Skip to main content
Search form
  • 00:01


  • 00:05

    PATRICK STURGIS: Hi, I'm Patrick Sturgis,and I'm Professor of Research Methods at the Universityof Southampton.[How would you define survey methods for a student?]Survey methods is the study of populationsby means of drawing samples from populations.And populations can be very broadly defined.

  • 00:27

    PATRICK STURGIS [continued]: We tend to think of populations in termsof the population of people in a country,but it could be the population of fishin the sea, the population of trees in a wood,pebbles on a beach, anything.But the key principle is that we can draw smaller subgroupsfrom the total population, which are then easier and morecost-effective to measure.

  • 00:51

    PATRICK STURGIS [continued]: And we can then make, what we call inferences,we can talk about the characteristicsof the whole population, just on the basis of this smallersample.So that's the key idea behind survey research.[What is the history of survey methods?]The history of survey research goes back a long while.

  • 01:14

    PATRICK STURGIS [continued]: If you think of the Bible, Jesus was on his way,in his mother's tummy, to be measured as part of a census.So the idea of measuring populations,systematically counting, usually for the purposes of taxation,goes back a long, long way from even before the time of Jesus.

  • 01:36

    PATRICK STURGIS [continued]: But the more modern, how we thinkof surveys in the modern era, probablycan be traced to the sort of mid-late 19th century, whena lot of people were becoming interested,for a variety of reasons, in issues of poverty, inequality,and so on.There was a lot of concern with the situationof the working classes, particularly in Britain,actually.

  • 01:60

    PATRICK STURGIS [continued]: And there were people at Charles Booth, Seebohm Rowntree,who were kind of dissatisfied with what you kind of thinkof as anecdotes about the poor, and so on.What Booth, in particular, set out to do,was to go out and actually make systematic measurementson whole tracks of London.

  • 02:21

    PATRICK STURGIS [continued]: And this where he did send out interviewers, knock on doors,ask people questions about-- and thenrecord this all systematically, and producedthese wonderful maps that you can stillsee in the British Library Political Economic Science.So these were the pioneers who startedsystematically measuring things, characteristics of populations.

  • 02:44

    PATRICK STURGIS [continued]: And then following that sort of early developmentthere was growing interest, again,now moving over to the US, particularly,in measuring audience perceptions of radio shows,of TV shows, cinema, and so on, because, of course,there's an advertising premium for getting audiences, gettingparticularly wealthy audiences.

  • 03:10

    PATRICK STURGIS [continued]: So this became more important to understand who was listening,and what they're paying attention to,so development further of, kind of, polling methods.And this was, in the early days, wasquite rudimentary in terms of how we would do surveys now.But probably one of the most important developmentswas by a chap, called, George Gallup,who still has a polling company named after him now.

  • 03:43

    PATRICK STURGIS [continued]: And he very famously got the election forecast rightin the 1936 presidential electionbetween Landon and Roosevelt, wherea very successful, at the time, magazine called, The LiteraryDigest, had 13 million readers, and it sent out cardsasking them to say who they were going to vote for.

  • 04:09

    PATRICK STURGIS [continued]: And they called the election heavily in favor of Landon.And Gallup, using a more systematic methodof drawing a small sample, but one which closely matchesthe population, called it for Roosevelt,and the rest is history.That was a very clear, prominent exampleabout how, it's not so much the size of the sample,it's how well it matches the population.

  • 04:36

    PATRICK STURGIS [continued]: And so those were some of the key early developments.[What kinds of questions about the world can be answered withsurveys?]Well certainly not every kind of questionthat you would like to ask about the world,but I think it's a-- when you are wanting to make inferencesto say things about the characteristics of populations,but you don't really have the time, or the resources,to go out and ask every single person.

  • 05:05

    PATRICK STURGIS [continued]: I mean, I think naively, that's how you would probablyapproach trying to understand whatthe average income in the population is,or how many people are going to votelabor or conservative in the next election,go out and ask everyone.Now, you can sometimes do that with some populations,if it's the population of studentsat Southampton University, maybe you can ask everyone.

  • 05:28

    PATRICK STURGIS [continued]: But surveys are useful when you needto probably get some information quite quickly,and you can't ask everyone.When you are asking about more concrete things,I think, surveys are better at measuring thingsthat are, let's say, concrete.

