Escolar Documentos
Profissional Documentos
Cultura Documentos
Data analysis
Research design
1
Source: Translated from Andersen, Hansen & Klemmensen, 2010:67
1. Interesting 2. Relevant 3. Feasible: achievable, realizable. (Factible, viable) 4. Ethical 5. Concise 6. Answerable ELEMENTS OF THE RESEARCH QUESTION: POPULATION: All the members of the category under investigation. The universe which we wish to say something about, can be more or less specific. (E.g. Germans, municipalities in Denmark, tabloid newspapers). THE UNIT OF ANALYSIS: What/who is to be investigated (e.g. individuals, countries...). THE VARIABLES: The characteristics of the units (e.g. gender, employment status, GDP...). Any unit of data collection whose value can vary. THE VALUES: The placement of the units on the variables (e.g. male/female, employed/ unemployed, high/low). Range of possible variations available to a variable. ESSENTIAL QUESTIONS IN THE PROCESS OF IDENTIFYING A RESEARCH QUESTION: What do we know: theory and extant knowledge? What do we need to know? What you would like to investigate? Why is this interesting, important and relevant to investigate?
TYPES OF RESEARCH QUESTIONS I: DESCRIPTIVE: Mode of research that doesn't seek to explain why things are as they are, only to show what is going on. How big a proportion of the news do crime issues make up? EXPLORATIVE: Research that is design not to test a hypothesis by deductive means, but which instead aims to explore a field in a more inductive way. How do young people talk about politics? UNDERSTANDING: How do journalists motivate their framing of political issues? EXPLANATORY: Account that suggest a casual process behind data collection. Does the gender of political leaders affect citizens' perceptions of their leadership competences? PREDICTIVE: A statement about the future that can be tested if stated in a rigorous form (that is, as a hypothesis). What will happen to citizens' democratic trust if the use of negative campaigning is increased in Denmark? CHANGING: How should political candidates act to maximize the exposure of their view points in the media? TYPES OF RESEARCH QUESTIONS II: TESTING OR EXPLORING. TESTING: When extant theory, research and knowledge enables you to build predictions (hypotheses) about what to expect. It is deductive (theory data) and usually is associated with the use of quantitative methods. EXPLORING: When you have no tentative predictions, only questions. Research that is design not to test a hypothesis by deductive means, but which instead aims to explore a field in a more inductive way. CAUSAL RELATIONSHIPS. CRITERIA OF CASUALTY: Empirical correlation between the two variables The cause precedes the effect in time. The observed empirical correlation between the two variables cannot be explained away as being due to the influence of some third variable that causes both of them. 3
DEPENDENT VARIABLE: Variable that a researcher predicts will be affected by the variation of another variable (independent variable). Referred to as the Y in mathematical notation. It is what we wish to explain. INDEPENDET VARIABLE: Often referred to as the cause, assumed to be the variable influencing changes in the dependent variable. It is what explains the variation in the dependent variable. Referred to as the X variable in mathematica notation. CONTROL VARIABLE (third variable): A variable that the researcher suspects might influence the relationship between the independent and dependent variables. Referred to as Z variable in mathematical notation. Control variable (Z)
theory from evidence, rather than generating testable theory from rational extensions from existing theory. DEDUCTION: Theory (abstract) results (specific). Ir requires a greater degree of pre-emptive structure in the data collection process. HYPOTHETICO-DEDUCTIVE: Derives hypotheses from theory and tests the hypotheses in the data. RESEARCH DESIGN. It is a plan for the collection and the analysis of the data that contains: Which units are to be investigated. How many units you wish to investigate. When you want to investigate the units. Which data sources you want to use.
The research design should ensure that the data you collect enables you to give an answer to your research question. DIFFERENT TYPES OF RESEARCH DESIGN
Experiments Cross-sectional studies Longitudinal studies Single case studies Comparative studies
EXPERIMENTAL DESIGN.
