Research Methods

Experimental method - involves the manipulation of an independent variable to measure the effect on the dependent variable. Must be laboratory, field, natural or quasi.

Aims -aims are developed from theories and they are general statements of what the researcher intends to investigate (purpose of study), e.g. theory is that energy drinks affect how talkative people are. The aim would be to investigate whether energy drinks affect how talkative people are.

Hypothesis -clear, precise, testable statement. States the relationship between the variables. Types of hypothesis - directional (states direction of difference/relationship), non-directional (just states that there is a difference), null hypothesis (there will be no difference).

Deciding which type of hypothesis to use - directional when there is previous research, non-directional when there is no supporting research.

Variables - anything that varies or changes within the investigation.

Independent variable (IV) - the thing that changes in the experiment. To test effect of IV both a control condition and experimental condition must be used.

Dependent variable (DV) - the thing that is measured.

Extraneous variable (EV) -the thing that needs to be controlled, and will have an effect on DV if it isn't controlled.

Operationalisation -clearly defining variables in terms of how they can be mesured (stated within the hypothesis).

Extraneous variables (EV) - variable other than IV that may have an effect on the DV if not controlled, e.g. age. They're essentially nuisance variables; where possible the researcher will identify them at start of study and take steps to minimise their influence. 

Confounding variables - variable other than IV that may have affected the DV, so we can't be sure of true source of changes to the DV, e.g. personality.

Demand characteristics - any cue from researcher or situation that can be seen as revealing the purpose of the investigation. Participant reactivity is an extraneous variable that makes participants search for these cues.

Investigator effects - any effect of the investigator's behaviour on the DV, e.g. design of study or selection of/interaction with participants. The investigator may also ask leading questions.

Randomisation - use of chance in order to control for the effects of bias when designing materials and deciding the order of conditions (reduce researchers influence); it is an attempt to control investigator effects, e.g. order of conditions should be randomised.

Standardisation - using exactly the same formalised procedures and instructions for all participants in a research study. Means that non-standardised changes in procedure do not act as extraneous variables.

Experimental design - different ways in which the testing of participants can be organised in relation to experimental conditions.

Independent groups -two seperate groups of participants experience two different conditions of the experiment; one group will experience the experimental condition, the other group will experience the control condition. The performance of the groups will then be compared.

Repeated measures -all participants take part/ experience all conditions of the experiment. The two sets of data will be compared.

Matched pairs -pairs of participants are matched on some variables that may affect the DV. One member of pair will be allocated to one condition, the other will be allocated to the other condition. This is an attempt to control participant variables.

Independent groups - limitation is that participants are not the same (participant variables); to deal with this researchers use random allocation (each participant has same chance of being in one condition as any other). A strength is that order effects are not a problem, and participants are also less likely to guess the aim.

Repeated measures -limitation  is that each participant has to do multiple tasks and there may be order effects. To deal with this researchers use counterbalancing (half experience one order, other half experience other order). Strengths are that participant variables are controlled and fewer participants are needed.

Matched pairs -limitations are that participants are never matched exactly, and it may be time consuming/expensive so is less economical. A strength is that order effects and demand characteristis are less of a problem as participants only take part in a single condition.

Laboratory experiment - takes place in controlled environment where researcher manipulates IV and records effect on DV. Strict control of extraneous variables.

Field experiment - takes palce in a natural setting, the researcher manipulates the IV and records effect on DV.

Natural experiment -change in IV is not brought about by researcher but would have happened anyway. Researchder records effect on DV, e.g. natural disaster.

Quasi-expermient -IV has not been determined by anyone; the variables simply exist, e.g. age. Strictly speaking this is not an experiment.

Laboratory - strengths - high control of extraneous variables (high internal validity). Limitations - may lack generalisability; environment can be artifical and lack realism (lacks external validity). Demand characteristis may also be present.

Field -strengths - higher mundane realism than lab experiments (represents real life experiences); also has high external validity as participants arent aware they're being studied.

Natural -strengths - have high external validity and provide oppourtunities for research, e.g. orphan studies; Rutter. Limitations - may happen rarely; reducing oppourtunity and generalisability. Participants may not be randomly allocated.

Quasi -strength - same as lab experiment. Limitations - can't randomly allocate conditions, so there may be confounding variables.

Population - group of people who are focus of researcher's interest.

Sample -drawn from target population and is presumed to be representative; take part in investigation.

Random sample - all members of population have equal chance of being chosen, e.g. random number generator.

Systematic sample - selecting every nth person from list (sampling frame).

Stratified sample - reflects proportion of people in different population strata (sub-groups).

Oppourtunity samle - chosing whoever is available,e.g. anyone standing nearby.

Volunteer sample - advertise for participants willing to volunteer.

Random - strength - no researcher bias as has no control over participants. Limitations - time consuming and may end up with unrepresentative sample.

Systematic - strengths - no researcher bias, usually fair and representative. Limitation - still possible for sample to be unrepresentative.

Stratified -  strength - no researcher bias; randomly selected from each strata. Representative. Limitation - complete representation of target population isn't possible.

