Skip to main content
Search form
  • 00:00

    [MUSIC PLAYING]An Introduction to Sampling

  • 00:10

    DR. CHARLES LAURIE: How can you develop reliable knowledgeabout a group without studying all it's members?A population is the group of peoplethat your research is designed to generate knowledge about.If the population you're investigating is small,then it may be feasible to study allthe members of the population, known as a census.

  • 00:29

    DR. ERIC JENSEN: Most populations,however, are too large to gather datafrom all members of the population.You can instead gather data from a subset of the population,known as a sample.

  • 00:41

    DR. CHARLES LAURIE: Both qualitativeand quantitative research use samplesto make knowledge claims about larger populations.In quantitative research, the goalis to make accurate generalizationsabout a population's characteristics, such as,do more people hold view x or view y?Or, has the proportion of people holdingview x changed over time?

  • 01:03

    DR. CHARLES LAURIE [continued]: Qualitative research uses sampling to address questionssuch as what is the range of views being expressedby a group of people?Or, how do people come to hold certain views or feelingsabout a topic?

  • 01:15

    DR. ERIC JENSEN: However, this raises the question,if you only talk to a few people for your research,how can you say anything about the broader population?Drawing conclusions about a populationas a whole based upon the results from studying a sampleis called generalization.When thinking about how to make legitimate new knowledgeclaims based on your research, you

  • 01:35

    DR. ERIC JENSEN [continued]: should first consider some basic principles of generalizationin social research.Firstly, look for patterns.Sometimes you may encounter a situation, participantperspective or behavior that appearsto be irrational or hard to understand.Your first step is to approach the topic systematicallywith the aim of identifying any patterns that

  • 01:55

    DR. ERIC JENSEN [continued]: might explain what's going on.In qualitative research, what is oftenmost important in the process is the kindof factors that underpin the situation.The kind of processes that affect people'sviewpoints or their actions.In quantitative research, the connectionsbetween different variables within your samplewill often be a primary focus.

  • 02:17

    DR. CHARLES LAURIE: You could also look for deviant cases.When doing any type of research it is completely normalto encounter exceptions to the norm.How you work with deviant cases depends onwhether you are conducting quantitative or qualitativeresearch.

  • 02:31

    DR. ERIC JENSEN: And also beware of the ecological fallacy.There is a natural human tendencyto presume that trends at the group levelapply to the level of the individual.For example, we might make the mistakeof thinking that because men on averageearn higher salaries than women, that an individual man willalways earn more than an individual woman.

  • 02:50

    DR. CHARLES LAURIE: You should alsounderstand the types of new knowledge claims.You can generally make two types of knowledge claimsin social research.First, you could make new knowledge claimsbased on the evidence you've collected.That is your research claims.Second, you can make larger claimsabout the way your new evidence fitsinto the broader picture of existing research

  • 03:12

    DR. CHARLES LAURIE [continued]: and knowledge.That is the implications of your research.

  • 03:15

    DR. ERIC JENSEN: Your research claimsneed to be narrow and specific.The scope of your claims is limitedby the data you've collected.The possible implications you identifycan be much broader than that data,because they're based on the total pictureof available evidence, not just your data.Implications are generally more speculativeand should usually be tentatively phrased.

  • 03:37

    DR. ERIC JENSEN [continued]: For example, one possible implication of this finding is.

  • 03:42

    DR. CHARLES LAURIE: Make the most of critical review.Seeking critical feedback from more experienced researcherswill help you feel assured that your reasoning isas solid as possible.Such feedback can reveal hidden assumptions and gaps thatmight otherwise go unnoticed.

  • 03:56

    DR. ERIC JENSEN: Your sample will almost certainly fallshort of the ideal in some way.This is fine, as long as you understand and acknowledgehow this affects your analysis and limitsthe kinds of knowledge claims youcan make about your population, based on your research.

  • 04:11

    DR. CHARLES LAURIE: You also wantto narrow the gap between research claims and evidence.When starting a research project,you face a gap between what you would like to knowand what available data can tell you.Constraints such as time and resources and skillsmean you will rarely be able to access data that perfectlyanswers your research question.

  • 04:34

    DR. CHARLES LAURIE [continued]: Even if everything in your research projectgoes perfectly, there will alwaysbe a degree of uncertainty when decidingwhat your research data means.

