WJEC A2 Psychology PY3 - Research Methods AS Recap

Experiments: in all experiments, researcher is trying to discover whether the IV (independent variable) causes an effect on the DV (dependent variable)

IV = manipulated either directly or indirectly in order to see if it affects the DV

DV = dependent variable is response to manipulation of the DV that is measured

Confounding variable = any variable other than IV that has affected DV which the researcher did not control

LABORATORY EXPERIMENTS

• conducted in a highly controlled environment (lab), variables carefully controlled
• IV is manipulated directly by the researcher and the effect on the DV measured
• always 2 conditions and pps are allocated randomly from the original sample

Advantages:

• variables highly controlled so less chance of confounding variables affecting DV = higher internal validity and cause-effect relationship established between IV and DV

• variables operationalised in study so easier to replicate, if study repeated and similar results found, increases validity of results = suggests original findings = true

Operationalising variables = clearly defining variable so easily tested or measured

Disadvantages:

• artificial environment, pps may not behave in same way as they do in real life, lacks mundane realism
• pps know they are in study = affected by demand characteristics, may try and guess aim of study and behave accordingly

Mundane Realism: for a study to have this, tasks must resemble types of experiences we have in everyday life

Demand Characteristics: occur when pps try to guess aim of study and alter behaviour accordingly, to support what they believe to be aim or refute it = reduces internal validity

FIELD EXPERIMENTS

• conducted in natural environment where more difficult to control variables
• IV directly manipulated by researcher and effect on DV measured

Advantages:

• less artificial as in natural environment, pps do tasks typical in everyday behaviour = higher mundane realism which increases validity
• pps unaware of being in study, reduces chances of demand characteristics

Disadvantages:

• less control over variables = greater chance results affected by confounding variables
• variables in study harder to replicate and so difficult to test validity

NATURAL EXPERIMENTS

• conducted in natural environment where researcher studies naturally occurring event
• IV not manipulated directly by the researcher, occurs naturally eg gender diff/head injury

Advantages:

• allows researcher to study 'real' problems = high mundane realism = high external validity

Disadvantages:

• little control of variables = greater chance confounding variables affect results = reduces internal validity
• pps aware in study = demand characteristics?

INDEPENDENT MEASURES/GROUPS DESIGN - when separate groups of pps undergo separate conditions

Advantages:

• fewer demand characteristics as pps less likely to guess the aim and change their behaviour (Help you/Screw you effect) as they only experience 1 condition
• more population validity as more pps used so more representative
• no order effects (eg practice, fatigue, boredom) as pps only do task once

Disadvantages:

• effects of pps variables (individual differences) harder to control (eg age, IQ, gender)
• bigger sample, twice as many people required, takes longer

REPEATED MEASURES - when same group of pps undergo each condition

Advantages:

• no pps variables as same group of pps for each condition
• smaller sample, quicker to obtain sample

Disadvantages:

• more demand characteristics as pps more likely to guess aim and change behaviour if they have to do it more than once
• less population validity = smaller sample, less representative
• more order effects (eg practice, fatigue, boredom) as doing study twice

How to overcome order effects: use counter-balancing technique = ABBA method, when half of pps do condition in 1 order (A + B) and the other half do conditions in another order (B+A) to avoid order effects affecting results

MATCHED PAIRS - when separate groups of pps undergo each condition but members of 1 group are similar in some way to pps in other group eg age, IQ, gender

Advantages:

• same task with different pps increases population validity and reduces risk of order effects
• fewer pps variables to influence results as similar

Disadvantages:

• takes a long time to match pps on characteristics like IQ
• will never have a perfect match, impossible to remove pps variables completely

CORRELATIONS

• link or association between two variables
• when using correlation analysis, measure taken of two variables concerned, 2 scores produced for each ps
• CV1 + CV2 = a covariable is a variable in a correlation which changes systematically with another covariable

Types of Correlation: (perfect, mild/modest, weak)

POSITIVE: where high values on 1 variable associated with high values on other variable (or low and low) - as one value increases so does the other

NEGATIVE: where high values on 1 variable associated with low values on other variable - as one value increases, the other decreases

ZERO: no relationship or association between the 2 variables

Advantages:

• provides precise, quantitative measure of strength of relationship between variables useful when trying to unravel complex relationships as allow for measurement of many variables and relationships at the same time

Disadvantages:

• impossible to establish cause-effect relationship between variables through correlational analysis - only measure degree of relationship between 2 variables
• even if relationship found, researchers cannot be confident other intervening variables (CV) are not responsible for relationship found
• correlations cannot measure non-linear relationships between 2 covariables

