Skip to main content
Search form
  • 00:00


  • 00:05

    DR. BENEDICT WHEELER: I'm Benedict Wheeler.I'm a senior research fellow at the European Centerfor Environment and Human Health,which is part of the University of Exeter Medical School. [Dr.Benedict Wheeler, Senior ResearchFellow at the European Center for Envirionment and HumanHealth, University of Exeter Medical School].And I do quite a broad range of research.But my main methods really are health geographyand environmental epidemiology, studyingthe impacts of the environment on human health and well being.

  • 00:27

    DR. BENEDICT WHEELER [continued]: [What research have you done examining the relationshipbetween health and environment?]So the project that we've been working on at the momentis called Beyond Greenspace.And this is about the relationshipbetween the natural environment and health and well being.There's quite a good history of research in this area,particularly in the last five or 10 years, it's really bloomed.

  • 00:49

    DR. BENEDICT WHEELER [continued]: But our take on this field of researchis, let's compare green space and built environments.How do they impact on people's healthin terms of where you live and where you spend your time?But what happens if we unpack that a bit more?And let's take urban green space.Say, does it make a difference if it's got water in it?

  • 01:09

    DR. BENEDICT WHEELER [continued]: If it's got trees in it?If it's got big open spaces wherepeople can run around in it?There are different types of spacesthat might afford different kinds of activities,different sorts of experiences.So that's what this piece of research is all about.[What methods have you used to do this research?]Well, this project is similar to lots of my other researchthat I've worked on in the past and other peoplein the team, which is based around secondary data analysis.

  • 01:35

    DR. BENEDICT WHEELER [continued]: So it's taking large data sets thathave already been collected by other peoplefor other purposes.And we take that data.And what we tend to do that for this kind of researchis link it to environmental data.So we can take data like the census-- whichcounts everybody in the country, or almosteverybody in the country-- and it gives usdata about health, and their age, and their sex,and where they live and so on.

  • 01:59

    DR. BENEDICT WHEELER [continued]: And we can link that to environmental databased on where people live.And so we use a lot of geographic informationsystems-- so GIS which is sort of computerized databasesand mapping data-- to link togetherdata about the environment, data about people,and joining it together to understand how those thingsrelate to each other.[What are the advantages and disadvantages in usingsecondary data in your research?]And that's one of the main thingsthat we have to grapple with is that it's secondary data.

  • 02:27

    DR. BENEDICT WHEELER [continued]: We didn't collect it.We didn't design the questionnaires.We have to work with what's there.And so for a lot of our research, it's OK.Something like the census asks a really good standard questionabout health.How is your health in general?It's very simple question, but it's actually well-established.It's very powerful.So it really works for us.

  • 02:48

    DR. BENEDICT WHEELER [continued]: And other surveys do that too.But obviously like you say, they can be collectedfor different purposes.So we're sort of repurposing it for our own needs.So we have to play around with it a bitto make it do exactly what we want to do.The advantages would be the scale and the cost.

  • 03:09

    DR. BENEDICT WHEELER [continued]: So the census is absolutely massive, obviously.It's a huge dataset that we could never collect on our ownas researchers.But similarly large longitudinal datasets--like the British Household Panel Surveycollected data on around 5,000 householdsevery year for 18 years from 1991--are hugely powerful datasets that then wecan tap into and use all that statistical powerin the large numbers without having to invest inthat data ourselves.

  • 03:40

    DR. BENEDICT WHEELER [continued]: Mostly what we have to do is invest in people's timeto sort of fund research time to actually use that data.[Did you supplement your research by using other sourcesof data?]Well I guess one of the advantagesnow is that a lot of this data is available online.We have great services-- like the UK Data Service--which sort of archives and houses a lot of these kindsof datasets and then makes it available for our kindof research.

  • 04:08

    DR. BENEDICT WHEELER [continued]: So that's really, really useful for us.And we can go online, we can go on to their website.Just the other day I was looking for some data on volunteering.So I just typed in volunteering to their search termsand came up with an interesting looking dataset thatmight be what I wanted to do.So once you've identified the data,then you might have to apply for special permissionto be allowed to access it.

  • 04:31

    DR. BENEDICT WHEELER [continued]: That's certainly the case for our usewhen we want to know geography, whenwe want to know something about where people live.Quite often that's quite protected data,so we have to apply for a special licenseto be allowed access to that to prove that we're notgoing to do anything that might breach data protectionconfidentiality.So we download that data.