  • 04:42

    DR. ERIC JENSEN: Humility and restraintwhen making research claims based on your datais always wise.Good academic social research does notclaim to have proved anything or discovered a universal lawof human behavior.Indeed, acknowledging the limits of your researchdoes not make it less valuable.The social world is messy, complicated

  • 05:04

    DR. ERIC JENSEN [continued]: and constantly changing.Therefore, making relatively restrained knowledge claimsbased on the best available evidenceis what social scientists do.You can narrow the gap between your available evidenceand your research claims in several ways.

  • 05:19

    DR. CHARLES LAURIE: Reduce the scope of your researchso that you aren't always making sweeping knowledge claims.

  • 05:24

    DR. ERIC JENSEN: Use representativesampling and acknowledge any limitations.Use existing knowledge to narrow the rangeof possible interpretations of your data.

  • 05:33

    DR. CHARLES LAURIE: Your main bridge across this gapwill be inference, the logical processof drawing general conclusions from available evidence.You rely on inference throughout the research process,including when you get to the stage of making knowledgeclaims at the end of your project.You must constantly work to recognize your biasesand be as rigorous as possible when

  • 05:54

    DR. CHARLES LAURIE [continued]: making sense of the evidence available to you.Establish the Basis for Your Knowledge ClaimsUsing a Methodology.

  • 06:02

    DR. ERIC JENSEN: You will also wantto think about how to establish the basis for your knowledgeclaims.When publishing or presenting your search results,you must always include a section describingyour research methods.This section allows your readers to independently assessthe validity of the claims you makeand to identify any flaws in your research designthat might undermine your conclusions.

  • 06:24

    DR. ERIC JENSEN [continued]: In this section, you should explain these kinds of things.Why you opted for a particular sampling strategy?Who you collected data from?How you gathered your data?How you identified and measured the variables that you'relooking at in your study?You also want to address how you analyzed your data

  • 06:45

    DR. ERIC JENSEN [continued]: and how you came to your conclusions.

  • 06:48

    DR. CHARLES LAURIE: Now you can startthinking about making knowledge claims beyond your sample.Social research is generally motivated by a desireto know something about a group of people,your population of interest.Sampling is based on the principlethat you can use data collected from a smallernumber of individuals to represent the larger groupfrom which they were drawn.

  • 07:09

    DR. CHARLES LAURIE [continued]: A representative sample then is one which characteristicsclosely match those of the population from which itis drawn.Accurate generalization from a quantitative researchperspective requires a precisely defined populationfrom which a representative sample can be drawn.For a sample to be fully representative,it needs to be theoretically possible to select

  • 07:31

    DR. CHARLES LAURIE [continued]: any member of the population for inclusion in the research.

  • 07:35

    DR. ERIC JENSEN: Many qualitative researchers alsohave an interest in making generalized knowledgeclaims that extend beyond their particular participants.However, qualitative research tendsnot to focus on a statistically representative sample,but rather on having a diverse sample thatshows a range of perspectives from within the population.A qualitative researcher may observe processes and patterns

  • 07:58

    DR. ERIC JENSEN [continued]: within a small group of individuals,and then build a working theoretical modelbased on those observations.This is then tested against additional observationswhich add to and refine the model basedon this new information.Thus qualitative research can be usedto build theoretical models that explain behavior in general

  • 08:18

    DR. ERIC JENSEN [continued]: and go well beyond the particular participantsin the study.

  • 08:22

    DR. CHARLES LAURIE: Some researchersargue that generalization beyond the contextin which the research was conducted is not legitimate.These researchers question whether we can directlymeasure any phenomenon using eitherqualitative or quantitative methods, giventhat our research participants are thinking,self-aware beings, that inevitably adapttheir responses based on how and in what context

  • 08:43

    DR. CHARLES LAURIE [continued]: they are approached.Researchers who follow this way of thinking emphasize thickdescription, that is, lots of descriptive detail,both of the interesting and seemingly mundaneaspects of the context in which the research was conducted.

  • 08:59

    DR. ERIC JENSEN: The first step of the sampling processis identifying and defining the populationyou're interested in.The next step is to decide what the most appropriate samplingmethod would be based on the details of your targetpopulation.You can choose from a range of different techniquesfor acquiring your sample.These are grouped together as probability sampling techniques

  • 09:22

    DR. ERIC JENSEN [continued]: and nonprobability sampling techniques.Every technique has its own benefits and limitations.Probability Sampling.

