Skip to main content
Search form
  • 00:00

    [MUSIC PLAYING]Hi, my name is Michael L. Dougherty.I'm an Assistant Professor of Sociology at Illinois StateUniversity.My areas of specialization include the sociology

  • 00:21

    of development and environmental sociology.In particular, I'm interested in mining conflictsin the developing world.So what I want to start with is talking about howone goes about taking two different data types--and, in this case, I'm talking about survey dataand ethnographic interview data--

  • 00:41

    and integrating these into one coherent storyfor one single manuscript.And in order to do this, I'm goingto draw from a recent article of mine,which was published in the Journal of Cleaner Production.And it was written together with my colleague Tricia D. Olsenat the University of Denver.The title of the article is, "They have good devices: trust,

  • 01:04

    mining, and the microsociology of environmental decisionmaking."There are three main advantages, the rationalethat I want to offer you for integratingdifferent types of data into one research project.

  • 01:27

    First of all, I would contend that the arguments that youdevelop when you draw from two different data typesare more robust.Secondly, I would argue that the findings aremore reliable because you have, basically,two different studies that corroborate one another usingtwo very different.Methodologies What I mean by the arguments being more robust

  • 01:51

    is that not only do you answer the "what"question or the "why" question, but you can answer the "what"and the "why" together.So you can tell a fuller story, youcan paint a fuller picture of the social phenomenathat you're interested in studying.

  • 02:13

    Mining investment in the developing worldhas grown dramatically over the past decade and a half,by 300% by some counts.And this, as you might imagine, hasled to a dramatic increase in mining conflictsin rural parts of the developing world that

  • 02:34

    are being asked to host mineral investment and mineraldevelopment, in many cases for the first time.My work, in general, tries to understandthese phenomena-- the social dynamics of mining conflicts.With this particular paper, we wereinterested in exploring the individual dynamics,

  • 02:56

    the rationale on the individual level for howpeople align in the context of a mining conflict.So lots of research looks at communitiesand asks questions about why certain communities supportor oppose mining.But very little of the research looks at the individualand asks questions about the social psychology

  • 03:18

    of the individual and how the individual makessense of mining, grapples with and comes to terms with miningin agrarian and, sometimes, indigenous communitiesin the rural developing world.So that was the focus of this particular paper.

  • 03:42

    I want to highlight three key lessons about mixed methodsdata analysis.First of all, mixed methods data analysis must be iterative.And I'll explain what I mean by that in just a moment.Secondly, mixed methods data analysis

  • 04:03

    must be somewhat inductive.And this is, of course, related to the first point,which is that mixed methods data analysis must be iterative.Thirdly, and finally, mixed methods data analysismust begin during field work.Now, what I mean by that is that we shouldbe careful not to make unnecessarily functionalistic

  • 04:25

    distinctions between the data collection phase and the dataanalysis phase.Rather, we should integrate these two.So while we are in the field collecting data,we should also be analyzing the data.We should be listening to it at the same timethat we collect it.But let me back up.And let me talk for just a minute about what

  • 04:47

    I mean, more specifically, when I say "mixed methods dataanalysis must be iterative."And to illustrate, to underscore this point I'mgoing to give you some examples from my process integratingthese two different data types-- ethnographic dataand survey data-- with this particular project.While I was in Guatemala conducting

  • 05:09

    exploratory interviews, I began to pick upon a theme across the interviews that I was conducting.And that theme was the idea of trust.In almost every interview in the early stage of data collection,the people that I was interviewingwould talk to me about trust.

  • 05:30

    Trust was very thematic, very important, very salientin these interviews.And as I listened to what people were saying,I began to realize that they wereusing the word "trust" to refer to two very different things.On the one hand, they were using the word "trust"to refer to what we came to call relational trust.

  • 05:54

    I trust you intuitively because youcome from the same walk of life that I do.The second type of trust that people would talk about,which is quite distinct from relational trust,is what we came to call institutional trust.And this is a more abstract kind of trust.And it's not reciprocal.It's uni-directional.

  • 06:15

    And this is trust in what we cameto call external institutions of authority.I started to hear about these two different kinds of trust.And then I decided that I needed to design variablesand measures and integrate these into a surveyin order to be able to measure this kind of trust,

  • 06:37

    in order to be able to operationalize it.And so here we have the first exampleof the iterative relationship between data types.I've got these themes that emergedfrom the ethnographic interviews at the exploratory phaseof data collection that then inform the survey design.So we conduct the survey.

  • 06:58

    And we leave the field.And now we're back in the United States entering the data,cleaning the data up, and analyzing the data.And what we found was that there weresome very strong relationships between someof these variables.We noticed that the trust variableshad really strong relationships to another variable that wehad called self efficacy.

  • 07:20

    And self efficacy is the idea that onebelieves in one's ability to affect changein one's own life.Also, these trust variables and the self-efficacy variablehad very strong relationships with the variablesthat measured mining support and mining opposition.

  • 07:41

    So there were some very strong relationshipsthat emerged through the inferential statisticalanalysis.But of course, we didn't know whythese things related to one another in the waysthat they did.We only knew that they related quite strongly.So once we had these relationships establishedin the inferential statistical analysis,

  • 08:02

    we went back to the ethnographic analysisto try to explain the causal mechanisms thatunderlie these relationships, which we were able to do.So when we were first listening to these interviewsand first analyzing this ethnographic data,we weren't listening at all for this idea of self efficacy.

