for human security in Turkey and the Western Balkans Belgrade meeting Research methods and methodologies Dr Sally Stares 8 November 2013 Introduction This research programme is rich in both its diversity and its shared core ID: 480653
Download Presentation The PPT/PDF document "Cross-border citizens’ network" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Cross-border citizens’ network for human security in Turkey and the Western Balkans
Belgrade meetingResearch methods and methodologies
Dr Sally Stares
8 November 2013Slide2
IntroductionThis research programme is rich in both its diversity and its shared corein project themesin methodological approachesIn this presentation, some ideas for how to strengthen the connections between the different projects, in terms of
research questionsmethodological approaches2Slide3
Quick sketch of methodologies proposedDesk research – e.g. To identify key themes, and background information, e.g. statistics, legal frameworks, institutional arrangements, advocacy contextCase selection – e.g.
Probability (random) sampling – crucial for generalisation, inferenceKey informants, volunteers?Comparative (or single) case studies: selected for characteristics of interestData collection – e.g.
unstructured/semi-structured interviews, focus groups, media structured interviews, questionnaires, secondary analysis of existing dataData analysis – e.g. content analysis, narrative analysis, discourse analysiscounting frequencies or proportions of certain behaviours or perceptions3Slide4
Illustrations of diversity of our research questions4Questions of exploration:
What kind of threats exist to young people in virtual spaces?What are perceptions of insecurity?Questions of distribution:
What is the scale of conventional crimes committed by Roma population in North-Northwest Bulgaria?Are legal limits/standards of working time and sick support observed?Questions of process:What is the mechanism of controlled voting in small municipalities in Bulgaria?How are schools compromised as safe spaces?Questions of causation:What are the consequences of housing reforms in Montenegro?
How has the privatisation process increased vulnerability of former workers
?Slide5
Examples of methods for exploration5Qualitative approaches are well indicated, especially where respondents are free to raise issues, to set the agenda
InterviewsFocus groupsDesk research and analysis of existing data sources may reveal key insightse.g. survey results, text data
Quantitative data collection methods not so directly useful, because of their highly structured format, predetermined by the researchersAlthough, multivariate analyses can reveal interesting patterns and associationsSlide6
Examples of methods for distribution6Quantitative data collection methods are the most obvious choice
e.g. surveysThen, sampling strategy is crucialFor results to be generalisable to the target population, probability (random) sampling must be used
Typically outsourced to survey agencies who have necessary sampling frame information, field force, etc.Desk research and analysis of existing data sources may be a key tool here, if primary data collection cannot be donePossibility of a quantitative data collection exercise at a later point in the research programme?Slide7
Examples of methods for process7Qualitative approaches again well indicated
InterviewsFocus groupsSuggest that a narrative approach to these methods might be useful? Encouraging respondents to relay sequences of events, etc.Desk research and analysis of existing data sources may reveal key insights, and suggests avenues of enquiry for primary qualitative data collection
Quantitative data collection methods typically not so directly useful; tend to comprise a snap-shot of components rather than information on the dynamic links between themSlide8
Examples of methods for causation8Gold standard: randomised control trials!
To make claims of a causal relation between two phenomena (say, housing reforms and insecurity), need to fulfil three requirements:Demonstrate an empirical association between themEstablish the one doing the causing happened before the one that suffered the effect
Rule out all other possible explanations for the association between themExtremely difficult to achieve claims of causality, and harder as the extent of desired generalisation increasesHowever…connects to approaches such as process tracing, which maps on nicely to a more general narrative approach…and, I think, is expressed in softer form in core human security questions…Slide9
Core research focus9Understanding the spaces and forms of insecurity in the region, using
violence as a proxy for insecurityKey overarching questions:How does the particular form of violence under study happen, how is it manifested?
How do people respond to it?What do people want to change? What would they wish the situation to be?If possible to answer: how could that be achieved?1 and 2 speak to questions of exploration and process3 speaks to process and tentatively to questions of causationNone speak directly to questions of distribution! Maybe 1 is indirectly linked?Slide10
Questions for discussion (1)10Can your existing research questions be reframed in terms of these?
I have a hunch that they can, but I may be wrong!Means in many cases a slight rearrangement, e.g.CRDP wants to inquire on conditions of occupational safety in infrastructure projects in Kosovo, and examine the institutional support available to the vulnerable workers of this industry
Please tell me what it’s like to work here. What’s good, what’s not so good?How about your working hours, what are they like? And holidays?If people have problems, e.g. injury, what happens then?How do you manage the difficult aspects of working here?Do you have any support, e.g. from the company, some security from contracts?In an ideal world, what would you like it to be like here?
In the abstract, and in practice - any thoughts on what you would like to change?
Consider effects of calling security into question; double hermeneutic?Slide11
Questions for discussion (2)11Can we arrive at a shared strategy/elements of a shared strategy for case selection?
I think case selection is keyBeing explicit about the communities we are studyingHow far we intend the results to be generalisable
Empirical generalisation, e.g. classic case of opinion pollAnalytical generalisation, mapping out and identifying themes and issuesSuspect that formal sampling may not be possible in many instances?For qualitative research, can adopt a strategy of corpus construction, i.e. trying to discover all the relevant themes – keep sampling respondents until you are not hearing any new themesWhichever way, explicit documentation of respondent selection will be key to methodological rigour and quality of our researchSlide12
Questions for discussion (3)12(Broader version of second question)…
Exactly which parts of the different projects are core, shared, and which are unique or peripheral or idiosyncratic?In terms of Substantive topic (e.g. youth violence)Social group (e.g. Roma)Methods (e.g. types of information gained from interviews, from focus groups, etc.)