  • 09:33

    DR. CHARLES LAURIE: Probability sampling techniquesare designed to ensure that in principle, anyof the individuals in your target populationmay be selected to be part of your sample.Generally this includes some formof random selection, which helps to ensure that the sample isstatistically representative.This kind of sampling is considered the strongestfor quantitative research and helps

  • 09:55

    DR. CHARLES LAURIE [continued]: to avoid the major sources of sampling bias.Qualitative researchers seldom use probability sampling.

  • 10:03

    DR. ERIC JENSEN: Simple random samplingis the simplest form of probability sampling.Members of your population are selected completely at randomfor inclusion in the sample, ensuringthat everyone has the same chance of beingselected for participation.This requires little or no informationabout your population's characteristics

  • 10:24

    DR. ERIC JENSEN [continued]: and is easily the most robust form of sampling in most cases.It can be performed with computer programsand can be very affordable for samplingsome populations, dependent on how geographically dispersedyour population might be.The main point is that every member of your populationmust have an equal chance of selection,

  • 10:44

    DR. ERIC JENSEN [continued]: which means you must be able to access everyonein your population.Therefore, in practice, you will at bestbe working with a nearly random sample, whichwill have limitations that you must acknowledgewhen reporting your results.

  • 10:58

    DR. CHARLES LAURIE: While it avoids sampling bias,simple random sampling also leavesthe demographic distribution of your samplecompletely up to chance.For statistical reasons you may need a minimum number of peoplein a particular category.One technique that can address this issueis called stratified random sampling,and it involves dividing the population

  • 11:19

    DR. CHARLES LAURIE [continued]: you are researching into exclusive subgroups, or strata.This can be along the lines of age, gender, country of origin,ethnicity or any other important characteristic that you need.You can then gather part of your samplefrom each of these subgroups, usuallyby selecting an equal number of people from each category.

  • 11:39

    DR. CHARLES LAURIE [continued]: This is useful if you're looking to comparethe attitudes of particular subgroupsby ensuring those groups are representedin sufficient numbers within your sample.Stratified random sampling is alsoused if you're trying to compare subgroups with drasticallydifferent sizes.As you might have guessed, stratified random samplingrequires significantly more information

  • 11:60

    DR. CHARLES LAURIE [continued]: about your population, and is much morecomplicated to perform than simple random sampling.You first have to know the number of individualsin each category in your population.

  • 12:11

    DR. ERIC JENSEN: Cluster sampling is usefulwhen your population is dispersedacross a large geographical area and would be expensiveand time consuming to survey them in person.In order to reduce the costs associatedwith surveying far flung participants,you can divide the population into geographical groupsor clusters.Often this is based on preexisting boundaries

  • 12:31

    DR. ERIC JENSEN [continued]: like counties, cities or city blocks.Then, rather than selecting individuals,you randomly select clusters.In basic cluster sampling your sample consists of everyonein the clusters you selected.If the clusters are proportionate in size,then this method is similar to simple random sampling,in that all members of the population

  • 12:53

    DR. ERIC JENSEN [continued]: have a roughly equal probability of being selected.This technique can also be used to gathera representative sample when you havea large group of people in one space, like in a festival,for example.

  • 13:04

    DR. CHARLES LAURIE: As you may have gathered by this point,these sampling techniques are not mutually exclusive.It is common, especially in large scale surveys,to use a combination of techniques.This will be referred to as multi-stage sampling,and one example of which is stratified cluster sampling.

  • 13:21

    DR. ERIC JENSEN: At this point, all the probabilitysampling techniques discussed so farhave used some form of random selection.Sometimes though, you'll find that it's just notpossible to use a truly random sampling method.At this point you can turn to systematic sampling, whichuses selection rules to mimic the objectivity

  • 13:41

    DR. ERIC JENSEN [continued]: of random selection.Often the selection rule is as simple as pickingevery n-th person to meet some criteria.While this approach is not as good as strict random sampling,it can come relatively close to producing a random sample.And it also avoids the key risk, whichis researcher selection bias.

  • 14:02

    DR. ERIC JENSEN [continued]: Nonprobability Sampling.

