Collecting the Right Data Dr Stephen Hincks Department of Urban Studies and Planning Todays Session… 1. Research Design – recap 2. Generating Data - methods 3. Sampling 4. Using Case Studies in Research The research design… ? Methodology vs Method • Methodology – Systematic framework that guides the research – Ontology – positivist, constructivist… – Steps and practices • Methods • The ‘tactics’ or ‘tools’ used by researchers to collect data (Robson, 2002) – Case studies – Interviews – Statistical modelling…. Quantitative and Qualitative Approaches to Data Collection Data… “In the developed world, we often have far more data than we can ever use. In most cases, what is lacking is not data but an understanding of what is important ....” (Lawrence G. 1997 Indicators for sustainable development, in Dodds F. (ed) The Way Forward : Beyond Agenda 21, London, Earthscan.) Data Collection and Analysis • Collection and analysis are likely to be closely intertwined缠绕的;错综复杂的 • Iterative 重复的process of collection and analysis Collection Analysis Collection Analysis Collection Analysis Methods to Generate Data: Some key decisions… Will I use: • Secondary data collection (assumptions, design, collection, presentation, and sometimes analysis by others). • Primary data collection (your own and original). • Quantitative data. • Qualitative data. • Mixed data and methods. – Philip, L.J. (1998) ‘Combining quantitative and qualitative approaches to social research in human geography – an impossible mixture?’ Environment and Planning A, 30, 261-276. Quantitative - Definition • A quantitative approach is one in which the investigator primarily uses post-positivist claims for developing knowledge (i.e. cause and effect thinking, reduction to specific variables and hypotheses and questions, use of measurement and observation, and the test of theories). (Creswell, 2003, p.19) Quantitative approaches • Attempts to explain phenomena by collecting and analysing numerical data • Tells you if there is a “difference” but not necessarily why • Data collected are always numerical and analysed using statistical methods – what are statistical methods? • Variables are controlled as much as possible • Randomisation 随机化to reduce subjective bias Gathering quantitative data • Data sources include – Surveys where there are a large number of respondents (esp where you have used a Likert scale) – Observations (counts of numbers and/or coding data into numbers) – Secondary data (government data; third party data etc.) • Analysis techniques include hypothesis testing, correlations and regression… Black swans and falsifiability可证伪性 • Falsifiability or refutability. 可反驳性 of a statement, hypothesis, or theory is the inherent与生俱来的,固有的 possibility that it can be proven false • Karl Popper and the black swan; https://www.flickr.com/photos/lselibrary/ deductive c.f. inductive reasoning IMAGELIBRARY/5 • Hypothesis testing • Start with null hypothesis i.e. H0 – that there will be no difference CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=1243220 Limitations of quantitative research? • Some things can’t be measured – or measured accurately • Doesn’t tell you why – does it need to? • Can be impersonal 没有人情味的– no engagement with human behaviours or individuals – mmmmm. • Data can be static – snapshots of a point in time – or not… • Can tell a version of the truth (or a lie?) – doesn’t all data? “Lies, damned lies and statistics” – persuasive power of numbers – difficult to argue against! Qualitative - Definition • A qualitative approach is one in which the inquirer often makes knowledge claims based primarily on constructivist perspectives 建构主义观点(i.e. the multiple meanings of individual experiences, meanings socially and historically constructed, with an intent of developing a theory or pattern) or advocacy/participatory perspectives (i.e. political, issue-oriented, collaborative or change oriented) or both. (Creswell, 2003, p.18) Qualitative approaches • Any research that doesn’t involve numerical data • Instead uses words, pictures, photos, videos, audio recordings. Field notes, generalities. Peoples’ own words. • Tends to start with a broad question rather than a specific hypothesis • Develop theory rather than start with one inductive rather than deductive • 归纳而不是演绎 Gathering qualitative data • Interviews (structured, semi-structured半结构化 的 or unstructured) • Focus groups • Questionnaires or surveys • Secondary data, including diaries, self-reporting, written accounts of past events/archive把..存档 data and company reports; • Direct observations – may also be recorded (video/audio) • Ethnography人种学 Gathering qualitative data • Tends to yield产生 rich data to explore how and why things happened • Don’t need large sample sizes (in comparison to quantitative research) • Some issues may arise, such as – Respondents providing inaccurate or false information – or saying what they think the researcher wants to hear – Ethical issues may be more problematic有疑问的 as the researcher is usually closer to participants – Researcher objectivity may be more difficult to achieve Limitations of qualitative research? • • • • It can be very subjective – if done badly It can’t always be repeated – or is that true? It can’t always be generalised – or is this true? It can’t always give you definite answers in the way that quantitative research can (seemingly 表明上看来) Perspectives on Secondary Data Collection Perspectives on Secondary Data. It would be wrong to think of social research solely in terms of the first-hand collection of