  • 04:51

    DR. BENEDICT WHEELER [continued]: And you're right.We'll go back to things like the original questionnaires justmake sure how the questions were asked, if they'reasking the things that we think they are, clean up the data,look at missing data.So quite often, there might be a bit of missing datawhere people haven't answered the question.And then once we've done all that cleaning up,we can get on with actually analyzing it.

  • 05:12

    DR. BENEDICT WHEELER [continued]: [Do you analyze it as a whole or do you take samples?]With those kinds of datasets, we will mostlyanalyze the whole lot.It depends on exactly what the question is.So some of our analyses-- say using the British HouseholdPanel Survey-- we've analyzed the entire sample followingpeople over time.For example, looking at whether peoplereport better mental health when they live in greener asopposed to less green urban areas.

  • 05:39

    DR. BENEDICT WHEELER [continued]: For other bits of that analysis, we justlooked at people who had moved house.Because that's the key way by which people's environmentchanges is they moved to a different location.And in that case, we'd really boil down the samples.And actually we ended up with about 600 people whohad moved to a greener urban area, and about 500who'd moved to a less green urban area.

  • 05:59

    DR. BENEDICT WHEELER [continued]: So sometimes we'll cut the data like thatto find the people that we actually want to study.[How deep do you think quantitative analysis can goin a study like that, and do you use any other methods?]There's a lot of depth you can get into now, especiallywith these complex longitudinal datasets,like the BHPS understanding society.And some of the cohort datasets are incredibly rich,have a huge amount of data that wecan, for our analyses around green space,adjust for all sorts of characteristicsof individual people and the places where they live.

  • 06:34

    DR. BENEDICT WHEELER [continued]: But I think that for some of these questions, particularlyaround say how people relate to their urban green spaces,taking a deeper qualitative approachcan be really powerful.And we have some research going on in our center thatuses those approaches and starts to tease those relationshipsapart a bit more.And there are opportunities to do that.

  • 06:55

    DR. BENEDICT WHEELER [continued]: I know there are some sort of qualitative secondary datasetsthat are available.They can be quite challenging to use,I think, because probably even more sothan with the quantitative data, qualitative data collectionhas set off with a pretty particular in-depth questionthat it's trying to answer.And so finding secondary qualitative datato answer those questions can be quite hard, I think.

  • 07:19

    DR. BENEDICT WHEELER [continued]: [What are the key things a student should know if they areusing secondary data in their research for the first time?]I think probably the main thing that I would recommendis to be quite specific about the datasets to start off with.So if you thought, maybe there issomething in the census that willhelp me answer this particular research question I've got.The best thing to do is a, to look at the source material.

  • 07:42

    DR. BENEDICT WHEELER [continued]: So you go to the ONS website.Look at all the material that's available around the 2011census.There's tons of stuff on there with questionnaires and so on.But also look for publication.So what have other people publishedusing the 2011 census, using weather science,or the usual sort of reference databases?What else is out there?And you can start to get a feel for what people have actuallydone with that data.

  • 08:05

    DR. BENEDICT WHEELER [continued]: So that's the kind of dataset-specific approach.I suppose then there would be very specificquantitative methods books.And I can't think of any offhand,but if you're interested in, say, quantitativehuman geography-- which is a lot of what do,that might be the area that you go to that bit of the libraryand just have a browse through the books.

  • 08:29

    DR. BENEDICT WHEELER [continued]: That can be quite a useful way to just geta feel for what kind of material is available.[What are common mistakes made by new researchers when usingsecondary data for the first time?]I suppose one common mistake-- and I'm as guilty of thisstill as anybody-- is that it can be so exciting to geta hold of this data.There's tons of stuff in there.

  • 08:50

    DR. BENEDICT WHEELER [continued]: And maybe you've got a question that nobody else has everthought of that could be addressed using that data.And it's quite tempting just to download the data from the dataservice, and just get on with it, just kind of-- youcan start putting in commands into your stats packageand sort of play with the data.And that is a really good thing to do.Because you have got to get a feel for the data, what'sin it, what's not in it.

  • 09:13

    DR. BENEDICT WHEELER [continued]: But at some point, you have to step back and go,hang on a minute.Is this actually doing what I think it's doing?Go back to the documentation, go back to the questions.Look at that question that's asking the question that youwere interested in.And think is that actually what I'm intending to do.Is that really capturing it?What are the limitations about that particular question.

