THERESA DE LOS SANTOS: Hello.I am Dr. Theresa de los Santos.I am an Assistant Professor of Communicationat Pepperdine University in Malibu, California.In my personal research on emotionin news and social media groups, I primarilyutilize quantitative research methods of experimentsand content analyses.
THERESA DE LOS SANTOS [continued]: As soon as you have a research question or hypothesis,you automatically have to answer the question of who or what,or how many will you study, in order to answer it.In other words, the question we have to ask ourselvesis, who or what do we want to generalize to?This is why knowing the basics of sampling is critical.
THERESA DE LOS SANTOS [continued]: As you are watching this video, keep these key pointsabout sampling in mind.Key point number one, you do not haveto gather data on every single eventin a population of interest to draw reasonable conclusions.Key point number two, there are two main sampling methods,
THERESA DE LOS SANTOS [continued]: probability and non-probability.The type of sampling a researcher usesdetermines their ability to know how likely itis that their results accurately reflect the entire population.And key point number three, samplingis important because of external validity.
THERESA DE LOS SANTOS [continued]: Expanding on the first key point,the process of selecting individual units for studyis called sampling.It works like this.Once a researcher has a research objective,they define the target population.There are no rules for this.Based on the goal of the research,
THERESA DE LOS SANTOS [continued]: the researcher relies on logic and judgmentto determine the population.A population can be defined as a complete set of elements,persons, or objects that possess some common characteristic.Populations can be very broad, such as all American adultsages 18 to 49, or more specific, such as American adults ages
THERESA DE LOS SANTOS [continued]: 18 to 49 with five or more social media accounts.With this, there is a target population,or the group of people or things a researcher ideally wantsto generalize findings to.And then, there is an accessible population, or the groupthat they actually have access to.This may be a subset of the target population,
THERESA DE LOS SANTOS [continued]: such as adults ages 18 to 49 with five or more social mediaaccounts in the city of Malibu.Sometimes, the entire population is small enoughthat the researcher can access and include every memberor object of it in a study.This is called the census.It's like the one done on the entire population of the United
THERESA DE LOS SANTOS [continued]: States every 10 years.But more often than not, the populationis too large for the researcher to attemptto survey all of its members.This is when a carefully chosen sample can be drawnto represent the population.It doesn't always work perfectly.Researchers can sample towards the center of a population,
THERESA DE LOS SANTOS [continued]: or they might accidentally sampleat the periphery of a population.I'll talk about types of sampling error laterin this tutorial.But careful sampling can often accuratelyidentify population characteristics, sometimeseven with very small numbers of events.This takes us to key point number two.Confidence in how accurately a sample represents a population
THERESA DE LOS SANTOS [continued]: increases with the type of sampling methoda researcher uses.There are two main types of sampling, probabilityand non-probability.The first form, probability sample,involves drawing samples to maximize chance,that each event in the populationhas an equal chance of being included.It is often referred to as random or representative
THERESA DE LOS SANTOS [continued]: sampling.This is because probability samplingstrives to obtain samples that are statisticallyrepresentative of the overall populationor are miniature versions of the population.To achieve this, a researcher whowants to collect a probability samplemust have an accurate master list of their whole population.
THERESA DE LOS SANTOS [continued]: Remember this, this is key.A sampling frame is the complete listof all of the population elements,usually people if one is conducting a survey,or a type of media content, like newspapers or TV episodes,if one is conducting a content analysis.There is more than one way to use probability sampling.
THERESA DE LOS SANTOS [continued]: I'll start with the simplest and most obvious form,simple random sampling.In simple random sampling, often usinga mechanical or electronic aid like a random number generatorto remove researcher bias and ensure that chance alonedictates selection, each person or item
THERESA DE LOS SANTOS [continued]: listed on the sampling frame is assigneda unique identification number.And elements are selected at randomto determine the individuals to be included in the sample.As a result, each element has an equal chance of being selected,and the probability of being selectedcan easily be computed.
THERESA DE LOS SANTOS [continued]: Statistical computing packages can be usedto determine random numbers.Excel, for example, has a built-in functionthat can be used to generate random numbers.Simple random sampling is the standardfor other forms of probability sampling.The second type of probability samplingis proportional stratified random sampling.
THERESA DE LOS SANTOS [continued]: Proportional stratified random samplingis a way to ensure that all segments of a populationare proportionately represented.To do this, researchers divide the populationinto mutually exclusive subgroups,like sex of the participants, political affiliation,relationship status.
THERESA DE LOS SANTOS [continued]: These subsets are called strata.Once the strata have been identified,a simple random sample is taken in each subgroup.In situations where there is unequal representationof one group or stratum in a population comparedto another group, to increase sample precision,proportional stratified random sampling
THERESA DE LOS SANTOS [continued]: is the preferred method.Let me give you an example.At private universities like Pepperdine Universitywhere I teach, the female to male student ratio is 60 to 40.Thus, if a survey researcher reallywanted to capture the pulse of the campuson a particular issue with a sample of 100 students,
THERESA DE LOS SANTOS [continued]: it should contain 60 female students and 40 male students.The next form of probability samplingis systematic sampling.Systematic sampling means sampling every nth personor item on a sampling frame, for example,taking every 10th or 100th person listed.The fixed interval that the researcher uses
THERESA DE LOS SANTOS [continued]: is called the sampling interval.To determine the sampling interval,the total population is divided by the number of elementsdesired in the sample.Let's say a researcher wanted to interview15 of the 75 journalism majors at Pepperdine University.They would divide 75 by 15.