Oppourtunity - strength - convinient; saves time, effort and money. Limitations - unrepresentative and no generalisability (all from one area). Researcher bias.

Volunteer - strength - less time-consuming. Limitation - researcher bias; attracts certain type of person, e.g. curious.

Ethical issues - arise when conflict exists between the rights of participants and the researchers needs to gain valuable and meaningful findings.

Informed consent - involves making participants aware of the aims, procedure and their rights (including right to withdraw), and also what their data should be used for. Researchers may feel like this gets rid of 'natural' behaviour.

Deception - deliberately misleading or witholding information from participants. Linked to a lack of informed consent. Deception can be justified if it doesn't cause the participant undue distress.

Protection from harm - participants should be protected from physcial and psychological harm.

Privacy and confidentiality - participants have right to control information about themselves.

BPS code of conduct - BPS code of ethics is a quasi-legal document instructing UK psychologists on acceptable and unacceptable behaviour when dealing with participants. Built around principles of respect, competence, responsibility, and integrity.

Dealing with informed consent - issue them with consent letter to sign containing all relevant information.

Dealing with deception and protection of harm: debriefing - tell them aims of study at the end of the investigtion and any other thing they weren't told. Give them the right to withold data if they wish, and offer counselling if required.

Dealing with confidentiality - maintain anonymity, and remind participants that their data will be protected throughout the process.

Pilot study - a small-scale version of an investigation that takes place before the real investigation is conducted. The aim is to check that procedures, materials, measuring scales, etc., work, and to allow the researcher to make any necessary changes.

Single-blind procedure - a particpnat is not aware of the aims of the research and/or which condition of the experiment they're recieving. It is an attempt to control the confounding effects of demand characteristics.

Double-blind procedure -neither the participant or the researcher are aware of the research aims or other important features (e.g. who's recieving placebo or real drug)

Control groups and conditions -a control condition provides a baseline and enables a comparison.

Naturalistic observation - watching and recording behaviour in the setting within it would normally occur. 

Controlled observation - watching and recording behaviour within a structured environment, i.e. one where some variables are managed, e.g. this was used in Ainsworth's strange situation study - two-way mirror.

Covert observation -participants' behaviour is watched and recorded without their knowledge and consent.

Overt observation - participants' behaviour is watched and recorded with their knowledge and consent.

Participant observation - researcher becomes member of group they're researching

Non-participant observation -researcher remains outside the group.

Naturalistic - high external validity, so findings can be generalised to every day life. Lack control over extraneous variables, making replication hard.

Controlled - finding's can't be as readily applied to real life situations, but replication is easier.

Covert - removes problem of participant reactivity; participants will behave naturally. Increasing validity of the data. However, there are ethical issues.

Overt - more ethically acceptable but participants behaviour may be influenced.

Participant - experience situation as participants do, increasing validity. But, researcher may identify too much with them and lose objectivity.

Non-participant - researcher maintain an objective psychological distance. However they may lose valuable insight by being too far removed.

Inter-observer reliability - single observers may miss important details, or only notice things that confirm their opinion/hypothesis. Introducing bias into the research process. Research should be carried out by at least 2 people.

Training to establish inter-observer reliability -

• observers should familiarise themselves with behavioural categories,
• observe the same behaviour at the same time, e.g. in a pilot study,
• compare the data they have recorded and discuss any differences,
• and analyse the data by correlating each pair of observations.

Structured and unstructured - an unstructured obsrvation is when the researcher simply writes down everything they see. A structured observation would consist of recording observations on a pre-determined list of behaviours, and the use of sampling techniques.

Behavioural categories -a target behaviour will be broken up into components that are observable and measurable, e.g. affection may be broken down into - hugging, kissing, smiling, hand holding, etc.

Sampling methods -

• Event sampling - target behaviour or event is first established then recorded every time it occurs
• Time sampling -

Structured vs unstructured - the data produced is more likely to be quantitative with structured observations, making analysis easier. With unstructured observations the data will be more qualitative. However, unstructured observations have more detail, but also olbserver bias (may only record what catches their eye).

Behavioural categories - must be observable, there should be no 'dustbin category' (all possible forms of target behaviour should be included in the checklist), and categories must not overlap, e.g. smiling and grinning would overlap.

Sampling methods -event sampling is useful when target behaviour happens infrequently, as behaviour could be missed if time sampling was used. But the event may be too complex and details may be overlooked with event sampling. Time sampling reduces number of observations, but this could mean its unrepresentative of the observation as a whole.

Self-report technique - method where a person is asked to state or explain their own feelings, opinions, behaviours, and/or experiences related to a given topic.

Questionnaire - a set of written questions (sometimes referred to as 'items') used to assess a person's thoughts and/or experiences. May be used in an experiment to assess the dependent variable.

Open questions -does not have a fixed range of answers and respondents are free to answer however they wish. These tend to produce qualitative data.

Closed questions -offers a fixed number of responses, e.g. yes or no answers, or rating from 1 - 10. Produces quantitative data.

