Research

Research

1. Sampling

   • What and why: sampling is closely linked to research design/method as well as its aim - also depends on whether the rs wants to transfer or generalise (Matthew and Liz)...sampling is the selection of some cases from a larger group of cases (Matthews and Ross) - larger group is called the population - the sample is therefore a version of the pop made for the purpose of the research as it is rarely possible to rs a whole target pop - by using a smaller more representative sample from the pop it allows to generalised to the findings to the whole populatio.
   • Common problems: small samples can lack representativeness and could include bias.. therefore aim for bigger more representative samples where findings can be generalised to rest of pop.
     • Sampling error: difference between the characteristics of a smaple and characteristics of the pop from which it was selected = the bigger the sampling error the less representative and generalisable the findings are.
     • Systematic sampling error: over-rep or under-rep of some pop characteristics in a sample due to the method used to select the sample - bias towards certain characteristics.
   • Probability sample: sample that is highly representative of the whole population - allows the rs to know in advance how likely it is that any elements of the pop will be selected... small samples however still hold possibility for bias or sampling error - pop should not be too large due to law of diminished returns - about absolute size not relative size.. best population is random selection with a fairy big sample (low chance of bias and generalisable.. closely associated with quantitative data.
     • Types: simple/random, systematic random, stratified
     • Random/simple: basic probability sample with the essential characteristic being that every element of the population has a known and independent chance of being selected :) representative :) generalisable :( accidental bias
     • Stratified random sample: this sampling methods uses information already known about the pop prior to sampling to make the sampling process more efficient - all elements of the pop are distinguished according to their value on some relevant characteristics - elements are then sampled randomly from within these strata :) representative :) generalisable :( accidental bias :) is used to enable the rs to minimise teh time spent and resources required (Matthews) :( but many rs do not have time and resources to design a stratified sample and then gain access to this sample.
     • Systematic random sample: variant of simple random sample where the first sample is selected randomly from a list and then a pattern is used to selected the others e.g. every nth term :) representative :) generalisable :( accidental bias
     • Cluster: a sample consisting of cases selected because of their proximity to 1 or another.. requires less prior knowledge - used when there are naturally occurring mixed aggregate of elements within the pop (with each element only appearing in one element) and then a random sample is taken from each cluster
   • Non-probability: does not allow the rs to know in advance how likely it is that any element of a pop will be selected from the sample - good if: there is no sampling frame or cluster, when exploring new areas instead of theory testing, if prioritising depth of understanding above capacity to generalise - no random selection.. size is not important and there it is usaly limited by saturation/law of diminished returns - Chamberlain: data colection stops once a rs is repeatedly collecting data which supports their conceptual framework without committing the large amount of time and resources associated with quant rs
     • Types: availability, snowball, purposive and theoretical
     • Availability - otherwise known as convenience is based on the samples availability to the s - also known as haphazardous sampling :( bias :( not representative :( not generalisable
     • Purposive (Matthews): the sample is selected from cases that will best enable the rs to explore the rs questions in depth... usually used in in-depth sample studied with rs designs based on qual info which doesn't need to be statistically representative/generalisable :( time consuming :( costs a lot :( diminished returns :( not concerned with no-response as they are seeking to fufill teh selection criteria rather than ensuring that their sample is based on probability :( not repesentative
     • Theoretical (Matthews): a sample is selected from cases that will best enable the rs to explore theoretical ideas - used for in-depth small studies with qual rs designs that don't need to be representative - initial case is usually selected on a relatively unstructured basis (as the theory begins to emerge), then further cases are selected to explore and test the emerging theory - until saturation.
     • Quota: technique selects a certain number or quota of cases on the basis of them matching a number of criteria needed for the rs - involves designing the pop into proportions of the groups that the rs wants to represented in their sample - sample represents certain characteristics that are in proportion to their prevalence in the pop :( difficult to choose criteria that is usable and relates to the question :( bias towards certain groupds :( not representative of other characteristics :( not generalisable to other pops
     • Snowball: member of first initial sample are asked to identify others with similar characteristics as them to join in :( initial sample may bias whole sample :( good for hard to reach groups :( not rep :( not generalisable
   • bias: prejudice in favour or against a group/individual/perspective.. it can come from accessibility to sample
   • Non-response - ppt decline to ppt
   • Generalisability: conclusion based on a sample, or subset of a larger pop that holds true for that pop - refers to the ability to use the findings to explain the behaviour of other pop - depends on sample quality - which is determined by amount of sampling error and size
     • Cross population generalisability: exists where findings about one group/pop of a setting holds true for other groups/pop or settings - determined by quality of smaple

2. Primaary and secondary data

   • Primary: information that is collected by teh rs themselves for their own rs purposes - includes questinnaires, surveys, experiements, observations and interviews :) rs is able to gather the exact data they want to collect :( costsly, time consuming and quantity of data is often samll and quality is often poor.
   • Secondary: information that has already been collected or created by someone else for heir own rs purposes but for which the rs uses for their rs - includes questionnaires, documents, surveys, diaries, official statistics etc :( rs may not be able to get exactly the info needed.. may be limited in space and time.. could cost.. access is difficult.. limited sources.. bias or unreliable rs.. hard to compare :) quick and easy.. provides quan and qual data that no rs could match... lots of data over different time periods and places.