Randomised division into test and control groups, manipulation of the independent variable Studies based on observations made at one time Studies based on observations made at many times In-dept investigation of a single case (but often several observations) investigation of a least two cases (the investigation focuses on differences and similarities i.e. comparison)
Research designed to test a hypothesis, usually through the establishment of controlled conditions and the manipulation of independent variables to measure changes in dependent variables. The scholars creates the situation which is needed to observe a phenomenon. Two or more groups are exposed to different stimuli and their subsequent response is measured. The scholar controls the values in the independent variable. The units of analysis are randomly assigned to values in the independent variable (ensures
control for third variables). The experiment is the logic which other types of research design seek to imitate.
CROSS-SECTIONAL STUDIES. Research design that involves collecting data from a sample at one point in time. The researcher is concerned with selecting many cases on the basis of variation in identified characteristics; these characteristics will often be the independent variable of the case, for example where an individual sex, marital status and age. Changes in the dependent variable, for example income, can be explored by examining the differences between men and women, or plotting the variation in income by age. The scholar collects data at one point in time focusing on two or more units of analysis. The values on the independent variable are given and must thus be observed. The units of analysis/cases can be selected randomly or based on theory. The scholar can comare many or few units of analysis. Example: a large-scale national survey which is collected at one given point in time to examine the sources of religious prejudice. LONGITUDINAL STUDIES. A research design that involves collecting data on cases over an extended period. The design will involve data collection from the same sample at two or more points in time. The data from each collection period can be compared to assess social change. The scholar investigates one or several units of analysis at different points in time. The values on the independent variable are given and must thus be observed. Tha units of analysis/cases can be selected randomly or based on theory. May include many or few units of analysis (e.g. individuals, countries, etc.). Example: Analysis of Supreme Court decisions over time.
SINGLE CASE STUDIES. Detailed examination of a single case. A case study may focus on one individual, one area, one group or one organization. It is not designed to compare one individual or group to another. Though it is possible to conduct a series of case studies, each study would not be designed specifically to enable comparison with others. Investigates one unit of analysis/case. 6
The case may be interesting in itself, but is often an example of a more general phenomenon. It is possible to include many different types of data/obervations. Can be analyzed over time.
COMPARATIVE CASE-STUDIES. Its focus is to identify differences and similarities between two or more different groups or cases. Comparison of a smaller number of cases where the main interest is to explain systematic differences and/or similarities. Often a very careful selection of cases: maximize variation in the independent variable control third variables do not choose on the dependent variable
Most similar system design (MSSD) vs. most different systems design (MDSD).
Y and not something else that is producing an apparent causal relationship? How confident can we be that the independent variable really is at least in part responsible for the variation that has been identified in the dependent variable? External validity: Concerns the question of wether the results of a study can be generalized beyond the specific research context; the extent to which research findings can be generalized to the population and different social settings. Generalizability. Ecological validity: Concerns the question of wether social research sometimes produces findings that may be technically valid but have little to do with what happens in people's everyday lives. RELIABILITY: Is concerned with the question of wether the results of a study are repeatable. When a data collection instruments records the same phenomenon, it is said to be reliable. This doesn't mean consistency of results every time, only consistency in the way, for example, a question is understood by interviewees. If a question was interpreted differently each time it was asked, or an experimental set of conditions were experienced differently by different participants, the responses generated could not be said to be reliable. Reliability can be tested during piloting (pre-testing of research instruments to identify weaknesses within it). Reliability is determined by how the measurement has been conducted. The concept refers to the accuracy of the phases in this process. If you repeat the measurement of the same phenomenon in the same way you should get the same results. Reliability problems often have the form of random mistakes.
High measurement validity and high reliability Source: Freely after Andersen, Hansen og Klemmensen, 2010:102
SINGLE CASE STUDY: CHARACTERISTICS: Intensive investigation of a single case/unit depth internal validity rather than generalizability. Research that is non-comparable (David & Sutton 2004). There is a limitation on this affirmation, every research is comparable to others to some extent. Often aiming a rich description and detailed understanding (holistic). Often inductive seeks to operationalize concepts and categorize in the process of data collection.