  • 05:54

    PATRICK STURGIS [continued]: How many qualifications have you got,rather than what do you think might happen if X, Y, and Z.People aren't so good at kind of putting themselves--probably not so good if we're interested in askingabout highly technical things, complex things.If we want to ask people about their attitudesto nanotechnology, and so on, thatcan be difficult in a survey because a lot of peoplehaven't thought about it.

  • 06:22

    PATRICK STURGIS [continued]: More towards the concrete and the tangible,the simpler kinds of questions that we mightwant to ask people.[When would you not use survey methods?]I'm a survey methodologist, but I wouldn't advocatesurveys for everything.I think precisely because you're dealingwith large numbers of people, many surveys,most surveys will have high hundreds, thousands, 20,000,even over hundred thousands, so youcan imagine that for each person that you interview,you can't spend too much time.

  • 06:57

    PATRICK STURGIS [continued]: And so there's a lot of pressure on space in a questionnaire,and that means actually you end upreally having to ask people quite simple questionsabout often complex issues.And also you have to ask the question in such a waythat it means the same thing to everyone.

  • 07:17

    PATRICK STURGIS [continued]: You can't ask some people one version of the question,and some people a different version.So again that tends towards askingsomething that's quite simple.And, by the same token, we can't really,or it's difficult, I should say, to let people chooselots of different answers.

  • 07:38

    PATRICK STURGIS [continued]: We want to be able to compare people and makefair comparisons.So we have to ask the same question of everyone,and we have to give everyone the same set of answers.And so often we'll end up saying,OK, we're going to ask you about this quite complex policyissue, and you can choose one of these four answers.

  • 08:00

    PATRICK STURGIS [continued]: So one of the common criticisms will be,well, people have much more diverse views,and we force people into choosingone of four categories, when it couldbe hundreds of different things that people say.And similarly, those four categorieswill be chosen by the researcher-- I'm offering youwhat I think in my range of possibilities,you may well want to choose something else,but it's not there.

  • 08:26

    PATRICK STURGIS [continued]: So that's one of the key criticism of the waythat we ask questions.There's also a criticism, which is that we are taking people--we're not observing people in their natural state,you know, we're putting them in an artificial situation,kind of like we are now.You're asking me questions, I'm answeringthem, that's not the way that we usually behave.

  • 08:53

    PATRICK STURGIS [continued]: And so there's something about the artificialityof the situation that we put people in,that we then worry about whether their answers reallyreflect the way that they genuinely feel,that we're somehow constructing their responses.Yeah, I guess those would be the two main things.[What advice would you give to students new to surveys?]One of the key things that distinguishesdifferent kinds of surveys is about how we draw the sample.

  • 09:22

    PATRICK STURGIS [continued]: And I think we can draw samples in a very, if you'd like,intuitive, and often low-quality ways,we might just stand on the street cornerand intercept people as they come by, and hopethey'll answer our questions.That wouldn't be a very good way of drawing a sample.On the other hand, you could do a more complicatedrandom selection of the full list of everyonein the population, possibly in a series of stages.

  • 09:52

    PATRICK STURGIS [continued]: So that's sort of a way of saying that sampling is key.Sampling is crucial in a way thatis not particularly intuitive, but is actuallyreally fascinating, and is, I think,something which is, if you like, somethingalmost magical about doing it.

  • 10:13

    PATRICK STURGIS [continued]: So I'm going to get enthusiastic about sampling.Because if you draw a sample in the right way,if you draw it according to random design, thenyou can make accurate inferences,you can say accurate things about the characteristicsof the entire population.

  • 10:34

    PATRICK STURGIS [continued]: So in Britain, if we draw a random sampleof 1,000 people, or even less, then wecan say accurate things about the whole populationof 60 million people.There are some other assumptions that wehave to get right in order for those estimates to be accurate.But in theory, if we draw our sample randomly,then there's a whole lot of math that wecan demonstrate-- enables us to make these inferences.

  • 11:06

    PATRICK STURGIS [continued]: If you don't draw your sample randomly,then you're not able to draw on that underlying mathematicaltheory.So that's a key distinction, and students reallyneed to understand that.It's not to say that you can't draw nonrandom samples,or that they're useless, but that'san important distinction.[How hard is it to draw a random sample?]If your target population is pebbles on the beach,it's very easy because pebbles on the beachdon't say, no, when you pick them up, well,not in my experience, anyway.