A while ago I developed a taxonomy of choices in research, which I find helpful for clarifying the scope and nature of different pieces of researchThen facilitates comparisons of different studiesOthers have found it useful, but don’t be obliged! Here it is anyway…Slide13
Taxonomy of choices13
Substantive research topic
Substantive theoretical framework
Research method / tool
Generalisability
Observations and data
Levels of analysis
Ways of representing
(analysing and reporting)
Relationship between researched and researcher / research instrument
ValidationSlide14
Taxonomy of choices14
Substantive research topic
Substantive theoretical framework
Research method / tool
Generalisability
Observations and data
Levels of analysis
Ways of representing
(analysing and reporting)
Relationship between researched and researcher / research instrument
Validation
Substantive research topic:
Concrete questions; Mary’s what, when, how, who, why?Slide15
Taxonomy of choices15
Substantive research topic
Substantive theoretical framework
Research method / tool
Generalisability
Observations and data
Levels of analysis
Ways of representing
(analysing and reporting)
Relationship between researched and researcher / research instrument
Validation
Substantive theoretical framework:
Human security
Other tacit frameworks of knowledge?
Contextual social, cultural knowledge
Theoretical framework may play a major or minor roleSlide16
Taxonomy of choices16
Substantive research topic
Substantive theoretical framework
Research method / tool
Generalisability
Observations and data
Levels of analysis
Ways of representing
(analysing and reporting)
Relationship between researched and researcher / research instrument
Validation
Research method / tool:
e
.g. survey, interviews…
Broadly,
How to select participants
What mode of data collection to use
Key variation in extent of:
P
ersonal contact
with subjects (internet surveys vs. in-depth interviews)
Intervention
(covert observation vs. action research)
Formality of structure
(experiments vs. participant observation)Slide17
Taxonomy of choices17
Substantive research topic
Substantive theoretical framework
Research method / tool
Generalisability
Observations and data
Levels of analysis
Ways of representing
(analysing and reporting)
Relationship between researched and researcher / research instrument
Validation
Observations and data:
Observations = information in its rawest form (e.g. tape recording of interview)
Data = information in analysis format
A set of observations can be converted into different types of data
Data not ‘given’, but involve creative choices
Sometimes observations = data (e.g. questionnaires); sometimes several steps from observations to data (e.g. text coded from interviews)Slide18
Taxonomy of choices18
Substantive research topic
Substantive theoretical framework
Research method / tool
Generalisability
Observations and data
Levels of analysis
Ways of representing
(analysing and reporting)
Relationship between researched and researcher / research instrument
Validation
Ways of representing
How to represent data to oneself as researcher:
analyse
How to represent findings to an audience
Sometimes synonymous, sometimes two distinct steps
Conceptual question: nature of representation (prose, numerical system?)
Technical questions: details of system (e.g. for statistical models)Slide19
Taxonomy of choices19
Substantive research topic
Substantive theoretical framework
Research method / tool
Generalisability
Observations and data
Levels of analysis
Ways of representing
(analysing and reporting)
Relationship between researched and researcher / research instrument
Validation
Validation
Reassuring audience and oneself of quality of research
For quantitative approaches, many angles
E.g. convergent validity, discriminant validity, reliability
For qualitative approaches, no direct equivalents, but…
Quality markers, e.g. richness of data, ‘thick description’, openness to surprise
Key that is often lacking: how would I know if this finding were wrong? i.e. guard against
verificationismSlide20
Taxonomy of choices20
Substantive research topic
Substantive theoretical framework
Research method / tool
Generalisability
Observations and data
Levels of analysis
Ways of representing
(analysing and reporting)
Relationship between researched and researcher / research instrument
Validation
Generalisability
Of substantive findings, to broader population/setting?
Requires probability sample
Of research instrument?
E.g. questions asked in interviews; standardised questionnaire
Empirical and analytical generalisationSlide21
Taxonomy of choices21
Substantive research topic
Substantive theoretical framework
Research method / tool
Generalisability
Observations and data
Levels of analysis
Ways of representing
(analysing and reporting)
Relationship between researched and researcher / research instrument
Validation
Levels of analysis
E.g. intra-individual, inter-individual, individual-group, societal-level, individual-societal
May employ multiple levels of analysis
Structural conditions for violence
Individual agency in the face of violence
Clarity on levels of analysis used is often lacking
Harre’s
‘distributed’ and ‘collective’ representationsSlide22
Taxonomy of choices22
Substantive research topic
Substantive theoretical framework
Research method / tool
Generalisability
Observations and data
Levels of analysis
Ways of representing
(analysing and reporting)
Relationship between researched and researcher / research instrument
Validation
Relationships between researcher, research instrument, research subjects
i.e. what happens during the research process, and how prominent this issue is
E.g. questionnaires assume respondents understand and answer questions in basically the same way; where they don’t it is a nuisance
E.g. in action/participatory research, relationship defines the project
Ethical concerns key here
Key nature of human securitySlide23
Possible next steps?23Desk research completed
To clarify substantive and theoretical frameworks; set out research questionsLevel and/or type of generalisability agreedLevels of analysis greed
Method(s) selectedTo best serve research questions, and given capacities, time frame etc.Cases selected‘Sampling’ procedure explicitly definedRelationship between researchers and subjects explicitly defined, especially re ‘transformative’ potential of human security as a topic; measures for managing expectations, duty of care to respondents and all affected; advocacy implicationsPlan made for forms of observations and dataE.g. interview notes coded in any way?Plan for how data will be analysed and reportedDifferent for individual projects than regional report? Can we devise a common core?
Can we devise a scheme for validation/quality assurance? E.g. a standard reporting frame for how the respondents were selected?