STUCTURED, FOCUSED COMPARISON (George & Bennett 2005) Structured: ensuring meaningful comparison by using the same template for data collection on each case. Focused: delimitation to certain aspects/variables of cases. Comparison: Similarities and differences between cases are used to explain outcomes (control of variation). CONTROL OF VARIATION: A FEW EXTRA WORDS. Experiments: variation in relevant independent variables (control variables) is handled by randomly dividing the units of analysis into experimental and control groups, and the manipulating the key independent variable. Alternative: to hold control variables constant: Comparative designs: separate analysis for each value on the control variable. Statistical controlled.
TWO CLASSIC VARIANTS OF COMPARATIVE CASE STUDIES: MOST SIMILAR SYSTEMS DESIGN: Attempts to exclude possible causes (control variables) and identify a causal relation (co-variation) between the key independent variable and dependent variable for otherwise comparable units. Recommendation: Maximize variation in the independent variable (X) and minimize variation in control variables (Z).
X 1 2 3 4 5 + + + Y + + + Z1 + + + + + Z2 Z3 + + + + +
10
MOST DIFFERENT SYSTEMS DESIGN: Attempts to exclude possible causes (control variables) and identify a causal relation between independent and dependent variables across variation in control variables. Recommendation: maximize variation in control variables (Z) and hold the independent variable constant (X). Problem of casualty: no variation in dependent variable (Y).
X 1 2 3 4 5 + + + + + Y + + + + + Z1 + + Z2 + + + Z3 + + +
3. CASE SELECTION.
CONSIDERATIONS FOR CASE SELECTION: SINGLE CASE STUDIES: Purpose of study? Ongoing case selection following the process. Snowballing? Variation and control of variation. Focus on internal or external validity? Intentional selection (different from random sampling). Reflect research objectives. Cumulation. Avoid selection bias. Pragmatism in data collection process.
BOTH:
11
Often more inductive approach: giving priority to issues raised by those researched; a degree of openness in questions... But can also be deductive: using qualitative methods to test hypothesis (e.g. group interviews with vignettes).
PROS OF QUALITATIVE METHOD: Is more holistic and context-sensitive. Can give us access to knowledge we cannot get in other ways (e.g. Geertz' blinking boys). Is flexible in terms of possibilities of readjusting focus of research, question asked, sample... 12
CONS OF QUALITATIVE METHODS. It is a trade-off: when you go deep you cannot also go wide! It is time- and space-consuming.
NB: Whether we should use qualitative methods or not depends on our research question and data sources! QUALITATIVE DATA SOURCES: Interviews, individual or focus groups. Documentary material: laws, white papers, parliament debates, archives, correspondences, secondary literature, etc. Newspaper articles. Field notes (observation). Blogs, websites, sms Flyers, brochures, posters Non-textual material: tv, movies, art works, photographies, etc.
particular interest to the researcher. SNOWBALL sampling: identifying subsequent members of a sample by asking current members of the sample to identify other participants with the required characteristics. Often used where no sampling frame can be identified or constructed. (E.g. interviewing until saturation). Remember: We choose our material according to our research question, not in order to confirm our hypothesis. In qualitative research we have fewer units of analysis, so our sampling must be even more precise. Qualitative research should be just as systematic in sampling as quantitative research.