  • 09:34

    DR. BENEDICT WHEELER [continued]: So I think that that's probably number one mistaketo make with secondary data is make sure you know it's doingexactly what you think it is.[How did you first become interested in research?]I guess that came about through my degree.So I did a BSE in environmental sciences.It was a very broad environmental science programwith lots of different aspects of the environmental sciences.

  • 09:58

    DR. BENEDICT WHEELER [continued]: One of the modules was in environmental epidemiology.And another one of the modules was in GIS.And these things started to really gel and make senseto me.And thinking about, OK well we canunderstand how the environment impacts on human health.And maybe we can use GIS as a methodologyto link those two things together.And so I think that's where it all came from.And I did my final year dissertationon air pollution and childhood asthma,and used GIS to bring together sort of quant datasetsaround those sorts of questions.

  • 10:29

    DR. BENEDICT WHEELER [continued]: And it's pretty much continued ever since then.[What advice would you give your younger self on socialresearch?]I really enjoyed doing my PhD.And it was a fantastic opportunity--particularly with these kind of approaches--to really get to grips with a lot of different datasets.I just loved the idea that I couldsit on the internet, or phone somebodyup and get them to send me a CD and have allthis data ready to be analyzed.

  • 10:55

    DR. BENEDICT WHEELER [continued]: If I went back and talked to myself,I might say to myself slow down a little bit.Just think about the research questionsyou're trying to answer.What's the best and most efficient wayto really capitalize on those datasetsto answer those questions?And I think the one critique I would have of my thesis--and well, there are various things to critique it on.

  • 11:20

    DR. BENEDICT WHEELER [continued]: But one thing I'd really pick up onis just how much I tried to do.Because there was so much.It's got three major datasets in itthat I analyzed using the census, and HealthSurvey for England, and the ONS longitudinal study.Really complex big datasets and lots and lots of analysis,pages and pages of appendices with tablesand-- which is all very informative and kind of useful.

  • 11:45

    DR. BENEDICT WHEELER [continued]: But maybe some of the things I could have done,I could have done a bit deeper and a bit less broad.[Do you think there is a place for social networking dataor big data in the health and environment field?]Big data is obviously a really big sort of current issueand a really good opportunity in all sorts of research fields.We actually just, a couple of weeks ago,hosted a meeting on big data for health and environment researchat our center.

  • 12:11

    DR. BENEDICT WHEELER [continued]: And really interesting different sorts of groups of people,different organizations coming togetherwho have very large potentially live kind of datasets.So one great example of that is the Met Office, also based downnear Exeter, have huge datasets clearly around the weather.And we have various research programsgoing on around climate and weatherand how it impacts on health and well-being.

  • 12:35

    DR. BENEDICT WHEELER [continued]: And so trying to bring together those really big datasetsand get the best out of them is a really good challengefor health and environment research.[If students were interested in becoming an academicin the health and environment field,would that be an area the field would take in the future?]I think it's one area where they could focus, definitely.And I think I'd always be wary about settingyour sights on something that's too much of a trendand only focusing on that one thing.

  • 13:04

    DR. BENEDICT WHEELER [continued]: There are lots of other opportunitiesas well to do research around these areas.Having said that, big data is going to be around for a whileas we get more and more data.The problem is not having data.It's having too much data.And how do you deal with that?And so there's a lot of methodological researchto be done on how do we get the best out of big datawithout getting completely lost in it.

  • 13:28

    DR. BENEDICT WHEELER [continued]: [What advice would you give a student who has a mathematicalbackground and would like to learn about quantitativemethods?]Yeah math is a really interesting one.I didn't do math a level.I would count myself as a quantitative research.I use a lot of statistics.But I wouldn't say my math is particularly strong.I think you've got to have a bit of ability with numbers,obviously it's quantitative.

  • 13:51

    DR. BENEDICT WHEELER [continued]: That's sort of the whole point of it.But I think you can do some introductory statskind of courses, read some introductory stats texts,and read around things like quantitative human geographythat are numeric but very contextualized as well.And find the bits that you're comfortable with,and also identify where you're goingto need to up your skills.

  • 14:12

    DR. BENEDICT WHEELER [continued]: And where you might need to go on a course or somethinglike that too, to really get to grips with specific issues.[What was the initial hypothesis for Beyond Greenspace?]Well, the initial hypothesis for that project, Beyond Greenspaceis the evidence around green space and healthand well being is quite mixed.It's all pointing in roughly the same direction.