  • 11:35

    PATRICK STURGIS [continued]: But where it's much harder is in the real world,when your sample is of human beings, and they're busy,they're not in, they've got better things to do,and so you do not get everyone who should be in the sampleto give you an interview.So a nonresponse is a particular threatto being able to make accurate inferences.

  • 12:00

    PATRICK STURGIS [continued]: [In addition to a good sampling design,what else do you need to do good surveys?]Well you need to be able to write good questions,you need to understand-- I think, again, lots of peopleapproach questionnaire design because it's asking questions,people think, I can write a question, I can ask a question,I'll write down what I think sounds reasonable.

  • 12:22

    PATRICK STURGIS [continued]: But there's a very big literature setsof studies, which show that you canask questions in right ways, and wrong ways,and understanding those principlesabout questionnaires, on how to write questions,how to present the response options that peoplecan choose from.

  • 12:44

    PATRICK STURGIS [continued]: Essentially how we measure things,when in the first part of it is howwe draw a sample, the second thing-- which is actuallyoften more overlooked because the sampling side of things,even though when students read their textbooks,that may look scarier because it has lots of equationsand mathematics in it, that's actuallyand some times is more tractable,the mathematical problems we can deal with.

  • 13:10

    PATRICK STURGIS [continued]: What is actually harder is measuring things accurately.You can imagine if I ask you how many times you have takenphysical exercise in the last four weeks,there probably is some true value therethat we could get right or wrong,but how I ask you that question isgoing to have a lot of influence on how accurateyour answer is that you give.

  • 13:38

    PATRICK STURGIS [continued]: [What would you advice an inexperienced student to dobefore sending out a survey?]I actually am quite an advocate of notrushing into doing your own data collection as the first thingthat you do.Because I think you can learn a lot about how to do surveys,and how not to do surveys, by looking at surveysthat have already been done, and done to a high quality.

  • 14:03

    PATRICK STURGIS [continued]: So I think that's one of the thingsthat I would advise students to do,who are interested in doing survey research would be,rather than rushing out, as I often see students do, sortof write the questionnaire and send it out [INAUDIBLE],is to spend time working with an existing survey, lookingat the documentation and-- I sound a little boring--but it's certainly good preparation if youare going to go and do your own data collection.

  • 14:29

    PATRICK STURGIS [continued]: [What kind of surveys would you suggest people take look at?]Surveys are quite country-specific in termsof how we draw our samples.I was talking earlier about drawinga random sample of people from the whole population, nowhow do you do that?Well, in some countries you've got lists of every individual,you've got a population register.

  • 14:51

    PATRICK STURGIS [continued]: We don't have that in this country,we have a list of addresses that the post office puts lettersthrough mailboxes.So we need to draw samples differently,and that tends to be true in different countries.So what I say is really based on what you might do in the UK,and most of the big surveys will be done broadlyalong the same procedures in drawing the sample broadlyin the same way.

  • 15:22

    PATRICK STURGIS [continued]: So you could go to the UK Data Serviceand download a survey data sets, and all the documentation,and you'd probably get a good ideaof how to do a survey to high standard.I think, having said that, you probably-- there aresurveys that perhaps have more engaging content than others.

  • 15:48

    PATRICK STURGIS [continued]: You might be more interested in something like the BritishSocial Attitudes Survey, which asksa random sample of the British populationtheir views about kind of contemporary issues,and questions of kind of controversy, like immigration,and attitudes to gay and lesbian people,and how that's changing over time.

  • 16:10

    PATRICK STURGIS [continued]: So I think a lot of students, especially in social sciences,would find that particular survey quite interesting.Of course, many others.[MUSIC]

Video Info

Publisher: SAGE Publications Ltd.

Publication Year: 2017

Video Type:Interview

Methods: Survey research, Populations, Sampling

Keywords: accuracy in communication; applications and contexts; communication challenges; comparison; criticism and critics; George Gallup; history; inferences; poverty and inequality; practices, strategies, and tools; preparedness; Selection strategies; taxation ... Show More

Segment Info

Segment Num.: 1

Persons Discussed:

Events Discussed:



Professor Patrick Sturgis discusses the development of survey research since biblical times. He explains the appropriate use of the survey as a research methodology, and he highlights the challenges of good sampling and survey design.

Looks like you do not have access to this content.

Patrick Sturgis Discusses Survey Methods

Professor Patrick Sturgis discusses the development of survey research since biblical times. He explains the appropriate use of the survey as a research methodology, and he highlights the challenges of good sampling and survey design.

Copy and paste the following HTML into your website