STANDARDIZED VS. STRUCTURED: Depends again on the focus of the research: Is it important to get directly comparable answers? - closed answers, ordered sequence. Is it important to tab interviewees life worlds in-depth? - Open answers, flexible sequence. Our focus here: Open-ended questions and semi-structured interview guides! 14
Structure refers to the degree to which the form and order of questions asked are kept identical from one interview to another. It seeks to maintain high levels of reliability and repeatability. The more unstructured interview seeks to emphasize the depth validity of each individual interview. Standardization refers to the level of closure placed around the answers interviewees can give. Closed answers allow greater scope for quantification. Open answers allow for greater depth and personal detail (more typical cualitative). Examples: Standardized questions: State your age in years... Have you ever been in UK? Yes/No How afraid are you of being robbed? (a) Not at all, (b) A little, (c) Afraid, (d) Very afraid. What type of accommodation do you live in? (house, flat, hotel, caravan...) What is your favorite sport? What university clubs and societies do you belong to? Please, describe the process that led you to study this course. Men should work, women should stay at home. Discuss. How do you feel the world has change since you were a child? Reliability Repeatable Measurement validity Internal validity Generalizability (external validity)
Semi-standardized questions:
Unstandardized questions:
INTERVIEWEES: WHO? Conscious/theoretical sampling Snowball sampling Needs justification Often difficult to say in advance (saturation)
HOW MANY?
15
Depends on the number of relevant distinctions (age, sex, etc.) Kvale: 15 (+/-10)
BEFORE THE INTERVIEW: Have much knowledge of the topic? Creating contact to interviewees Briefing (introduction and summary at the beginning) Think of location: comfort, tranquility, naturalistic setting... Creating and testing of interview guide (piloting)
THE INTERVIEW GRUIDE: THEORETICAL GUIDE: Builds on theoretical concepts and research questions. Lists concepts and questions as interview topics. Consider the ordering of the questions. The actual question one plan to ask interviewees. Normally short and easily understandable questions (no theory). Focus on concrete actions, experiences, events, attitudes, etc.
OPERATIONALIZED GUIDE:
- Do you find the subjects you learn important? What type of motivations for learning are - Do you find learning interesting in itself? dominant in high schools? - What is your main purpose in going to highschool? - Have you experienced a conflict between what Do the grades promote an external instrumental you wanted to read and what you had to read to motivation at the expense of an intrinsic interest obtain a good mark? motivation for learning? - Have you been rewarded with money for good grades? THE STRUCTURE OF THE INTERVIEW: Briefing
16
Control and power asymmetry Creating confidence and a good atmosphere How critical can/should one be? Warm up questions to establish rapport. Demographic questions. Core questions covering all key themes within the research question. Prompts and probes to elicit more detail and depth that are initially forthcoming. Clarifying questions to check interviewees' understanding of what they have been asked. Debriefing
CONDUCTING THE INTERVIEW: The dynamic dimension: flow in conversation Types of questions Being sensitive: Listen! Avoiding bias/interviewer effects Ongoing validation
PILOTING: Pretesting of research instrument such as the questionnaires or interview schedules with a small subsample of the target population to identify weaknesses within the data collection instrument.
18
Text
Concept/Codes
Category
Data material that we want to The immediate associations our The topic which our concepts/ analyze reading generates codes can be subsumed under
Focus:
TEST
Dilemma
between
ideals
and
EXPLICATING AND VALIDATING CODING PROCEDURES: Memos Code book/code framing Inter-coder reliability Intra-code reliability
19
to test it with our research. WHEN TO USE DEDUCTIVE ANALYSIS OF QUALITATIVE DATA? When you have formulated a clear hypothesis (test). When you want to contain a theoretical focus in the investigation. When you want to maximize the contribution to specific research field.
PROS AND CONS OF USING DEDUCTIVE ANALYSIS OF QUALITATIVE DATA: PROS: A clear question generates a clear answer A clear answer generates a clear contribution to the literature Having clearly defined categories/codes in advance makes it easier to cover large bodies of texts CONS: Is reality ordered after our theoretical categories? Would different coders reach the same result?