  • 14:07

    DR. CHARLES LAURIE: Now let's talkabout nonprobability sampling.Nonprobability sampling techniquesdo not have equal probability of selectionfor all members of your target population.You can obtain a representative sample using nonprobabilitysampling, but there's no guarantee nor any wayto measure the statistical probability that the sample isrepresentative.

  • 14:29

    DR. ERIC JENSEN: Nonprobability sampling techniquesare generally much cheaper and easierthan probability sampling, but theyshould be used with caution.Because you don't know if your sample is representativeof the population, you can't use the most powerfulinferential statistical tests to generalize from your sampleto the population as a whole.Your knowledge claims about your sample data

  • 14:49

    DR. ERIC JENSEN [continued]: will therefore need to be more tentative, especially if you'remaking quantitative claims.

  • 14:54

    DR. CHARLES LAURIE: Convenience samplingis a technique that selects the members of a population whoare the easiest for the researcher to access.This sampling technique is very cheapand requires minimal effort, but has obvious drawbacks.It is usually unlikely that the people sampledthrough this method are representativeof a wider population.That said, because convenience sampling is so inexpensive,

  • 15:17

    DR. CHARLES LAURIE [continued]: it can be an effective way to test survey questions,or to conduct other preliminary pilot testing activities.

  • 15:24

    DR. ERIC JENSEN: Next its quota sampling.So quota sampling is the selection of a sampleaccording to a predetermined quota on a non-random basis.Quota sampling is often compared with stratified randomsampling, as both of these types of sampling methodscontrol the sample based on variables of interest.For example, gender or ethnicity.

  • 15:46

    DR. ERIC JENSEN [continued]: Quota samples are much less expensiveand they differ in their ability to accurately representthe population of interest.Whereas stratified random sampling stilldelivers a probability sample, quota sampling does not.

  • 15:59

    DR. CHARLES LAURIE: There are two ways to use quotas.Proportionately and non-proportionately.Proportional quota sampling involves selecting participantsfor your sample in proportion to some characteristicof the population as a whole.Once you have reached this quota,you would stop sampling this category of participant.This allows you to generate a sample thatis on the surface reflective of the distribution

  • 16:22

    DR. CHARLES LAURIE [continued]: of some characteristic across the population.However there's no way to ensure that the sample isrepresentative of the larger population.

  • 16:31

    DR. ERIC JENSEN: Snowball samplingis a technique where participants in your researchare used to find additional participants.It relies on discovering a strategically important contactinitially.This contact is used to introduce youto other contacts that fit the characteristicsin your population.The technique is helpful for identifying study participantswith very specific characteristics,

  • 16:53

    DR. ERIC JENSEN [continued]: rare experiences or hidden population groups whomay be difficult to identify with other recruitmentmethods, such as those involved in illicit or taboo phenomena.This almost guarantees though that your samplewon't be representative.And it's really only a justifiable methodwhen there's no other good way to access participantswithin your population.

  • 17:15

    DR. CHARLES LAURIE: Purposive or theoretical samplinguses the researcher's judgment to select participantswho are likely to offer particularly valuable insights.To do this, you will need to have a special knowledgeor expertise about a group to select subjectswho represent the population.The selection process can be guided earlier data collectionwithin the project, prior research or theory, as well

  • 17:37

    DR. CHARLES LAURIE [continued]: as the researcher's instincts.This involves selecting for the extremes in a populationor selecting individuals to representcategories of interest.

  • 17:46

    DR. ERIC JENSEN: There's certainlyno shortage of effective sampling strategies.To decide on appropriate sampling strategyfor your project, you need to consider the population you'reinterested in, the scope of claimsyou want to be able to make about the populationand the time and resources that you have available.

  • 18:03

    DR. CHARLES LAURIE: It is important to understandthe principles of quantitative sampling.Quantitative research involves applying inferential statisticsto your sample in order to make generalizationsabout your population as a whole.To support quantitative knowledge claims,your sampling techniques should be robust.Methods should be consistent, well justified

  • 18:24

    DR. CHARLES LAURIE [continued]: and clearly spelled out in advance.They should be objective.It should not be dictated by the personal views or assumptionsof the researcher.And they should be representative.It should match as closely as possiblethe characteristics of your population.

  • 18:38

    DR. ERIC JENSEN: Sample size effects your optionslater on when you go to conduct your statistical analysis.A larger sample size will increase the sensitivityof your analysis, making it easierto detect patterns that exist within the population.The smaller your sample, the less likelyyour sample will reflect your population.

