Stratified Random Sample: Example & Definition
What is a Stratified Random Sample?
A stratified random sample is a population sample that requires the population to be divided into smaller groups, called 'strata'. Random samples can be taken from each stratum, or group.
Example of a Stratified Random Sample
Suppose that you were a researcher interested in studying the income of American college graduates one year after graduation. The population that you are interested in is American college graduates. A population consists of all members of a defined group who possess specific characteristics that you are interested in studying.
You find out that each year there are over 1,750,000 people graduating from college, which means that your population size is almost two million! It's impossible for your research team to collect data from every member of your population, so you decide to collect data from a sample, which is a subset of the population that is used to represent the whole population. You and your research team decide that you want to take a sample of 3,000 American college graduates.
How do you choose your sample? You could choose a random sample, in which each member of the population has the same chances of being selected for the sample.
Assume you chose a random sample of 3,000 college grads. You look at the demographics of your sample participants and find that 2,034 are Caucasian, 832 are African-American, and 134 are Asian-American. You start to wonder if there are any differences in income one year after graduation between the different racial subgroups. You also wonder if the demographics of your sample are truly representative of the demographics of American college graduates. One way of examining these questions is by using a stratified random sample.
Suppose that instead of conducting a random sample, you decide to use a stratified random sample. You look up the demographics of American college graduates and find that 51% of American college graduates are Caucasian, 22% are African-American, 9% are Asian, 8% are Native American, 5% are Hispanic, 3% are Pacific Islander, and 2% are multiracial. You divide your population into strata based off the participant's racial demographics. Here you have seven strata, one for each of the seven racial categories. Using this information, you can conduct either a proportional stratified sample or a disproportional stratified sample.
In a proportional stratified sample, the size of each stratum in the sample is proportionate to the size of the stratum in the population. In a disproportional stratified sample, the size of each stratum is not proportional to its size in the population.
Let's assume that you have decided to conduct a proportional stratified sample. Your next step is to ensure that the proportion of the strata in the population is the same as the proportion in the sample. This means that 51% of your sample needs to be Caucasian (since 51% of the population is Caucasian), 22% should be African-American, 9% Asian, 8% Native American, 5% Hispanic, 3% Pacific Islander, and 2% multiracial. By calculating the proportions, you determine that the racial demographics of your 3,000 study participants should include:
- (3,000 x 0.51) = 1,530 Caucasians
- (3,000 x 0.22) = 660 African-Americans
- (3,000 x 0.09) = 270 Asians
- (3,000 x 0.08) = 240 Native Americans
- (3,000 x 0.05) = 150 Hispanics
- (3,000 x 0.03) = 90 Pacific Islanders
- (3,000 x 0.02) = 60 multiracial
You can see that the demographics of the stratified random sample are very different than those of the random sample.
Now that we have our population broken into strata and we have identified how many study participants we need from each strata, the next step is to use random sampling in order to pick the specified number of participants from each of the seven strata. In other words, we randomly select 1,530 Caucasians from all of the Caucasians in the population. We randomly select 660 African-Americans from all of the African-Americans in the population, and so forth.
In order for our samples to be truly random, three conditions must be met:
- Each member can only be assigned to one stratum.
- Each member of the stratum must have an equal chance of being selected.
- The selection of one member of the stratum cannot influence the selection of another member.
Once we have chosen our study participants from each stratum, what we have is a stratified random sample.
Advantages of Stratified Random Sampling
Advantages of using a stratified random sample include:
- The stratified random sample is more representative of the actual population than a random sample because it follows the same proportions of the population.
- Stratified random samples give more precise information than a random sample.
- Because of the improved precision, you don't need as many study participants as you would with random samples and other sampling methods. This, in turn, saves money.
- Dividing the population into strata allows researchers to draw conclusions not only about the general population, but also about the subgroups of the population.
Disadvantages of Stratified Random Sampling
Disadvantages of using a stratified random sample include:
- It can be difficult and time-consuming to select relevant strata groups for your research study.
- It can require more time to analyze stratified data than it would take to analyze data that has not been categorized into groups.
Lesson Summary
In order to select a stratified random sample, members of a population must first be divided into strata, then randomly selected to be a part of a sample. Stratified random samples can be either proportional or disproportional to a subset's representatives in the general population. The advantages of stratified random samples include increased precision and lower costs. The disadvantages include difficulty in selecting appropriate strata and analyzing the results.
Learning Outcomes
When you are finished, you should be able to:
- Recall what a stratified random sample is
- Discuss the process to produce a stratified random sample
- State the advantages and disadvantages of using stratified random sampling
To unlock this lesson you must be a Study.com Member.
Create your account
Using a Stratified Random Sample with Different Strata
Suppose that, just as in the lesson, you wish to study the income of American college graduates one year after graduation. The demographics are as follows: 51% of American college graduates are Caucasian, 22% are African-American, 9% are Asian, 8% are Native American, 5% are Hispanic, 3% are Pacific Islander, and 2% are multiracial. Suppose also that in looking up the demographics, you find that 52% of American college graduates are female and 48% are male and the racial distribution across the genders is equal to the racial distribution overall (so 9% of the females are Asian and 9% of the males are Asian, etc). In each example below, state how many individuals should be selected in each strata if you want to survey a total of 5000 college graduates.
Examples
- strata defined by race/nationality
- strata defined by gender
- strata defined by both gender and race/nationality
Solutions
- We would want the demographics of our sample of 5000 graduates to match the demographics of the overall population of graduates. So, we would need (5000 x 0.51) = 2550 Caucasians, (5000 x 0.22) = 1100 African-Americans, (5000 x 0.09) = 450 Asians, (5000 x 0.08) = 400 Native Americans, (5000 x 0.05) = 250 Hispanics, (5000 x 0.03) = 150 Pacific Islanders, and (5000 x 0.02) = 100 multiracial.
- We want the gender representation in our sample to match the gender distribution of the overall population of graduates. So, we would need (5000 x 0.52) = 2600 female graduates and (5000 x 0.48) = 2400 male graduates.
- We want both the racial and gender demographics of our sample to match that of the overall population of graduates. From the second example, we know we want 2600 females and 2400 males. For the females, we want (2600 x 0.51) = 1326 female Caucasians, (2600 x 0.22) = 572 female African-Americans, (2600 x 0.09) = 234 female Asians, (2600 x 0.08) = 208 female Native Americans, (2600 x 0.05) = 130 female Hispanics, (2600 x 0.03) = 78 female Pacific Islanders and (2600 x 0.02) = 52 females who are multiracial. For the males, we want (2400 x 0.51) = 1224 male Caucasians, (2400 x 0.22) = 528 male African-Americans, (2400 x 0.09) = 216 male Asians, (2400 x 0.08) = 192 male Native Americans, (2400 x 0.05) = 120 male Hispanics, (2400 x 0.03) = 72 male Pacific Islanders and (2400 x 0.02) = 48 males who are multiracial.
Discussion
In your examination of the income of college graduates a year after graduation, you want to see the effects of gender and race on income. Which strata above would you choose to give the most accurate representation?
Guide to Discusssion
The best strata to choose would likely be the third option - dividing by both gender and race - so that comparisons can be made based on gender, race, or both.
Register to view this lesson
Unlock Your Education
See for yourself why 30 million people use Study.com
Become a Study.com member and start learning now.
Become a MemberAlready a member? Log In
Back