psychology - research methods

Psychology - research methods

Experimental method – involves the manipulation of IV to measure the DV

Hypotheses

-          Directional (two tailed) – researcher has made clear the sort of difference anticipated between conditions

-          Non-directional (one tailed) – researcher states that there is a difference between conditions but the nature of the difference is not specified

Operationalise variables – make them more measurable (e.g. 300ml)

Type I error - when null hypothesis rejected when it should have been accepted (too linient)

Type II error - when null hypothesis accepted when should have been rejected (too stringent) - misusing a real effect in an experiment - real difference in data overlooked

5% used as it strikes a balance between risk of making type 1 and type 2 errors

Research issues

Extraneous variables – any variable other than IV that MAY affect DV if its not controlled

Confounding variables – a variable other than IV that DOES affect DV

Demand characteristics – any cue from researcher/research situation that may be interpreted by participants as revealing the purpose of an investigation (participant may change behaviour)

Investigator effects – any effect of investigators behaviour on research outcome

Standardisation – using exactly the same procedure for all participants

Experimental designs (the different ways in which participants can be organised into conditions

Independent groups – participants allocated to different groups where each group represents one experimental condition

Repeated measures – all participants take part in all conditions of the experiment - avoids effects of individual differences in frequency of nightmares

Matched pairs – pairs of participants are first matched on some variables that may affect DV (e.g. age) then one member of the pair is assigned to each condition - avoids order effects and less likely to guess aim so improve validity

Random allocation – attempts to control for participant variables in an independent groups design – each participant has equal chance of being in each condition (no bias)

Counterbalancing – attempts to control for the effects of order in repeated measures design – half participants experience condition in one order and the other half in the opposite order (ABBA)

Types of experiment

Lab experiment – takes place in a controlled environment within which the researcher manipulates IV and records effect on DV whilst maintaining strict control of extraneous variables

Strengths – high control over confounding and extraneous variables (can establish cause and effect = high internal validity)

Weaknesses – lack generalisability – artificial and not like everyday life (low external validity and mundane realism), also participants are aware they are being tested on which could lead to demand characteristics

Field experiment – takes place in a natural setting within which the researcher manipulates IV and records effect on DV

Strengths – more natural = high mundane realism – producing more valid/authentic behaviour

Weaknesses – loss of control of CV and EV – cause and effect more difficult to establish, ethical issues as participants may not be aware they are being studied = haven’t consented (lack of privacy)

Natural experiment – change in IV is not brought about by researcher (would’ve happened even if researcher not there) – researcher records effect on DV

Strengths – provide opportunities for research that may not otherwise not be partaken due to practical/ethical reasons, high external validity (real-world problems/issues studied)

Weaknesses – rare events = reducing opportunities for research and limits scope for generalising findings to other similar situations, participants may not be randomly allocated to experimental conditions (less sure of cause effect)

Quasi experiment – a study that is almost an experiment but the IV is not manipulated by a researcher and cannot be changed (e.g. age)

Strengths – controlled experiment usually = high internal validity

Limitations – cannot randomly allocate participants (CV)

Sampling

Population – group of people who are focus of researcher’s interest (from which a smaller sample is drawn)

Sample – group of people who take part in research investigation – sample drawn from a target population (presumed to be representative)

Bias (in context of sampling) – certain groups being over/under represented within sample selected (limits generalisability to rest of population)

Generalisability – extent to which findings and conclusions from a particular investigation can be broadly applied to population

 Sampling types:

-          Random – all members of target population have an equal chance of being selected

+Unbiased (CV and EV have no effect as they should be equally divided between conditions = increased IV)

-difficult and time consuming to conduct

-          Systematic – every nth member of targeted population is selected to produce a sampling frame (a list of people in target population organised e.g. into alphabetical order)

+objective

-time consuming

-          Stratified – composition of the sample reflects the proportions of people in certain subgroups (strata) within target population

+produces representative sample (designed accurately to reflect composition of population) = generalisability is high

-doesn’t reflect all the ways which people are different

-          Opportunity – selecting anyone who is willing and available

+convenient, less costly and less time consuming

-unrepresentative of target population as it is drawn from a specific area (lacks generalisability), researcher has control over selection of participants (researcher bias)

-          Volunteer – participants select themself to be part of sample

+easy and less time consuming, the participants are more engaged

-volunteer bias – a certain ‘profile’ of people may be attracted (curious and tries to please researcher – affecting generalisability)

Ethical issues

Informed consent – participants are aware of the aim of the research, procedures, what the data will be used for and their rights

Deception – deliberately misleading/withholding information from participants

Protection from harm – participants should not be placed at any more risk than they would in their daily lives and protected from physical and psychological harm