  • 18:57

    DR. CHARLES LAURIE: You will alsoneed to track your response rate, whichis an important metric in quantitative survey research.The response rate is the percentageof respondents actually responding to your surveyor research questions.In order to determine whether thereare obvious biases entering your sample at this stage,you should keep a refusals log, where you write down

  • 19:18

    DR. CHARLES LAURIE [continued]: basic information.This includes the age, gender, ethnicity of the respondentand the date and time of their refusal.And the reasons for their refusal,both explicit and possible other reasonsthat you've decided upon.Tracking this refusal informationenables you to answer questions about sampling biasin a robust manner.

  • 19:39

    DR. CHARLES LAURIE [continued]: With online surveys, there may be fewer optionsfor tracking refusals.

  • 19:43

    DR. ERIC JENSEN: Qualitative samplesare different from quantitative symbolsbecause they tend to be smaller, as they involvemore intensive data analysis, theycan be selected proposively, and tendto be less focused on statistical representativenessas a criterion for quality.They tend to focus more instead on saturation,that is, reaching the point whereyou encounter similar phenomena, again and again.

  • 20:06

    DR. ERIC JENSEN [continued]: Qualitative research involves exploring patternsand processes across a relatively narrow rangeof cases, rather than focusing on the distributionof certain characteristics across an entire population.

  • 20:18

    DR. CHARLES LAURIE: Qualitative researchers sub-consciouslydraw upon their own experiences as a resourcein their inquiries.Good qualitative research can build upvalid explanations that make no mentionof the percentage of the population for whomthese explanations apply.Nevertheless, if you are doing a qualitative research project,you do need to consider how representative your sample is.

  • 20:41

    DR. CHARLES LAURIE [continued]: You can choose to focus on particular categoriesof individuals, but you need to justify the selectionand adjust your analysis and knowledge claims accordingly.

  • 20:49

    DR. ERIC JENSEN: Unfortunately, thereare no simple rules for sample size in qualitative inquiry.Sample size depends on what you wantto know, the purpose of inquiry, what will be usefuland what will have credibility, and whatcan be done within the available time and resourcesthat you have.Before you even start your data collection,qualitative research can require a significant time commitment.

  • 21:12

    DR. ERIC JENSEN [continued]: Even a relatively simple research designlike interviewing can take more than 10 hoursto carry out due to the amount of time necessary to contactparticipants and conduct the data collection.This includes preparing the audio recording equipment,organizing and conducting the interview,transcribing the data once you have it,analyzing the interview and so on.

  • 21:33

    DR. ERIC JENSEN [continued]: The time demands can quickly escalateif you don't keep a tight limit on your sample size.

  • 21:38

    DR. CHARLES LAURIE: Use emergence, saturationand representativeness to guide youthrough the sampling process.It's important to stay flexible while conductingqualitative research.You must allow for both emergence and saturation.Emergence involves following interesting, oftenunanticipated, lines of inquiry thatcome up during the qualitative research process.

  • 21:59

    DR. ERIC JENSEN: A technique for decidingyou can stop collecting data is called saturation.Qualitative research is an ongoing processof developing an explanation that will effectivelyaddress your research question.You've reached saturation once all previous cases canbe explained and new cases are no longerproviding additional clarification or changes

  • 22:19

    DR. ERIC JENSEN [continued]: to your basic explanation.Develop your Qualitative Knowledge Claimsby Doing a Literature Review.

  • 22:25

    DR. CHARLES LAURIE: How do you develop qualitative knowledgeclaims?Qualitative research is often in an excellent positionto develop context rich, detailed and descriptiveaccounts, aided by a strong understandingof the specific context affecting your participants.Yet identifying your project's broader implicationsis also important, both in establishing its value

  • 22:46

    DR. CHARLES LAURIE [continued]: and ensuring its contribution to knowledgebeyond your particular case that you're researching.

  • 22:52

    DR. ERIC JENSEN: One way to do thisis to look for processes that mayfeature in a wider range of casesthan the one you investigated.Generalizing in a qualitative research contextcan range from explaining the how and whyof your particular case to explaining general processesand developing theoretical models.Some have suggested that qualitative research should not

  • 23:14

    DR. ERIC JENSEN [continued]: attempt to offer causal explanations or predictions.As such, qualitative researchers seekto use observations of the presentto describe past occurrences that haveled to the present situation.The processes you identify may over timeapply to other context and other studies,leading to development in theory and understanding.