  • 14:34

    DR. BENEDICT WHEELER [continued]: But there's some contradictory evidence around.It's not a straight forward as we might think it is.And maybe one of the reasons for thatis that green space is different in different places.It has different characteristics.So the hypothesis is, are their relationshipsbetween different types and qualitiesof different kinds of natural environments with healthand well-being?

  • 14:56

    DR. BENEDICT WHEELER [continued]: [What conclusions did you draw from the Beyond Greenspaceproject?]Well we're still at the stage of preliminary findings, I guess.We're just winding up the project at the moment.But I think our conclusions wouldbe, a, that methodologically it's stilla really interesting area to pursuethese linkages of geographical and particularly longitudinaldatasets are really interesting and good opportunities.

  • 15:22

    DR. BENEDICT WHEELER [continued]: Some of our findings are really interesting.So for example, some of our research is around the coast.So access to the coast as a sort of natural environmentthat lots of people have experience of visiting.And in fact, we have-- I think-- the last count wasover 10 million people in the UK livewithin five miles of the coast.No, sorry five kilometers of the coast.

  • 15:43

    DR. BENEDICT WHEELER [continued]: So that's a lot of people.And those people are much more likely to visit the coastmore often.Because it is much closer by.And there may be physical activityand psychological benefits that peoplegain from having that kind of environment on their doorstep.But there are also other people can gain even on their visits.You know, trips on holidays and so on.

  • 16:05

    DR. BENEDICT WHEELER [continued]: And so the coast appeared to be something important,particularly in this country.It might well be different in other countries.[Do you want to change how social policy is created aroundGreenspace, or would that be a secondary issue?]No, absolutely.Policy context is really important for this research.It's already happening.You look at the 2010 white paper on public health,it talks about green space and public health.

  • 16:27

    DR. BENEDICT WHEELER [continued]: There's a public health outcomes frameworkindicator on time spent outdoors in natural environments.And there's also stuff in natural environment policyas well around measuring and takingadvantage of the health and well-being benefitsof natural environments.So it's already in there, but I thinkthere's a real need for a more nuanced understanding of whatthose relationships are.

  • 16:49

    DR. BENEDICT WHEELER [continued]: Who's benefiting?How are they benefiting?What different kinds of environmentsdo different kinds of things for people?And so it's kind of picking this apart a little bit isreally important for the policy agenda, I think.[What is the difference between a research project witha policy agenda, and writing up a research project to bepublished in the academic world?]I think in an ideal world, they would all come together.

  • 17:09

    DR. BENEDICT WHEELER [continued]: And we try to do that as much as we can.So if we publish a paper, we'll tryto also have a briefing on our websitethat talks about the basics of it, what's in there,and try to have a summary in plain language.And increasingly, we're already doing thingslike writing short policy briefings.And also I think really understandingwhat it is that policymakers and decision makers and the publicactually want out of the other end of research.

  • 17:36

    DR. BENEDICT WHEELER [continued]: So actually asking people, what'sthe most useful format for you to get research findings in?Is it a one-page piece of paper?Is it a website?Is it tweets?I don't know.And trying to get a grip on what'sthe best sort of way to deliver the messagefor the different audiences.[What questions around this issue remain unanswered?]There are a lot of things left to answer.

  • 17:59

    DR. BENEDICT WHEELER [continued]: I mean, like I say, the research is-- the evidence baseis actually quite mixed.And so if we look at something like urban green spaceand physical activity, an area that you would thinkwould be reasonable to presume that peoplewho live in green or urban areas get more physical activity.It may not be the case.

  • 18:19

    DR. BENEDICT WHEELER [continued]: It's probably a bit more complicated than that.And certainly there are studies that say yes, thereare studies that say no.We know one of the best predictorsof accessing green spaces and doing physical activity in itis owning a car.So actually having the capacity to go somewhere else to do it.And also walking a dog is the other thing.So there are different motivations,different opportunities.

  • 18:43

    DR. BENEDICT WHEELER [continued]: But I think some of the key questionsare around mechanisms.There are lots of proposed mechanisms.And some of them have more evidence than others.So, thinking about natural environments, healthand well-being, it could be physical activity.It could be psychological processes,so there is the attention restoration theory.It's quite a well-known psychological theoryaround time in nature.

  • 19:08

    DR. BENEDICT WHEELER [continued]: Restoring what's called directed attention,so having your capacity to focus on stuff is limited.And maybe time outside helps with that, potentiallyas a stress reduction.There's also potentially the absence of a negative.So maybe greener urban areas have better air quality.So actually they're just sort of greenerareas have less negative impact of poor air quality,if you like.