3. Questionnaires: use of questionnaires grew is the 20th century - they are a list of questions each with a range of answers and a formate that enables standardised and relatively structured data to be gathered about each of teh large number of cases... can be done on the phone, by emails, for posted for self-completion.. are often used to gather data within a cross-sectional or longitudinal rs design where the pop of of substantial size - collect both qual and quan data - depends on research aim and purpose - e.g.'s include 10 yearly census, victim surveys and CSEW

   • Pews research centre: the choice of words and phrases in a questionnaire are critical in expressing the meaning and intent of the question to the respondent and ensuring that all respondents interpret the question in the same way - as even small wording differences or understanding or phrases can substantially effect the answers people provide and research findings... for example leading questions - these are questions that suggest the answer that the rs is looking for from respondents - questions should therefore be as value neutral as possible - wording should be clear and simple to avoid confusion.... lack of answer options when using closed questions could also cause problems - lead to unreliable answers.
   • strengths: data can be gathered in a standardised format - do large scale surveys collecting generalisable findings - do cross sectional comparisons - most effective way of gathering data from a range of people - can be confidential if done correctly - ppt can back out as any moment (ethical) - quick to complete - easily entered into a computer and analysed - can collect and analyse diferent types of data e.g. facts, descriptions, opinions, knowldge etc - can ask different types of questions to gain a variety of responses (closed and opened) - cheap and quick - easy to repeat in other places and different times.
   • Types of questions: open - more detailed and complicated answers that are difficult to analyse - time consuming meaning low response rate. Close questions = quick but answer reliability is low (lie, forget and social desirability) used in large scale surveys
   • Limitations: limited access to qual data - not good for sensitive issues (ethics) - exclude certain groups (homeless, BME, prisoners, children, elderly).. often suffer from low responses rate.. questionable validity.. problems with social desirability or acquiescene bias and bias... not good for sensitive topics
   • Open and closed questions

4. Interviews

   • Unstructured: qualitative research methods - a qual way of understanding teh complex behaviour and opinion of individuals whilst avoiding any pre-determined questions or categories that may limit field of enquiry (Punch)... provide flexibility as lots f topics can be discussed in the order both ine ppt and ppt feel is appropriate... rs will often have a short list of topics to discuss at the start of the interview but the interview flows freely... Douglas - for unstructured interviews there must be flexibility and creativity...
     • Stengths: detailed in depth data with expansion and digression on certain topics.. authentic data as it is not influenced by researcher opinions or pre-set questions.. flexible and versatile allowing sensitive topics to be discussed ethically and appropriately ... offer dynamic effects of interactions and expressed opinion... non-standardised so can be used to examine non-motivations.. can promote and probe for answers.. ask indirect questions which may make the ppt more willing and prepared to reveal negative feelings or answers.. good for focus groups.
     • Limitations: bias info... miss info.. lie.. social desirability.. acquiescene bias... over-politeness.. silence.. lack of awareness or understanding.. fall to follow topic.. time consuming and costly.. lack of control over interview
   • Semi-structured: qual and quan data - qual methods that combine a pre-determined set of open questions with the opportunity for the rs to explore particular themes or responses further... rs can add or remove questions or change order to make the interview to how they see most appropriate at the time (after having prepared questions)... most common interview used... there is an interview schedule but it can either be followed or deviated from.. more like a convo so can build trust and rapport.. invites people to open up and expend answers.. e.g. mays policing cannabis as a class a drug.
     • Types of questions (Kvale): intro (tell me about or have you ever)... follow up (what do you mean by that).. probing (could you tell me about)... specifying (what did you do then, how did you react)... direct questions (do you feel bad after beating someone up)... structuring topics (i'd like to move on to)... silence...interpreting questions (do you mean that).
       • Limitations: interviewer effect - can shape the answers.. Hyman found that white interviews received different responses that black interviews when interviewing different races - this is also thr same for male and female interviewers... not complete control.. where do all the questions come from - does this make it harder to compare results later on... social desirability effect.. acquiescene bias... can our thoughts really be portrayed fully by language in an interview situation.. interview situtaion = false... confused by collective facts, subjective experiences or cultural narratives (Silverman).
     • strengths: authentic data - flexible allowing rs to collect detailed data that is not influenced by rs.. not limiting respondents to a set of pre-determined answers.. rs can retain some control over the interview making it easier for the rs to compare results later... allow past events to be reconstructed (flexibility).. possibile longitudinal rs... ppt speak for themselves.. data based on human understanding of social reality.
   • Structured: quan - pre-determined to produce data that is easy to quantify.. clear structure.. often closed ? or multiple choice questions... narrow quan data - focus on frequency, relationship, cause and effect.. dtaa that can be easily analysed... e.g. CSEW
     • Limitations - no flexibility - not good for sensitive topics - no rapport built - awkward environment... social desirability bias.. lack of info
     • Strengths: easy analysed, quick to answer, cost effective, not as time consuming, more control over interview, stick to research purpose.
   • Other types of interviews: telephone, email, focus groups etc
     • Focus groups: allow rs to investigate multiple people at once - produce rich data and time/cost effective.. some may not speak as much however.. false consensus may develop around dominant group members.. may be harder for rs to control schedule.. dont know work dynamic... good for hard to reach groups however.