Right to withdraw – participants can leave the study whenever they want

Confidentiality – personal data is protected (anonymity)

BPS code of conduct (British Psychological Society) – ethical guidelines implemented by ethics committees who use a cost-benefit approach

Types of study

Pilot studies – a small-scale version of an investigation that takes place before the real investigation is conducted (checks that procedures, materials, measuring scales etc work to allow for the researcher to make modifications if necessary) - identify any methological weaknesses and see if any modifications in design e.g. whether interview questions sufficiently relevant to dream content

Single-blind procedure – participants don’t know the aim of study/which condition they are in

Double-blind procedure – neither participant or researcher is aware of aim/conditions – often used in drug trials (placebo)

Control group – nothing is changed for this group – used for comparison

Observational techniques

Naturalistic – watching and recording behaviour in the setting within which is would usually occur

+high external validity (generalised to everyday life)

-lack of replication due to lack of control over situation, also CV/EV uncontrolled

Controlled – watching and recording behaviour within a structured environment

+CV/EV less of a factor = replication easier

-less external validity (less generalisability)

Convert – ‘closed’ – participants behaviour is watched and recorded without their knowledge or consent

+removes demand characteristics and ensures for natural behaviour (increased internal validity)

-ethical issues – privacy, lack of consent

Overt – ‘open’ – participants behaviour is watched and recorded with their knowledge and consent

+more ethical

-demand characteristics present

Participant – researcher becomes a member of the group whose behaviour they are watching and recording

+researchers get an increased insight into lives of people being studied = increased external validity

-researcher may identify too strongly with participants and loose objectivity

Non-participant – the researcher remains outside the group whose behaviour they are watching and recording

+researcher remains an objective view

-lose valuable insight to be gained in participant observation

Observational design (how researcher would plan an observational study)

Behavioural categories – when a target behaviour is broken up into components that are observable and measurable

Event sampling – a target behaviour or event is first established then the researcher records this event every time it occurs

Time sampling – a target individual or group is first established then the researcher records their behaviour in a fixed time frame

Self report - any method in which a person is asked to state/explain their own feelings, opinions, behaviours and experiences related to a given topic

Self report techniques:

-          Questionnaires

Open questions – produce qualitative data

Closed – produce quantitative data but lacks depth

+time effective – collect large amounts of data quickly, statistical analysis/comparison easy

-people may lie (social desirability)

-          Interviews

Structured – predetermined questions asked in a fixed order

+straightforward to replicate as standardised

-limited information as no follow ups

Unstructured – no set questions just a general aim

Semi -structured – list of questions but follow up questions asked

+more flexibility (richer information)

-interviewer bias possible, analysis is harder

self reports contain social desirability bias so do not reflect what really happens so data lacks validity

Correlations: AO1

Positive correlation - as one covariable increases so does the other

Negative correlation - as one covariable increases the other decreases

Zero correlation - no relationship between covariables

Correlation coefficient - number between -1 and 1 showing how strong correlation is (1=strong, -1=weak)

CORRELATION DOESN’T ALWAYS MEAN CAUSATION

Correlations: AO3

+provide a precise and quantifiable measure of how 2 variables are related - suggesting ideas for possible future research if variables demonstrate an interesting pattern or strong relationship

+quick and economical to carry out - secondary data can be used

-studies only tell us how variables are related but not why - doesn’t demonstrate cause and effect

-intervening variable - another untested variable causing relationship between the 2 covariables

-therefore correlations can be misinterpreted and therefore misused

Types of data

Quantitative/Qualitative - numerical/descriptive (more detail so more understanding)

Primary/secondary - first-hand/second-hand

Meta-analysis - process of combining findings from a number of studies on a particular topic to produce an overall statistical conclusion based on a range of studies

Nominal – most basic level of measurement, used when data is put into a tally chart/categories, gives very little information (only tells how many people are in each group)

Ordinal – used when data can be put into order – but units of measurement are not of equal size and it is usually based on opinion (subjective)

Interval – most complex level of measurement, equal gap between each unit of measurement (e.g. cm)

Measures of central tendency

Mean/median/mode

Measures of dispersion

Range/standard deviation

Presentation of quantitative data

• Table

• Bar charts - used for discrete (categoric) data - bars don’t touch

• Histograms - used for continuous data - bars touch

• Scattergrams - depict relationships/correlation between covariables

Distribution of data

• Normal - bell shaped curve (symmetrical) - mean, median and mode occupy same midpoint of curve, tails of curve never touch 0

• Positive skew - distribution leans to left side (tail on the right) - mode remains at highest point of peak, median lowered slightly, mean is moved towards tail more. Order is mode, median, mean