CODE LISTS/CODE BOOK Structured, systematic presentation of codes and sub-codes (definitions and descriptions). Sub-codes can de types of the main-code or values on a variable. Start code list vs. end code list (often not the same)
Variable PROGNO1 (PROGNO2, PROGNO3) Variable label: prognostic frame of claim Note: A prognostic frame describes the claimants vision of what should be done to solve a problem/the issue at hand. Thus, a prognostic frame is the presented ways of solving the identified problem, the hypothesised new social patterns or the suggested alternatives. Often a prognostic frame will be a call for specific or more general action Value labels: 10 FREEDOM OF SPEECH 11 limiting freedom of speech (including legal revisions) 12 widening freedom of speech (including legal revisions)
21
20 EQUALITY/PARITY 21 general call for equal treatment of Muslims and non-Muslims 22 abolishment of the state church 23 no special rights/treatment 24 Muslims should treat non-believers with respect in order to receive respect for 25 fair treatment of Muslims in the Media 29 other equality/parity prognosis their believes
30 DISCRIMINATION 31 general call for anti-discrimination/islamophobia measures 33 changing the tone of the debate 39 other discrimination prognosis
CLOSED/DEDUCTIVE CODING Hypothesis: Closed political opportunity structures made Danish Muslims perform a scale-shift in the Cartoons.
Text Interviewer (I): Tell me about the delegations that went to Egypt and Lebanon in December 2005? Interviewperson (IP): Well, after having contacted the ambassadors, the Prime Minister, after we petitioned and demonstrated for three months, and nobody took us serious, we thought we had to try a last possibility. I: The delegations? IP: Yes, we went abroad and tried to raise awareness of our demands there. Code
Upwards scale-shift
2. DISPLAYS:
It is a technique that helps to illustrate and draw conclusions. Condensed, visual depiction of
22
qualitative data, which facilitates analysis and helps to detect patters, anomalies, trends and themes. Matrix displays: tables that present cross-tabulations of variables or codes. Network displays: graphs that depict flow of events and processes of connection.
DISPLAYS: WHY? Problem 1: Qualitative data analysis often leads to extended texts, which never gets around to analyzing data at a higher level of abstraction. Problem 2: Qualitative research reports tend to glose over the move from data to conclusions. Solution: Displays can condense qualitative data, visualize key findings and aid analysis.
TYPES OF DISPLAYS: Within-case displays: present characteristics of flows of one case/unit of analysis. Cross-case displays: highlight similarities and differences between cases/units of analysis. Time ordered displays. Conceptual displays.
Rejection of ostentatious Fight for Islamic symbols in religious symbols in public public Religion integrated in everyday life Assimilation Acceptance of cultural modernity Methodist and systematic religiosity Integration Search for alternative modernity
- discontinuation of - Government - state sponsored the legal case against decision to no longer Dialogue JP by the Danish State work with imams. conference in DK Prosecutor - Intervention by the - reprints of the Danish excaricatures - US, NATO and EU - Manifesto by ambassadors supports Rushdie, Hirsi Ali & The Pakistani bounty - flag burnings co. on the illustrators - Demonstrations - Reopening of Danish against the caricatures embassies abroad. all over the world - Fogh distances himself from the drawings JP apologizes To To To To
Time period
24
Sociological:
25
Moderate version Critical discourse analysis LANGUAGE AND SOCIAL REALITY: 2. POINT Social reality is only accessible through language Social relations are relations of power
LANGUAGE
REALITY
Thus, linguistic analysis of social relations are power analysis LANGUAGE AND SOCIAL REALITY: 3. POINT Linguistic analysis of social relations are power analysis Both linguistical relations and social relations are constantly negotiated
Thus, linguistic analysis provides a snapshot of power relations in a given context. CLASSICAL DISCOURSE ANALYSIS A linguistic analysis (point 1), which aims at making visible dominant power relations (point 2), which are constantly changing (point 3). CLASSICAL DISCOURSE ANALYSIS: HOW? Analyzing: Articulation: e.g. smoking used to be cool, now it is considered disgusting Nodal-points: key concepts, central/main ideas Antagonisms: opposition, polarization (how the discourse constructs it) Hegemony: dominant discourse having the monopoly on the articulation Chains of equivalence: associations, things that become alike Empty signifiers: concepts that are often so wide that are filled with many meanings and characteristics. CHAIN OF EQUIVALENCE. EXAMPLE: In Denmark we have for generations built a safe, rich and free society. The decisive glue of society has been and still is our common values. Freedom to be different. Equal opportunities for men and women. Responsibility. Democracy. Respect for the laws of society. Moving the Ghetto back in society, Regeringen 2010 Society= Denmark for generations = safe = rich = freedom = equality = responsibility = democracy 26