5. P.E.R.V.R.T

   • Practical issues: time consuming, costly, easily achievable, lack of response, training needed, resources needed, is it sensitive topic or group, gaining access
   • Ethical: informed consent, confidentiality, harm and privacy, deceit and legal issues
   • Reliability: consistency, replicability
   • Validity: true reflection, is it measuring what is says it is, is it detailed
   • Representativeness: typical, large scale and high repsonse
   • Theoretical issues: positivist and interpretivist.

6. Ethnography

   • Featurs: identifies problem, concpets or issue.. samples are often small and non-representative.. obs and record obs in field notes.. build a theory from obs.. primary data collection through obs but data often qual (think descriptive).. sometimes an interview follows.. deception... rs should not impose their beliefs, views or attitudes onto rs subjects
   • Origins: study of groups of people in their natural settings.. comes partly from the work of anthropologists such as Malinowski and urban reportage by victorian social reformes... Ethnography is going out into the environment that the rs aims to study and taking part is covert or overt rs aim in which he becomes one of the group.. rs emerge themselves in a social setting or social groups in order to understand that work from their perspective
   • Types: total obs = completely detatched from individuals being studied (used in courts or police).. obs ppt = not completely detached or completely involved (slightly included).. total ppt = completely involved with individuals being studied... Overt = easier to make notes, rs subjects can choose to conceal some things, rs avoids difficult ethic and legal choices, is teh behaviour being observed actually natural - social desirability...... Covert = no obs effect, hard to make notes, hard to ask questions, rs may end up doing things they shouldnt - going native and ethical problems
   • :( limitations: access is hard to get (gatekeeper).. time consuming.. tiring.. have to keep up act.. dangerous for rs (going naive and hurt).. need lots of background info.. difficult to gain trust.. could influence rs findings... deception.. confidentiality.. no informed consent... not representative.. not generalisable.. not reliable.. hard to record..
   • e.g - Hadfield Bar wars rs - covert rs in pubs and clubs but changed to overt when joining police patrols.. E.g. Lister and bouncers - covert rs

7. Ethics

   • Ethics are moral principles/a system of moral principles (OED 2014)... The moral principles guiding rs from its inception through to completion and publications (Finch and Fafiski) - relevant to all stages of rs projects.. constrain the ethic of ultimate ends (May)
   • Key ethical concerns - BSC code of ethics: 1) Harm - rs should ensure the physical, social and psychological well being of an individual ppt in rs and not adversely affected by ppt in rs.. 2) informed consent - rs should base rs on the freely given informed consent of those studied in all but exceptional circumstances.. 3) privacy - rs should not breach the duty of confidentiality and not pass of identifiable date to 3rd parties without ppt consent... e.g. of bad ethics = milgrams experiement
   • Strengths of following ethics: protect ppt, reliable data.. authentic data.. rs has integrity.. valid results.. rs can be compared and shared... consistent... helps navigate tricky issues.. codes of ethics provides useful guide and strengthens rs whilst protecting ppt.
   • Limitations of not following: rs with sensitive topics of people may cause harm... duty to stops harm.. is there is duty of public disclosure or confidentiality... what if they can only carry out teh rs without informed consent of deceit.
   • Situationalism: says ethics depend on the situation the rs is in.. no universal moral code/ethics meaning each case deserves a unique solution.. someone who practices situational ethics would approach ethical problems with general moral principles rather than a rigorous set of ethical laws - prepared to give up some principles but context is key - often a trade of betwen ethical principles and situational requirements of rs... Winlow used situational ethics in his rs... Universalism = opposite - strict moral principles that all rs must adhere to.