  • 23:34

    DR. CHARLES LAURIE: Alternatively, youcould use qualitative research findingsto build your own theoretical explanation, elaborateor to challenge an existing theory or explanation.You could show what a theoretical concept lookslike in practice and evaluate its strengths and limitationsin your research context.Or you could also use your qualitative study

  • 23:55

    DR. CHARLES LAURIE [continued]: to demonstrate gaps and limitations in an existingtheory.Develop Your Quantitative Knowledge Claimsby Using Inferential Statistical Tests.

  • 24:04

    DR. ERIC JENSEN: So how then do you developquantitative knowledge claims?In quantitative research, knowledge claimsare often developed using inferential statistics.Here the goal is identify statistical patternsand relationships between variablesat the population level using data you'vecollected in your sample.

  • 24:25

    DR. CHARLES LAURIE: Be a savvy user and consumerof statistical inference.Quantitative data drawn from a representative sampleallows you to make claims about a population's characteristicswith a known level of confidence.You may identify differences thatare statistically significant, one that is very likely, again,95% confidence or better, to apply to the larger population.

  • 24:47

    DR. CHARLES LAURIE [continued]: A statistically significant resultis unlikely to result from chance or from random variationthat occurs whenever you draw a sample from a large population.It is a real difference that shouldbe observable in the larger populationif you were able to collect data from allthe members of the population.Statistical significance and confidence intervals

  • 25:09

    DR. CHARLES LAURIE [continued]: both come from inferential statistics.

  • 25:12

    DR. ERIC JENSEN: Whenever you draw a sample to representa larger population, there'll be differences between the sampleand the population.This gap is known as sampling error.When making inferences based on quantitative datathere are two main types of inferential errorthat you can make.The first is a type one error, alternatively knownas false positives.

  • 25:33

    DR. ERIC JENSEN [continued]: This error involves the rejectionof the null hypothesis that is actually true.For example, if you were to concludethat a new medical treatment helps to cure cancerwhen it does not, the result will be tragicand this would be a type one error.

  • 25:48

    DR. CHARLES LAURIE: The second istype two error, alternatively known as false negatives.This error is the failure to reject a null hypothesis thatis actually false.For instance, it will be a very serious errorif you were to dismiss a potential cure for canceras ineffective, when it actually cures the disease.

  • 26:07

    DR. ERIC JENSEN: A great deal of the focus in social researchis on type one error and it's generallyconsidered the most important type of error to control.Therefore, we would want to reducethe risk of a false positive by convention to less than a 5%probability before accepting that thisis a real statistically significant result.

  • 26:28

    DR. ERIC JENSEN [continued]: This is a precautionary position.Things stay as they are until there'sa sufficient level of evidence to showthat a change is needed.

  • 26:36

    DR. CHARLES LAURIE: You will alsoneed to explain statistical inference to your readers usingconfidence intervals and p values.Confidence intervals are used to indicate the levels of type oneerror that would apply if you were generalizing a findingfrom a sample to a population.Generalization within explicitly identified confidence intervalsis informed by probability theory.

  • 26:58

    DR. CHARLES LAURIE [continued]: It provides a robust basis for making inferencesabout populations based on samplesand for analyzing relationships between variables.

  • 27:07

    DR. ERIC JENSEN: Having less than a 1in 20 chance of making an inaccurate generalizationhas been set as a tolerable level of type one errorwithin quantitative social scienceresearch by general convention.This is expressed in statistical notationas p is less than 0.05, with p standing for the risk of making

  • 27:27

    DR. ERIC JENSEN [continued]: a type one error if you reject the null hypothesis.0.05 stands for 5% or 1 in 20 probability of occurrence.If a result is found to be statisticallysignificant in this way, that doesn't necessarily meanit's an important finding.You would still need to determine the effect sizeto establish how much is explained by the statistically

  • 27:50

    DR. ERIC JENSEN [continued]: significant difference.If only a small amount is explained,then the fighting could be statistically significant.That is, there's a better than 95%chance that it's a real effect, but relativelyunimportant for answering your research question.In other words, not substantially significant.