= respect for the law However, there exist in Denmark today areas where the Danish values are no longer fundamental. And where the rules that apply in the rest of society do not have the same effect. This is the situation in those residential areas that we in everyday language call Ghettoes. The Ghetto = different from the rest of society = (unsafe) = (poor) = (unfree) = (inequality) = (no responsibility taken) = (undemocratic) = (no respect for the law)
% %
()
Social practice - macro-sociological analysis of context Discursive pratice (production, distribution and consumption) - micro-sociological analysis of how text relates to existing discourses Text (discursive event) - Linquistical analysis of how a discourse/genre is constituted
Source: Fairclough: Discourse and Social Change (1992) 27 Jrgensen & Phillips: Diskursanalyse som teori og metode (1999)
ANALYSIS OF THE TEXT: Choice of wording Grammar (tense and person) Textual structure (analysis of genre)
Central question: Why exactly this word, this tense and this structure rather than others? ANALYSIS OF DISCURSIVE PRACTICE: Production: How is the text produced? Consumption: What audience is the text aimed at? Inter-discursivity: How does the text draw upon other existing discourses? Interpretation: How is the text read/understood by the readers? Intertextuality: How does the text interact with existing texts? ANALYSIS OF SOCIAL PRACTICE: Ideology: How does the text relate to existing ideologies? Hegemony: Which discourses are dominant in the social order/discursive order? Social processes: a) Social stability reproduction of discursive/social order b) Social change challenge or re-formulation of discursive/social order
28
ADVANTAGES OF QUANTITATIVE CONTENT ANALYSIS: In the most simple form (computer counting of word/phrases) can give an overview of patterns in data and developments (e.g. Google Trends). From words to numbers and back again (combining quantitative and qualitative analysis), gives the opportunity to go from abstract, quantified patterns, to concrete ddetails. Makes it possible to analyze data in-depth as well as at large (e.g. analyzing an entire debate, one actor, one topic, one phase, etc.) Deduced categories/variables and induced sub-codes/values Enables the use of descriptive statistics and statistical comparison (e.g. between actors, time,
29
place, etc.) Can be used for both explorative studies and hypothesis testing. Gives the opportunity to answer research questions such as: to what extent..., how often..., what is the typical...
DISADVANTAGES OF QUANTITATIVE CONTENT ANALYSIS: Reduction of reality data/information is lost Says little about how meaning in texts is produced (discourse analysis). Methodological over-kill: cab't we get the same results from coding much less text? Time consuming Problems of coding reliability Selection bias: e.g. newspapers will only report on certain events.
CLAIM-ANALYSIS: WHAT? A version of quantitative content analysis, based on deductive coding of qualitative data Focuses on claims-making = the expression of a political opinion by physical or verbal action in the public sphere (Statham et al. 2004) Combines protest event analysis and elements of discourse analysis (frame analysis) Sampling units are typically newspaper articles, while coding units are instances of public claims-making. CLAIM-ANALYSIS: WHEN? When we are interested in the amount, character, change, etc. of public claims-making on a given topic, in a given area or by a given actor. When we want to study coalition formation, conflict lines, relative positioning, etc. within a field. Ideal when we want to study a delimited public controversy or period of public debate. When we want to be able to make generalizable statements (external validity). When your data is not produced with quantification in mind.
CLAIM-ANALYSIS: HOW? 30
Identification of instances of claims-making in a text: an actor who expresses a political opinion about the topic of interest. Coding of only the factual elements of e.g. a news artivle (the events/actions described and actual statements made). This means that we don't code evaluations, judgments and commentary by the journalist.