data by means of, say, observation or asking questions. A great deal of information is already available, having been collected by others and often for other reasons. BUT Both the production of official statistics and their secondary analysis are not unproblematic enterprises. LIMITS : ACCESSIBILITY and PROVENANCE The 'Provenance' of secondary data – How was the data collected and for what purpose? This will enable an initial assessment as to the likely reliability of the data, and an initial assessment about its robustness稳 健性. • • • Questions to consider include: If the data were collected for a specific reason, what are the implications of the rationale of data collection for the coverage and reliability of the data set? What are the implications of the data collection methodology for the coverage /degree of detail available in the data set? What period does the data relate to? If the information is not current, but is being used as a proxy for the prevailing situation, is there any reason to expect that the there have been substantial changes in the period since data collection? Perspectives on Primary Data Collection Interviews: Structured, Semistructured and open • Particularly useful for gathering qualitative data • Face to face and telephone • Highly structured questions are good for eliciting诱发,刺 激 quantitative data – Can be difficult to gain ‘deep’ understanding of a specific issue • Open or unstructured interviews allow access to unpredictable information – Flexible - allows interviewer to probe answers – Can become overloaded with irrelevant information • Semi-structured interviews can be designed to cater for open and closed questions I’m Going to Generate My Own Data: Using Surveys… Survey design What information is needed? • Factual/attitudinal; complex/detailed; long/short; standard/measurement. Multiple interests… • Respondent – length / perceived value. • Interviewer – ease of use, instructions. • Researcher – hypothesis, research aims, ideas. • Analyst – coding and processing. What resources are available? • Skills, competencies. • Equipment, funding. On-line … • Eg Survey Monkey Some Basics… • Questions should be designed so that it is easy for respondents to understand them and to answer accurately and clearly. • Questionnaires should be easy for interviewers to administer. • Questionnaires should be constructed so that recorded answers can be easily edited, coded and transferred onto computer files for analysis. • Questionnaires should encourage and keep respondents’ interest. Too many or too complex questions can either generate bad responses or result in the ‘waste bin’ response! Questionnaire Design • • • • • • • Introduction and purpose – why fill it in? Sequencing and filtering – ease of use? Question types – open, closed? Question wording and construction. Instructions – clear or confusing? Appearance and layout. Consent and confidentiality. Question wording…avoid: • • • • • • • • Ambiguous questions Imprecise questions Presuming questions Questions that rely on memory Double questions Leading questions Hypothetical questions Overly sensitive questions OBSERVATION as survey research. Observation is planned and conducted in a systematic way, rather than happening spontaneously and haphazardly as it usually does in every day life. Appropriate techniques are carefully selected for the purposes at hand. Observations are systematically recorded rather than stored only in a personal memory, and are carefully interpreted and analysed, again employing systematic and planned procedures. Moreover, the data produced by observational research are subjected to checks on validity so that we can be more confident about their accuracy than is usually the case with observational data produced routinely in everyday life. 10 MINUTE BREAK Sampling…. Sampling • Units of analysis. • Selection and sampling of cases or sources of data in qualitative and quantitative research. • Random and non-random sampling, representative samples and statistical inference. • Sampling error and bias. • What to sample, how many, how do I choose ? What to sample? • Sample units – people; households; sites; businesses; planning applications; local governments; professionals; buildings; neighbourhoods; town villages, cities; etc. • Do you know the ‘ideal’ population – A finite, total known population. • Can you create or have access to a sampling frame – The collection of units from which sample will be drawn. Boundaries, definitions, period, accuracy of frame – inbuilt bias, timing of selection. • Sample to frame (statistical test) / frame to population (non-mathematical justification) Sample size and statistical relevance… • As a general rule, the larger the sample size, the more precisely the sample statistics will reflect the population parameters. • The standard error is inversely proportional to the square root of the sample size ie to double the accuracy means the sample size will need to be 4 times bigger; to treble it, then the size needs to be 9 times bigger etc. • You need big enough samples and ‘cell sizes’ in statistical analyses eg chi-square method – filtering, randomness, resources. Probabilistic Sampling methods… • Simple random sampling: each case of the population has an equal chance of selection – sample errors and over / under representation. • Systematic/interval sampling: an interval is chosen randomly, and every nth case selected – periodicity. Probabilistic Sampling methods… • Stratified sampling分层抽样: population divided into subsets according to some