Correlation Co-efficient: identifies the strength of a relationship between 2 variables; it is between -1 and +1, 0 = zero correlation, -1 = strong negative correlation and +1 = strong positive correlation

OBSERVATIONS

• research method where researcher watches and records behaviour shown by individual/group of individuals

Types of Observations:

• Naturalistic: takes place in a natural setting
• Controlled: often takes place in lab where certain variables controlled
• Participant: researcher becomes part of group being studied

(+) researcher gains insight into emotions/motivations felt by pps, producing more detailed/meaningful data

(-) observer part of social experience with pps, becomes too involved with other pps, can lead to researcher bias/subjective interpretations of data gathered rather than objectivity

• Non-participant: researchers do not become involved with those being studied (being a ps) but just watch and record behaviour (often from distance) to avoid having an influence on the pps

(+) any observations/records of behaviour can be made more objectively as researcher will not have any influence on the pps behaviour or become emotionally involved which could make records more subjective 

(-) however difficult to eliminate all subjectivity and bias as the observer may only select behaviour they deem significant based on their own expectations or predictions for the observation; may interpret behaviour based on personal opinions/beliefs = subjective

• Covert: pps not aware they're being studied by a researcher

(+) demand characteristics reduced as unaware of being studied so don't change behaviour = increased validity = behaviour observed is how ps would naturally behave in that setting 

(-) difficult to be 100% confident that IV causes DV as cannot directly control behaviour of pps without becoming an overt observation

• Overt: pps aware they're being studied by a researcher

(-) observer effects may affect results, the presence of an observer may cause the pps to act differently because they know they're being observed = demand characteristics

• Unstructured: researcher records all relevant behaviour but has no specific system for doing so
• Structured: systems (eg coding systems) are used to organise observations

Sampling procedures: continuous observations often not possible so researcher decides on one of the following method:

• event sampling - counting no. of times certain behaviour occurs in target individual(s)
• time sampling - recording behaviours at regular intervals eg every 30 seconds

Advantages:

• more valid data because person's actual behaviour recorded rather than ps self-report
• naturalistic observations = high ecological validity, behaviours observed naturally

Disadvantages:

• Observer bias = researcher may only record behaviours they're expecting to see
• difficult to control CV = may account for behaviour being observed

CONTENT ANALYSIS

• "indirect observation" where artifacts people have produced are observed and recorded (rather than their behaviour) eg soap operas, adverts (LH, newspapers) diaries = looking for themes, patterns or trends in the artifacts
• DATA = Quantitative: counting no. of times a theme appears in the artifact; Qualitative: describing (with egs) a theme/trend that appears in an artifact

Coding in Content Analysis: themes/patterns being observed need to be split up into different behavioural categories, which are then operationalised - broken down into a specific set of components to provide researchers with a clear coding system for recording trends they observe

Advantages: high ecological validity - based on observations of what people do, focuses on real communications + easily replicated by others

Disadvantages: observer bias = researchers may have different interpretations of behavioural categories = findings not objective or valid, also cultural bias = researcher's culture may influence the behavioural categories used

QUESTIONNAIRES

• comprises of a set of qs designed by the researcher to collect info about topic
• can include open qs - give qualitative data (rich in detail but more difficult to analyse)
• can include closed qs - give quantitative data (easier to analyse but answers may not reflect real thoughts/behaviour as only choose from fixed set of responses)
• Likert Scale: given more than 2 options for a response, measures strength of feeling (strongly agree, agree, agree nor disagree, disagree, strongly disagree

Advantages:

• collect same info from large no. of people easily
• gives researcher opportunity to find out what people think and feel
• pps more likely to reveal more confidential/personal info that they would in an interview

Disadvantages:

• pps may not answer qs truthfully due to social desirability bias, reduces validity of data
• sample may be unrepresentative = only certain type of ppl fill in qs - low pop. validity

INTERVIEWS

• interviews can be structured or unstructured
• structured: researcher has pre-determined set of qs which pps are asked face to face
• unstructured: researcher may have a couple of initial qs but no pre-determined set of qs - allows new questions to be developed as interview progresses which can be tailored to pps responses

Advantages of Structured Interviews:

• standardised qs easily repeated
• easier to analyse as answers are more predictable

Disadvantages of Structured Interviews:

• interviewer bias (expectations) may influence answers
• different interviewers = low reliability if behave differently when asking qs

Advantages of Unstructured Interviews:

• gain more detailed info from pps than in structured interviews
• lead to new insights not gained from use of structured qs

Disadvantages of Unstructured Interviews:

• interviewer bias greater issue than in structured interviews - researcher will follow up leads they want to - less objectivity
• requires well trained interviewers with specialist knowledge who can generate appropriate qs throughout interview