Coding of both physical actions and speech acts referred to in a text, speech acts being the discursive statements in a debate (found in press releases, speeches, interview comments, complaints, etc.)
Often coding of claims-making by all actors in a debate institutional actor as well as noninstitutional actors (multiorganizational field)
FRAME ANALYSIS THE DISCURSIVE CONTENT OF CLAIMS. Focuses on how actors try to construct an argument, and frame a message so that it mobilize support and de-mobilize adversaries. Deconstructs arguments what is the analysis of the situation, what needs to de done, and why. FRAME, FRAMING AND RESONANS. FRAMES are (according to Goffman) schemata of interpretation, which enables individuals to localize, perceive, identify and name events in their everyday life and the
31
world at large. By ascribing meaning to events/situations frames organize our experience and guide our actions (frames as strategic vs. unconscious constructs). FRAMING refers to the process by which actors ascribe meaning to events/situations in a way that aims to mobilize support and demobilize adversaries. RESONANCE emerges when framing is successful when frames successfully speaks to individuals existing perceptions and situations, and makes them responsible to the content of the message. (See Snow & Benford for constraints on resonance-building). THREE CORE FRAMING TASKS: DIAGNOSTIC FRAMING: identification of a problem and ascription of the blame/explanation for the problem. Often one explanation/factor is highlighted instead of others. Example 1: The focus within Marxism on capitalism/the capitalist, and the exploitation of the worker as the main problem. Example 2: Al-Qaeda's blaming of the West oriented elites or crusaders as the root of all bad in Muslim societies. PROGNOSTIC FRAMING: the proposed solution to the identified problem. The indication of strategies, tactics and goals. Example 1: The suggestion within Marxism of class struggle, revolution and the communist society as the solution. Example 2: Al-Qaeda's yihad by the sword against the crusader alliance. NOTE: is most often around prognoses that organizations, groups or movements fight internally. MOTIVATIONAL FRAMING: the indicator of a rational for action. Motivational frames are registers of motives for action. Example 1: The stressing within Marxism of the believe that the revolution is near, and capitalism is unjust (injustice frames). Example 2: Al-Qaeda's attempt to make violent yihad against the crusaders a religion duty to individual Muslims.
32
13-02-06, Copenhagen
criticises
for inviting only Democratic Muslims to an official meeting for blasphemy and racism
because Democratic Muslims only represents secular Muslims arguing a violation of international conventions of human rights because violence is not the solution, and because the prophet Muhammad is sacred to all Muslims
17-03-06, Copenhagen
Ahmed Akkari, spokesperson of a coalition of Muslim organisations Imam, Ahmed Abu Laban of The Community of Islamic Faith
(press charges)
03-02-06, Copenhagen
that a consortium of independent actors from Denmark and the Middle East should be created to solve the conflict as certain roumours are circulating explaining how the political system works in Denmark
08-02-06, Copenhagen
Democratic Muslims
e.g. positive ,negative or neutral evaluation of Muslim protests Who were Muslim friends and foes in the debate?
Who does the claim support who is it in agreement with Which alliances were formed?
Who does the claim criticize who does it disagree with Where were the major fault lines in the debate?
Is the claim stated in reaction to other claims or proactively Who is setting the agenda in the debate?
33
INTERVIEWS? CRITERIA ADVANTAGES OF SELF- DISADVANTAGES OF SELFCOMPLETION COMPLETION (or disadvantages of structured) (or advantages of structured) Cheaper Fewer questions The respondents can complete Fewer complex questions the survey at a time convenient to themselves No interviewer effects No help
GENERALLY
RELIABILITY
MEASUREMENT VALIDITY More honest answers (specially Questions can be read and to sensitive issues) answered in a different order (not internet-based surveys) EXTERNAL VALIDITY Distribution over geographical area limited timespan a wider We don't know who answers within a Difficult to reach the illiterate Often low responses rates 34
QUESTION FORMATS Closed-ended: questions which require the respondent to select from a range of stated answers. (are you male or female? A. Male B. Female) Open-ended: unstandardized questions which enable the respondent to enter a response in their own words. (what words you use to describe people who receive social welfare?)