• Negative skew - distribution leans to right side (tail on the left) - mode remains at highest point of peak, median lowered slightly, mean is moved towards tail more. Order is mean, median, mode

  • A way to remember this is that mode is always at the highest point and median is always in the middle (of mean and mode) - therefore mean is lowest

Statistical testing

Sign test

• Determines significance and looks for a difference

• USING A SIGN TEST: DIFFERENCE, REPEATED MEASURES DESIGN, NOMINAL DATA

• Performing sign test:

  • subtract category 2 from category 1

  • add up positive and negatives signs

  • S = less frequent sign (this is the calculated value)

  • Calculated value more than critical value = no significance

Choosing stats test:

• difference or correlation - usually comes from hypothesis wording

• unrelated design = independent groups

• related groups = matched pairs, repeated measures

Peer review (the assessment of scientific work by others who a specialists in the same field, to ensure research intended for publication is high quality)

Aims of peer review

• allocate research funding

• validate quality and relevance of research e.g. methodology

• suggest amendments or improvements

• ensure accuracy

process

• other psychologists check research report before deciding whether it could be published

• independent scrutiny by other psychologists working in similar field

• work considered in terms of validity, significance and originality

• assessment of appropriateness of methods and designs used

• reviewer can accept manuscript, accept with revisions, suggest author makes revisions then resubmits or rejects

• editor makes final decisions whether to accept or reject research based on reviewers comments

+establishes validity and accuracy of research

-some reviewers may use their anonymity negatively e.g. criticising a rivals research (competition for limited funding)

-publication bias - editors want to publish significant findings with positive results so some research may be disregarded if doesn’t meet this criteria - creates false impression of psychology

-reviewers are more critical of research that contradicts their own view and more favourable to those who match it - slows down rate of change if opposition to mainstream theories is being buried

Reliability - measure of consistency

ways to assess reliability

• test- retest - administering same test to same people on different occasions and seeing if obtained results are similar

• inter-observer reliability - observers should conduct observations in teams so results have more reliability as 1 person deciding the results could lead to subjectivity bias

• where possible use lab experiments as researcher has strict control over variables and it improves ability to replicate study

measuring internal reliability

• split half reliability used for test, questionnaires or interviews - comparing 2 halves of test by randomly selecting and putting these halves onto 2 forms and each form should yield same score if test reliable - can calculate coefficient to compare (looking for coefficient of +0.80 or more)

measuring external reliability

• test retest - giving same test to same person on 2 separate occasions to see if same results are obtained - calculate coefficient

• interobserver reliability - researchers observe same behaviour independently and compare data - if data similar then it’s reliable - can be improved by training observers and ensuring categories operationalised

Validity - the extent to which an observed effect is genuine and can be generalised to real world, does it measure what it was meant to measure

• internal validity - the extent to which the effects of investigation are due to manipulation of independent variable

• external validity - relates to factors outside of study:

  • ecological validity - extent to which findings from a research study can be generalised to other settings and situations

  • temporal validity - extent to which findings from a research study can be generalised to other historical times and eras

• face validity - does the study appear to measure what its supposed to measure

• concurrent validity - extent to which a psychological measure relates to an existing similar measure

• investigator effects is where investigator affects result of studies e.g. they may influence participants answer - one way to minimise is to give interviewers provide standardised questions to ask (script) to avoid bias in responses

Improving validity

• include more than 1 question which would make the aim of experiment less obvious to guess which would in turn reduce demand characteristics and improve validity of experiment

consent form: CCRADAL

• consent

• confidentiality

• right to withdraw

• activity

• disadvantages

• advantages

• length

• explain general purpose of research

• check box and signature

scientific reports

• to give credit to other researchers

• enables readers to track down sources used

• can avoid plagiarism

content analysis

• a method of quantifying qualitative content via coding/categorisation

• sampling = researcher decided what material to use

• familiarise with material = become familiar with the types of material likely to be encountered and construct a system for categorising the data based on aims of study

• common themes = research must decide how to categorise the analyses material by making coding categories

• tally = record number of occurrences of a particular coding category - produces data at nominal level of measurement

• +generates quantitative data so stats test so significance tested

• +reliability as people can repeat

• -subjectivity - choice of categories and definitions of what comes under each category decided by researchers

• -validity as reduces rich qualitative data to numerical values - wider picture lost

thematic analysis

• avoids lack of validity that comes from content analysis - examines data without turning qualitative data into quantitative data

• sampling = researcher decided what material to use

• familiarise with material

• common themes to coding categories

• combine simpler codes into larger categories/themes

• use new categories by applying to new data they should fit the new data if they represent the topic investigated