THERESA DE LOS SANTOS [continued]: Then, after randomly selecting a starting point on the samplingframe, the researcher would cycle through the listand select every 5th student to be interviewed,as 75 divided by 15 is 5.Suppose you wanted to sample American collegestudents in clubs.You would need a whole lot of time and money
THERESA DE LOS SANTOS [continued]: to put together a sampling frame of every clubat every university across the United States.By the time you had this list, it would probably be outdated.So there's an easier way.With cluster sampling, you would first randomlyselect universities from a sampling frame,
THERESA DE LOS SANTOS [continued]: then select clubs at those universitiesand finally, students in the clubs.The advantage of this method is the relative easeof identifying people or items.The catch is that in every stage of sampling,the potential for error in the final sample increases.
THERESA DE LOS SANTOS [continued]: Probability sampling methods are lesslikely to be biased than non-probability samplingmethods.But using one of the four samplingmethods I just discussed is not always possible.Thus, researchers may use non-probability samplingthat reduces their ability to generalize their findings
THERESA DE LOS SANTOS [continued]: but may have theoretical relevance and/orthe advantage of convenience.There are several types of non-probability samplingmethods.Five of these include, first, convenience sampling.As the name implies, based on convenience,the researcher simply samples anyone
THERESA DE LOS SANTOS [continued]: who is willing and able to participate.More specific purposive sampling involves non-randomly selectingparticipants or items to fulfill or meet a specific purposeor characteristic that the researcher has in mind.Volunteer samples are made up of peoplewho self-select for a study, typicallyin exchange for a reward.
THERESA DE LOS SANTOS [continued]: The non-probability form of stratified samplingis a quota sample.In both stratified and quota sampling,participants or items are selectedto match population proportions.But this is important.In quota sampling, they are not randomly selected.To make this clear, let's return to the exampleof sampling 100 students from Pepperdine University where
THERESA DE LOS SANTOS [continued]: the known female to male student ratio is 60 to 40.Instead of stratifying by sex and then randomly selectingparticipants, a researcher using quota samplingwould also include 60 female students and 40 male studentsin their study.But the students selected would be their choiceor whoever is readily available.
THERESA DE LOS SANTOS [continued]: Finally, snowball or network samplingoccurs when researchers rely on members of a networkto introduce them to other members of a network.This sampling technique is often employedwhen the research demands people who are difficult to access.After trust is built with one individual,the researcher asks for referrals to other individuals.
THERESA DE LOS SANTOS [continued]: It should be obvious, but snowballor network sampling and volunteer samples onlyapply to human participants.The other three non-probability sampling techniquescan be used with non-human subjectsas well, most typically for media content and contentanalysis research.Overall, non-probability samples rely on the judgment
THERESA DE LOS SANTOS [continued]: of the researcher because sampling error cannot becomputed for them.Although a non-probability sample may closelymatch a population of interest, the researcherhas no way of knowing how much confidenceto place in the results.This takes us to key point number three.Sampling has to do with the credibility of research
THERESA DE LOS SANTOS [continued]: findings, particularly with regard to external validity.Although, we want to be careful to avoid strict categories.Whereas internal validity is typicallythe highest priorities for experimental researchers,survey researchers and quantitative content analysisare typically concerned with external validity
THERESA DE LOS SANTOS [continued]: with regard to sampling.As Campbell & Stanley's commonly accepted 1963 definitionstates, "external validity asks the questionof generalizability: to what population settings, treatmentvariables and measurement variables can this effectbe generalized?"Generalizing results requires that the sample in a study
THERESA DE LOS SANTOS [continued]: is very representative of the population to which the resultsare to be generalized.Typically, generalizing results requiressome kind of probability sampling that we just reviewed.Still, due to chance variation, if we randomlydraw multiple samples from a population,we would discover the samples are not identical.
THERESA DE LOS SANTOS [continued]: Sampling error involves the difference between samplesand population characteristics.The good news is that sampling error can be reducedby increasing sample size.So let me ask you a question.If you sample an entire population,do you have a sampling error?The answer is no.You perfectly matched your population of interest.
THERESA DE LOS SANTOS [continued]: And when probability sampling is used,it can be estimated using statistics.This is perhaps the critical advantageof probability sampling over non-probability sampling.It permits us to calculate how likely itis that a given sample differs from a populationon any question of interest and by how much.
THERESA DE LOS SANTOS [continued]: These calculations are called the margin of sampling errorand the confidence interval.Non-probability sampling does notpermit the computation of a margin of sampling errorin the same way that probability sampling does.As a result, there is much greater uncertaintyabout the accuracy of results from such samples.
THERESA DE LOS SANTOS [continued]: The bottom line is only probability sampleswith really good response rates allow researchers to makeestimates about sampling error.In conclusion, this tutorial covered the basics
THERESA DE LOS SANTOS [continued]: of sampling by focusing on the main differencesbetween probability and non-probabilitymethods of sampling.I hope you found it useful for better understandingthe consequences of the sampling strategy youutilize in your own research.Thank you.
Theresa de los Santos explains sampling techniques and how to use sampling research most effectively.
Looks like you do not have access to this content.
Theresa de los Santos explains sampling techniques and how to use sampling research most effectively.