35
RANKING QUESTIONS
DEVELOPING THE WORDING OF QUESTIONS: Challenges: Questions are answered very quickly. The respondents answer question based on their extent knowledge and the associations which are triggered by the question. Your task: To make it simple and easy for the respondents. To make the question wording clear, direct and simple.
You want to measure differences in opinions, values and believes, not differences in people's understanding of the question.
RULES: SURVEY QUESTIONS Avoid ambiguous terms Avoid technical terms Avoid asking two questions in one question Avoid negations Avoid long questions Avoid too general questions Avoid leading questions
RULES: RESPONSE CATEGORIES Exhaustive Mutually exclusive categories Response categories should match the question balanced Neutral response category? (don't know/ don't answer / don't care) Don't know option? Be careful with dichotomous response categories
36
AN
EXAMPLE:
ATTITUDES
TOWARDS
THE
GOVERNMENT'S
ECONOMIC
RELIABILITY AND MEASUREMENT VALIDITY Take-away point: the rules for questions wording and response categories are important because violations may damage the reliability (increase random errors in the measurement) and the measurement validity (increase systematic errors in the measurement). SURVEY LAYOUT From the individual question to the survey questionnaire things to consider: Effects of the question order The use of routing (enrutamiento) questions and funneling (canalizacin) questions Maintaining the respondent's motivation. Avoid response-set
VARIABLES AND LEVELS OF MEASUREMENT: A general way of categorizing variables. Different statistical analysis techniques available for the four data types: 1. Nominal It is a level of measurement where response categories cannot be placed into any specific order and no judgement can be made about the relative size or distance of one category to another. When you only have two possible answers it is a dichotomy. Examples: gender, party preference. 2. Ordinal Level of measurement that is applied to categorical variables whose response categories can 37
be placed into a rank order of importance: for example rating scales. No mathematical calculations can be made in relation to the distance between the categories. Example: Attitudes (Likert Scale: completely agree/ agree/ don't care/ disagree/ completely disagree) 3. Interval/Ratio. Level of measurement where data are measured on a continuous scale. Data can be placed in rank order and can be subjected to mathematical calculations. The distance between the observations is known and can be calculated. Ratio has a true zero point, while interval doesn't. Example: age, sympathy (on a 10-point scale), number of sources in a news story.
4. SAMPLING
POPULATION DATA OG SAMPLE DATA The objective of data collection is to collect information about the units of analysis dictated by the research question. The population data /census: all relevant units of population of interest are observed. It is often the case with individual level register data e.g. criminal records, education, health... Sample survey: a sample of the population is observed. Typically used in opinion surveys, assessments, user surveys. POPULATION: Every possible case that could be included in your study. It will be defined by the nature of your enquiry (David & Sutton, 2011: 226) SAMPLING FRAME: A list of every unit in the population. EXAMPLE: Population
Danish municipalities Countries Danes
Sampling Frame
List of Danish municipalities List of all countries acknowledged by the UN? Civil registration system, the phone book?
38
TYPES OF SAMPLING: Probability samples: Simple random sampling: random selection from a sample frame. Systematic sampling: random selection mathematically done. Stratified sampling: random selection within strata which reflect the population. Cluster and multi-cluster sampling: sample based on naturally occurring groups.
Non-probability samples: Used when no convenient sampling frames of the population are available or because of time or cost restrictions. Limited generalizability.
Technique
Characteristics
Representativeness
Our point of reference
Simple random sampling Systematic sampling Stratified sampling Cluster sampling Non-probability
Random selection from a sampling frame The first unit is selected by random numbers and every xth subsequent unit is then selected Random selection within strata which reflect the population Selecting a sample based on naturally occuring groups within the population -
The same
Stronger
Weaker
Much weaker
39