characteristic which you know (or suspect) is significant to the study. Separate random/systematic samples drawn from each subset • What do you know, prior to your research, about your ‘population’ and which of its characteristics are important to your study or research question, hypothesis/theories? RANDOM SYSTEMATIC STRATIFIED 1 1 1 2 2 2 3 3 3 4 4 4 5 5 5 6 6 6 7 7 7 8 8 8 9 9 9 10 10 10 11 11 11 12 12 12 13 13 13 14 14 14 15 15 15 16 16 16 17 17 17 18 18 18 19 19 19 20 20 20 21 21 21 22 22 22 23 23 23 24 24 24 Non-probabilistic Sampling… • In many cases of our research, random sampling is not possible or appropriate. • In these situations non-probabilistic methods of sampling can be used, and justified. • Typically used in qualitative research, but consequently no statistical relevance and unable to reliably say much about representativeness. Non-probabilistic Sampling… • Snowball or chain sampling - technique where existing study subjects recruit future subjects from among their acquaintances. • Quota sampling – researchers identify characteristics of the sample (say age and gender) to be selected in advance, then select desired number of respondents who fit the characteristics (say 25 old female; 25 young female; 25 old male; 25 young male). • Volunteer sampling – participates in the study volunteer to do so, in response say to an advertisement or general request to play a role in research – self-selecting. • Judgement or purposive sampling – researcher selects a ‘typical’ or ‘interesting’ case or cases for study – might be based on theoretical or practical justification. Case Study Approaches Case study research • A research design where the case(s) exhibit the operation of some identified general theoretical principle – theory led. • Case studies use multiple methods of data collection or generation to examine a phenomena in detail – a choice of methods. • Simply ‘cases’ referring to situations/ behaviours/ events of interest to your topic, research question or area of study. • It HAS to be more than the setting(s) of your research – “my research was done in Sheffield”. Main ‘case based’ research study designs… Single-case designs Multiple case designs Holistic (single e.g. Sheffield city e.g. Nottingham, unit of analysis) centre. Derby and Leicester city centres. Embedded e.g specific office e.g. specific (multiple units developments in office of analysis) Sheffield city developments in centre. the above. Why use case based studies ? A Single case study: • Critical case, testing well-formulated theory. • Unique/extreme case of a rare or specific phenomenon. • Revelatory of previously inaccessible phenomenon although commonly present. • In conjunction with other research approaches. Why use case based studies ? Multiple case based studies: • Evidence is often more compelling. • Study offers a more robust analysis. • Comparative studies. • Replication of similarity or contrary results rather than sampling • But extensive resources and time. Comparative Studies “An aim of comparative research is to understand and explain the ways in which different societies and cultures experience and act upon social, economic and political changes, plus how these views relate to more general changes and thus shared experiences and actions in the face of similar concerns and pressures.” May T. (2001) Social research, 3rd ed. Milton Keynes, OUP. SEE: Healey P. & Upton R. (eds) (2010) Crossing borders: international exchange and planning practices. Routledge: London. Sanyal B. (ed) (2005) Comparative planning cultures. Routledge: London. Some Reading… Bryman A. (1988) Quantity and quality in social science. London, Routledge. Black T. (1999) Doing quantitative research in the social sciences. London, Sage. De Vaus D. (1991) Surveys in social research. London, UCL Press. Fink A. (1995) How to design surveys. London, Sage. Fink A. (1995) How to ask survey questions. London, Sage. Fink A. (1995) How to sample surveys. London, Sage. Fink A. (1995) How to analyse survey data. London, Sage. Gillham B. (2000) Case study research methods. London, Contiuum. Gorard S. (2003) Quantitative methods in social science. London, Continuuum. Gray D. (2004) Doing research in the real world. London, Sage. Hakim C. (2000) Research design : successful designs for social and economic research. 2nd ed. London,Sage. Hedrick T. et al (1993) Applied research design : a practicalguide. CA, Sage. Kumar R. (1996) Research methodology. Sage, London. Litwin M. (1995) How to measure survey reliability and validity. London, Sage. May T. (1997) Social research. Milton Keynes, OUP. Moser C. & Kalton G. (1985-) Survey methods in social investigation. London, Heinemann. Oppenheim A. (1996) Questionnaire design and attitude measurement. London, Heinemann. Punch K. (1998) Introduction to social research. London, Sage. Sapsford R. & Jupp V. (1996) Data collection and analysis. London , Sage. Yates S. (2004) Doing social science research. Open University/Sage. Yin R. (1984 -) Case study research design. London, Sage. Further links • Some content borrowed from SkillsYouNeed website (http://www.skillsyouneed.com/learn/research-methods.html) Other useful links: • Introduction to Quantitative and Qualitative Research Models (William Bardebes). PDF at http://tinyurl.com/qq-models • Methods Map: http://www.methodsmap.org • Ready To Research: http://readytoresearch.ac.uk • Methods@Manchester: http://www.methods.manchester.ac.uk/resources/categories • Research Data Management training: http://datalib.edina.ac.uk/mantra/