  • 19:32

    DR. BENEDICT WHEELER [continued]: So those are some of the really important questionsto get stuck into.[Do the variables suggest themselves to you at the theorydevelopment stage, or do you discover them as you analyzedata?]I suppose this speaks to all secondary dataanalysis, which I think is usuallya bit of an iterative process.Ideally, you start out with your research questionsand your specific issues that you want to look into,which variables would be ideal.

  • 19:59

    DR. BENEDICT WHEELER [continued]: Then you find your secondary datasetthat contains that data when you find the variablesand you do your analysis.But the process is limited by what'sin the secondary dataset.So you have to have an idea of what'sin there already before you can draw up your hypotheses.Because if the data aren't there,you can't test the hypotheses.So you have to go around the houses a little bit with it.

  • 20:20

    DR. BENEDICT WHEELER [continued]: I think to find those variables and to pick outthe ones that really matter.[Have you ever attempted your own large scale survey?]It can be a useful thing to do if there are specific questionsto be answered in specific methodologies as well.So we've done a bit of research, for exampleusing accelerometers, which measurephysical activity in GPS loggers thatlog people's position in space.

  • 20:44

    DR. BENEDICT WHEELER [continued]: And so actually by getting peopleto carry those things around with them for a week,we can plot on a map exactly where they goand how much physical activity they're getting.And combining that with a questionnaire surveygives us a lot of deep informationabout what people do with their livesand how active they are in different spaces.So that kind of thing isn't somethingthat is generally available on secondary datasets.

  • 21:07

    DR. BENEDICT WHEELER [continued]: Although there's a bit of accelerometer on somethe Health Survey for England, for example.Sometimes with that kind of stuff,you have just got to do it yourself.But obviously, it's pretty time- and resource-intensive.[For students just setting out doing research,is there the opportunity for innovative methodologicalapproaches in that area of research?]Absolutely there is.

  • 21:28

    DR. BENEDICT WHEELER [continued]: And I think as particularly as those sortof technological approaches to individual monitoringbecome more and more ubiquitous, most people nowhave a smartphone in their pocket.So you can potentially do projectsthat track people, as long as people actually sign up to thatand that has all the right sort of ethical approvals.

  • 21:52

    DR. BENEDICT WHEELER [continued]: But people are more and more usedto having that data captured.In fact, lots of people do it themselves.There are lots of websites and appswhere you can record your own walking routes or runs or cyclemites or whatever.So people are used to doing that stuff themselves.So there's a real research opportunitythere to really make the most out of that.[Do you see researchers in the future working with companieswho collect big data?]I would think so, yeah.

  • 22:14

    DR. BENEDICT WHEELER [continued]: I certainly know there are various collaborations goingon where the big datasets that companies collect-- whichare clearly highly commercially valuable--tapping into those for scientific and socialscientific research can be potentially very fruitful.Clearly it takes a lot of negotiation, because it'svery, very commercially confidential,but also a massive resource.

  • 22:38

    DR. BENEDICT WHEELER [continued]: [When you wrote about your research,how did you decide which data to present?] Yeah that's a greatquestion.How do you pick and choose?For example, if writing a paper, youmight only have three tables and two figures.You would have to include you can't tell everybodyeverything that you've done.And you can't tell them all of your results.

  • 22:58

    DR. BENEDICT WHEELER [continued]: So I guess the important thing isa, telling the story that makes sensefrom your background and your hypothesis through your methodsto your results, to your discussion.So that has to flow.And it has to all tie together, but alsotelling a truthful story.So saying, being honest about the factthat you have generated other results as well.

  • 23:19

    DR. BENEDICT WHEELER [continued]: And we presented these results in these tables.We did do some other analyses as well.And they showed this kind of thing.Or increasingly now, what's hugelybeneficial is you can say, this is all in the paper.But see this online appendix for everything else that we did,so that you can be really transparent about your researchand what you've done, and how selectiveyou've been about the results that you're showing.

  • 23:41



Dr. Benedict Wheeler describes his work with secondary data, looking at how place affects health and well-being. He explains some of the pitfalls of secondary data research and offers advice for new researchers.

Looks like you do not have access to this content.

Ben Wheeler Discusses Using Secondary Data

Dr. Benedict Wheeler describes his work with secondary data, looking at how place affects health and well-being. He explains some of the pitfalls of secondary data research and offers advice for new researchers.

Copy and paste the following HTML into your website