pick every 5th student as you move down the list.

D. Cluster Sampling

If you are using cluster sampling, the selection of sections should be random and should include both full and part time faculty sections.

You randomly select clusters or groups (e.g. classes or sections), and you evaluate the assignments of all the students in those randomly selected clusters or groups.

Example: The artifact that represents an SLO in your program is a paper that is produced in the highest level course of your program, and there are 15 sections offered a semester. Each section has 30 students. You would like your sample to be 20% (90 students) of your overall student population (450 students) for one semester. Therefore, you randomly select 3 sections, and you evaluate the assignments of all 90 students enrolled in those 3 sections.