CASE STUDIES

• detailed study of a single individual, institution or event, over a long period of time
• info gathered from range of sources, including individual concerned or family/friends
• gathered using range of methods including interviews, observations and questionnaires

Advantages:

• provides researcher with rich, in depth data, less likely important info will be overlooked
• provides researcher with opportunity to investigate instances of human behaviour and rare experiences (eg Genie - able to study her in depth over long period of time and see how being locked away had effects on her development)

Disadvantages:

• case study involves analysis of one person's behaviour, difficult to generalise findings as other people may not behave in similar way
• long period of time - researcher becomes very involved, and loses objectivity = may become biased in how they record details = overlook key aspects of findings

RELIABILITY - how consistent something is

Internal Reliability - what happens inside an experiment and whether tools used to measure something will always produce the same result

• to ensure internal reliability = test-re-test = ask same pps to complete same test and if same/similar result, test is reliable

Issues of Internal Reliability:

• 1. Inter-Rater Reliability: if more one observer/researcher is used, do they all record results in the same way?
• DEALT WITH = clear operationalised variables to be measured, a clear coding/coding system, or a pilot study to identify any researchers who are inconsistent with the majority that record measures in the same way = followed by more training/removal
• 2. Measuring Tool Reliability: whether the tools are consistently measuring the same thing
• DEALT WITH: use split half method = divide pps answers to questions in half, if individual scores on both halves are similar (highly correlated), high reliability

• 3. Procedural Reliability: whether each ps experiences the same thing
• DEALT WITH = standardising the procedures, the instructions and the measuring tools used for all pps

External Reliability - whether the study would produce similar results if it were replicated or if similar study carried out, would researcher gain similar findings?

• to ensure external reliability = replicate the study on another occasion with similar tasks and pps = if similar results, findings/study said to be reliable

Issues of External Reliability:

• 1. Ways of ensuring Reliability: whether the study measures the same thing over time and therefore if a pps response is consistent over several different occasions
• DEALT WITH = do a test re-test method = give same pps the same task later, 2 sets of scores should be highly correlated if the test is reliable

VALIDITY - how true/accurate the study is

Internal Validity - what happens inside the experiment and to what extent the experiment is measuring what it is supposed to be measuring

Issues of Internal Validity:

• 1. Extraneous Variables: change in DV may not be due to manipulation of the IV but caused by confounding variables (pps or situational variable), situational variable = temperature/background noise eg, pps variable = individual differences like age, gender, IQ
• DEALT WITH = if possible, all pps complete all different conditions (to eliminate PV) and then any confounding variables will be situational when then need to be controlled
• 2. Measuring Tool: may be inaccurate, the tools in expt may not measure what they are meant to be measuring
• DEALT WITH = use content validity = ask an independent expert to evaluate measure and assess validity, OR use concurrent validity = compare tool with another tool that has been validated, if results are similar, that tool can be said to have concurrent validity

External Validity - whether the study is a true representation of behaviour in all settings and if it can be applied to other situations outside the lab and people in target pop

Issues of External Validity:

• 1. Population Validity: whether the sample of pps who take part in the study truly represent all people in the target pop, if they don't results cannot be generalised, often arises from sampling methods used that aren't representative enough of target pop
• DEALT WITH = increase the representativeness of sample by increasing the number of pps, broadening the sample by using stratified sample and broadening the target pop
• 2. Ecological Validity: findings from study may only apply in that environment (esp in lab - artificial) the behaviour in lab may not mirror behaviour in everyday life so results can't be generalised, ALSO mundane realism = whether the task within study is reflective of everyday tasks, if not findings cannot be applied to everyday tasks
• DEALT WITH = conduct the expt in a more natural setting and covertly, ALSO ensure tasks mirror those in everyday life

Crazy - Consent (informed)

Dogs - Deception

Drink - Debriefing

Ice - Invasion of Privacy

Cold - Confidentiality

Water - (Right to) withdraw

Puddles - Protection of Participants (Psychological/Physical)

LACK OF INFORMED CONSENT - pps may have agreed to take part in study without knowing the full aims of investigation and so haven't given their fully informed consent to take part (often occurs to avoid demand characteristics to ensure behaviour is as natural as possible)

How to control: researchers must tell pps exactly what they will be required to do in the study so they can make an informed decision about whether they want to participate

DECEPTION - pps sometimes deceived as to what the true nature of the research is; researchers may withhold info about aims or deliberately mislead pps to think study is about something else (links to lack of informed consent - deception causes lack of informed consent)

How to control: study should only be carried out if patient will not suffer from psych harm as a result of taking part without knowing the aims, if use of deception is justified = pps should be debriefed at end of study  = told full aims of research and have opportunity to withdraw data from study