  • 28:08

    DR. CHARLES LAURIE: In general, social researchfocuses on the overall typical patterns,known as the aggregate level, rather thanthe unique or unusual aspects of individuals views and lives.In quantitative research, particulars from people's livesare conceptualized as variables thatcan be investigated statisticallyacross your sample.

  • 28:28

    DR. CHARLES LAURIE [continued]: This can reveal patterns that apply to the population level.But it's important not to overstatewhat can be concluded from such aggregate research.Indeed, you have to be careful about making claimsat the individual level based on statistics thatapply to the whole population.Conclusion.

  • 28:49

    DR. ERIC JENSEN: Overall, we've lookedat how to develop reliable knowledge about the groupyou're studying without gathering data from allits members in this video.You'll need to have a look for patterns,you'll need to identify deviant casesand avoid the fallacy of presumingthat trends at the group level apply at the individual level.As you decide what kind of sampling strategy

  • 29:09

    DR. ERIC JENSEN [continued]: is right for your study, consider the following issues.Firstly, the scope of claims you wantto make at the end of your study.Secondly, the strengths and limitations of yourproposed methods for getting at the relevant evidence.And third, whether there any discrepanciesbetween one and two.

  • 29:26

    DR. CHARLES LAURIE: Bias in samplingoccurs when the participants you've selected for your sampleare not representative of the population as a whole.This is an issue no matter what sampling technique you use.Gathering a truly representative sample of a populationis very difficult, and in some cases may be impossible.So there will always be the risk of some sampling error.

  • 29:47

    DR. CHARLES LAURIE [continued]: You should acknowledge this risk and attemptto minimize sources of bias to the greatest extent possible.Be open and self critical about such limitationsand keep in mind how they will be affecting your analysisand the knowledge claims you're making.While probability samples are the gold standard,they will sometimes be unachievablewithin the practical constraints and limits

  • 30:09

    DR. CHARLES LAURIE [continued]: of a particular project.In such cases, the most important principleis to be systematic in order to avoid accidentally introducinga sampling bias.

  • 30:18

    DR. ERIC JENSEN: In research projects wheretime and resources are in scarce supply,cutting back on the sample size is often the best optionfor immediately reducing costs.You may need to use existing data sources rather than doingyour own original data collection,or take a different, less costly sampling approach in orderto save time and resources.

  • 30:39

    DR. ERIC JENSEN [continued]: In some cases, such changes to your research planwould mean reducing the scope of your research questionand certainly the scope of your claims.

  • 30:47

    DR. CHARLES LAURIE: Finally, keep in mindthe fundamental differences between qualitative andquantitative sampling.For qualitative research, consider the time you'll needversus how much time you have available for the project.

  • 30:59

    DR. ERIC JENSEN: Also be preparedto adjust your plans based upon the emergingfindings from your qualitative researchand avoid sampling past the point of saturation.For quantitative research, having a large sample sizeand robust sampling methods will give youthe basis for making better generalizations.

  • 31:16

    DR. CHARLES LAURIE: Understandingthe common pitfalls and errors associatedwith statistical inference can helpensure that you're making accurate knowledge claims.

  • 31:24

    DR. ERIC JENSEN: Finally, keep in mindthat all knowledge is provisionaland open to change when new and better evidence is presented.Keep your research on a solid footing by only making claimsthat you can support with your data and referencesto existing research.

Video Info

Publisher: SAGE Publications Ltd

Publication Year: 2017

Video Type:Tutorial

Methods: Sampling, Representative samples

Keywords: ecological fallacy; emergence; expertise; false positive; knowledge gap; practices, strategies, and tools; quotas; theory development ... Show More

Segment Info

Segment Num.: 1

Persons Discussed:

Events Discussed:



Dr. Eric Jensen, Professor of Sociology at the University of Warwick, and Dr. Charles Laurie, Director of Research at Verisk Maplecroft, present a thorough introduction to sampling in both qualitative and quantitative research.. They discuss various methods, best practices, and pitfalls to avoid to produce consistent and high quality research.

Looks like you do not have access to this content.

An Introduction to Sampling

Dr. Eric Jensen, Professor of Sociology at the University of Warwick, and Dr. Charles Laurie, Director of Research at Verisk Maplecroft, present a thorough introduction to sampling in both qualitative and quantitative research.. They discuss various methods, best practices, and pitfalls to avoid to produce consistent and high quality research.

Copy and paste the following HTML into your website