  • 08:24

    That became salient to us once weanalyzed the quantitative data.So then we went back to the interview data,particularly listening for this idea of self efficacy.And we saw that it was there.So here we have a story that quite clearly demonstratesthe iterative nature of doing mixed methods data analysis.

  • 08:47

    We have interview data that then informsthe design of the survey and then analysis of the surveythat then informs the way we read the interviewdata once we return to it.So this is what I mean by "mixed methods dataanalysis must be iterative."

  • 09:07

    So the second theme that I want to highlightis that mixed methods data analysismust be somewhat inductive.And this is related to the assertionthat mixed methods data analysis must be iterative,but it's distinct.I don't think we should be fully inductive whenwe go into the field to do mixed methods data analysis.

  • 09:27

    In other words, I don't think youwant to approach this from a grounded theory perspective,for example.You should have some idea of what it is you're looking for.But you also need to be open, otherwiseyou will close off the possibility of discovery.So if you go into the field with really strong researchquestions or really strong hypotheses or really strong

  • 09:48

    theoretical frameworks, you close offthe possibility of discovery that you need in orderto properly cultivate the iterative, dialogic feedbackprocess between the two different types of data.The third lesson that I want to highlightis that mixed methods data analysis

  • 10:10

    must begin during field work.And I've talked about this before,but let me just reiterate briefly.It's very important, while in the field,that you're already listening to the data,that you're already analyzing the data.And it's important that we don't have really strong distinctionsbetween the data collection phase and the data analysis

  • 10:33

    phase.It's really effective if we integrate thosebecause that's the way we get this iterative dialogic kindof conversation going between these data types.So just to recap, three main lessonsthat I want to highlight-- mixed methodsdata analysis must be iterative.It must be inductive, or at least somewhat inductive.

  • 10:53

    And it must begin during fieldwork.Now, I would like to move to discussingsome of the key challenges that wemight encounter in conducting mixed methods data analysis.First of all, there's a challenge to coherence.

  • 11:16

    And there's also a challenge to how you actuallygo about organizing your manuscript.And I've done this in different waysat different points in my career.It can often be a struggle.And I think we might understand this intuitively.And, in fact, this might be one of the reasonswhy it's relatively uncommon that, as researchers,

  • 11:37

    we undertake mixed methods analysis.It can be a struggle to integrate these two datatypes into a single narrative thatmakes sense, that tells a coherent story,and that still weighs in at less than 8,000 wordsor whatever it is that the journal that you're writing forneeds from you.

  • 11:58

    I've organized mixed methods manuscripts in different waysat different times, as I've said.In this particular article, "They have good devices,"I organized the analysis by a data type.So there was one section for the statistical analysis.And there was one section for the ethnographic analysis.

  • 12:18

    And then there was a discussion section that sort of integratedthe two.But in previous articles, I have organizedthese things along themes.And then, within one subsection that's organized along a theme,I've got quotations from the ethnographic data.And I've also got data tables from the quantitative dataintegrated into the same section.

  • 12:40

    And you can do this however you like,whatever kind of suits the story that you want to tell,and what works best for the articlethat you're trying to produce.But regardless of how it's organized,the key is to make sure that you'retelling a coherent story.And the key is to make sure that the data types complement

  • 13:00

    one another but don't duplicate one another.Finally, it doesn't always work.And that kind of goes without saying,but it's worth mentioning.Don't force it.If you're qualitative and your quantitative findingsdon't dovetail in the way that you imagined they would--

  • 13:22

    if they don't talk to each other in the waythat you had hoped for, if they don't corroborate each other--that's OK.Don't force it.Don't make them tell a story that they do not organicallytell.Instead, maybe the story becomes somethingabout why the findings from these two different data types

  • 13:43

    don't, in fact, corroborate one another.And that can be a perfectly legitimate story to tell,as well.So we should always be open to the possibilitythat our mixed methods data analysis might notwork out exactly how we intend.And keeping an open mind about thisis part of the research process.

  • 14:09

    Finally, I'd like to close with a few reflective questionsfor us to think about.I have been advocating, over the course of this video, the valueof doing mixed methods research and of integratingmixed methods data.But there might be circumstances in whichmixed methods data analysis is not always the best approach.

  • 14:32

    Are there some questions that lend themselvesto a different approach?Are there some logistical or other circumstancesin which a mixed methods approach might not be ideal?And then, finally, if, as I claim,mixed methods data analysis is so superior to other kinds

  • 14:53

    of data analysis because your findings are more reliableand your arguments are more robust,than why is it not more common that we, as scholarsand researchers, conduct mixed methodsresearch and mixed methods data analysis?If it's so important and so valid and so robust,why is it so infrequently employed?

  • 15:16



Academic Michael L. Dougherty presents a case study on mixed methods research. In his examination of mining conflicts, Dougherty used both survey data and ethnographic data to give depth and context to local responses to mining. He describes the key benefits and challenges of mixed methods research, particularly highlighting the difficulties it can pose in reporting findings.

Looks like you do not have access to this content.

Researching Mining Conflicts Using Mixed Methods

Academic Michael L. Dougherty presents a case study on mixed methods research. In his examination of mining conflicts, Dougherty used both survey data and ethnographic data to give depth and context to local responses to mining. He describes the key benefits and challenges of mixed methods research, particularly highlighting the difficulties it can pose in reporting findings.

Copy and paste the following HTML into your website