LACK OF PROTECTION FROM PSYCHOLOGICAL OR PHYSICAL HARM - researchers fail to protect pps from harm during research - ask them to complete tasks which cause anxiety, embarrassment (psych) or cause physical harm like smoking

How to control: researchers must ensure they avoid any situation that could cause psych or physical harm and should make sure pps in same psych/physical state after study as they were before

LACK OF THE RIGHT TO WITHDRAW - researchers fail to tell pps they have right to leave study at any time, and may perhaps fail to inform pps they have the right to withdraw their data

How to control: researchers should ensure pps told before study they can leave at any time and request to have data withdrawn

LACK OF CONFIDENTIALITY - researcher may fail to protect confidentiality of pps by publishing findings where pps can be identified; even if names withheld, doesn't prevent anonymity

How to control: researchers should not record names of any pps, use false names or numbers and ensure it is impossible to identify pps if results are published

LACK OF PRIVACY - researcher may invade privacy of pps without their awareness, eg a ps being observed in their own home

How to control: researcher should ensure they do not observe anyone without gaining informed consent first unless in a public place where they may be expected to observe others without pps being aware = in this case, researcher should ask for retrospective consent to use data gathered

Target Population: group of individuals the researcher is interested in

Sample: group of pps from the target population who take part in the study; the sample should be representative of the target population

Sampling Methods:

Opportunity

Self Selected/Volunteer

Systematic

Random

Quota

Stratified

OPPORTUNITY - where you select those who are available to use eg in the street

(+) quick, convenient = easy to find pps as they are readily available to select

(-) doesn't produce a representative sample, biased because research is only being drawn from people in a small area at a particular place and time

SELF-SELECTED/VOLUNTEER - researcher advises ppl to take part, so pps select themselves by replying to advert and putting themselves forward

(+) pps willing and likely to take it seriously

(-) volunteer bias - certain type of people reply to these adverts so may be a 'type' of person, not atypical and not representative of target population

SYSTEMATIC - researcher uses pre-determined system for selecting pps, involves getting an entire list of target pop and selecting every nth person

(+) potentially unbiased as researchers using objective system = more representative

(-) could involve some unintentional bias - depends on how list is drawn up, and who is selected may still not be representative

RANDOM - every member of target population has an equal chance of being selected, need a list of everyone in target population and use random/unbiased method of choosing sample eg names out of a hat

(+) largely unbiased as all members of target pop equal chance of being chosen

(-) time consuming, still end up with biased sample that is not wholly representative

QUOTA/STRATIFIED - various sub groups in target pop identified, pps obtained from each group in proportion to their occurrence in the target population, if stratified = selection from sub groups is random, if quota = selection done through opportunity sampling

(+) more representative than other methods, less biased as proportionate to target pop

(-) very time consuming/complex + with quota, still some bias as selection of pps in sub group = opportunity, may not be completely representative

Measures of Central Tendency - summarise how close together data is

MEAN - taking all scores from set of data, adding it up and dividing by the number of values there

(+) most sensitive measure as uses all scores in the data set

(-) affected by extreme scores/outliers so may not be representative

MEDIAN - middle value in set of scores when presented in rank order

(+) not affected by extreme scores so can be more representative than mean

(-) not as sensitive as doesn't take all scores into account

MODE - most frequently occurring score in a data set

(+) useful measure for describing data that is in categories

(-) may not be very useful when there is more than one mode in data set (multi-modal)

Measures of Dispersion - measures how spread out the data is

RANGE - measure of dispersion which shows the spread of data over any given data set (difference between highest and lowest score)

(+) very easy to calculate, quickly gives researchers idea to how wide data is spread

(-) may not be representative value as data can be affected by extreme scores

STANDARD DEVIATION - mean distance each score in a data set is from the mean score from all the data; the bigger the value of 1 standard deviation, the bigger spread of scores

(+) sensitive measure - uses all scores in the data

(-) time consuming, takes a long time to calculate

Qualitative Data = non-numerical data that is normally words or written data. It produces rich, descriptive details that can then be analysed by a researcher

(+) provides rich and descriptive detail which can then be used to determine results from a study and lead to further research studies

(-) it can be difficult to analyse due to its subjective nature; one researcher may interpret some qualitative data in a different way to another researcher

Quantitative Data = numerical data that can be counted or expressed numerically; it is often collected in research studies and then manipulated and analysed

(+) easier to analyse and manipulate, especially with larger amounts of data, which also makes it more generalisable = allows for greater objectivity and accuracy of results, less likely to be affected by researcher's subjective opinions/values

(-) the emotions, motives, opinions etc of pps cannot be measured with quantitative data, lacks rich and descriptive detail unlike qualitative data