Strengths - 

• cost-effective; can gather large amounts of data quickly as can be distributed to large amounts of people.
• Data produced is usually straightforward to analyse, especially with closed questions.
• Open questions produce data that is rich in detail.

Limitations - 

• responses may not always be truthful; form of demand characteristics called social desirability bias.
• Often produce a response bias; may always answer in the same way, e.g. always ticking yes or answering at favoured end of rating scale.
• Acquiesence bias; agreeing with items regardless of content.

Interview - a 'live encounter' where the interviewer asks set of questions to asses a persons thoughts'experiences. Interviews can be structured, unstructured or semi-structured.

Structured interviews -made up of pre-determined list of questions in a fixed order.

Unstructured interviews -no set questions, general aim is that a certain topic will be discussed, and interaction tends to be free-flowing.

Semi-structured interviews -list of questions have been worked out, but more questions can be asked and there isn't necessarily a preferred order, e.g. a job interview is semi-structured.

Structured interviews - straightforward to replicate. However, it is not possible to deviate from topic or elaborate points, which may be a source of frustration.

Unstructured interviews -flexible, and enable interviwer to gain insight into worldview of interviewee. However, analysis is not straightforward and conclusions may be hard to draw. There is a risk the answers will be lies, but an experienced interviewer will be able to establish sufficient rapport so the answers will be more truthful.

Designing questionnaires - Closed questions can be divided into different types:

• likert scales - respondent indicates agreement, e.g. strongly disagree - strongly agree,
• rating scales - identify their strength of feeling, e.g. very entertaining - not at all entertaining, 
• fixed choice option  - included list of all possible options and respondents tick all that apply to them.

Designing interviews -should be a standardised interview schedule to avoid interviewer bias, establishing rapport should be the first priority, and there should be a good understanding of ethical issues.

Common errors -

Overuse of jargon - would be confusing and unnecessarily complex for anyone who isn't specialised in that area.

Emotive language and leading questions -emotive language will show the interviewers attitude towards the topic, and leading questions such as "is it not obvious that student fees should be abolished?" guide a respondent towards a particular answer.

Double-barrelled questions and double negatives - a double-barrelled questions contains 2 questions in one, and respondents may agree with one part but not the other, e.g. "do you agree that footballers are overpaid and should give 50% of their wages to charity?" And a double negative can be difficult for the respondent to understand.

Correlation - a mathematical technique that shows the association between co-variables (variables investigated, e.g. weight and height).

Positive correlation - as one co-variable increases the other increases.

Negative correlation -as one co-variable increases, the other decreases.

Zero correlation -there is no-relationship between the co-variables.

Difference between correlations and experiments - in an experiment the researcher controls/manipulates the independent variable in order to record the effect on the dependent variable. With a correlation, there is no manipulation of a variable, and other 'intervening variables' may have an influence.

Strengths -

•  Useful preliminary tool for research, and they show how 2 variables are related. Useful starting point to assess patterns.
• Relitively quick and economical to carry out as theres no need for controlled environment or manipulation of variables.
• Secondary data (data collected by others) can be used , meaning they're less time consuming than experiments.

Limitations -

• cannot demonstrate cause and effect, and can only say how they're related, not why (due to lack of control).
• Intervening variables (third variable problem). Correlations may be misused and misinterpreted because of these intervening variables.

Qualitative data - data that is expressed in words and non-numerical (may be converted to numbers for purpose of analysis).

Quantitative data -data that can be counted, usually given as numbers.

Primary data - information obtained first hand by the researcher for purpose of the research project. Data arrives first hand from the participants themselves. 

Secondary data - data that has previously been collected by someone other than the researcher, it has already been subject to statistical testing so the significance is known, e.g. work of other psychologists or government statistics. 

Meta analysis - research about research; combination of results from a number of related studies.

Qualitative data - richness in detail and broadness in scope, giving it high external validity. However, it can be difficult to analyse, so conclusions often rely on the subjective interpretations of researcher (could be biased).

Quantitative data - simple to analyse, conclusions are easily drawn. Much more objective and less open to bias. However, narrower in scope so may not represent real-life.

Primary data - authentic data obtained for the purpose of particular investigation from participants themselves; good validity. Targets relevent information. But, requires time and effort from the researcher.

Secondary data - easily required with minimal effort, and the desired information may already exist. But, information may be outdated or incomplete.

Meta-analysis - generalisation is a +, bias as can pick and choose is a -. 

Descriptive statistics - use of graphs, tables and summary statistics to identify trends and analyse sets of data.

Measures of central tendency -any measure of the average value in a set of data (mean, median and mode).

Mean -all values in a set of data added up, and divided by the number of values there are. Representative of data as a whole, but affected by anomalous results.

Median -central value when values are arranged from lowest to highest. Anomalous scores don't affect it, but not all scores are included in final calculation.

Mode -most frequently occuring value in the set of data. There can be 2 modes (bi-modal). Easy to calculate, but it is very crude (not always representative of the data as a whole).

Measures of dispersion - general term for any measure of the spread or variation in a set of scores.

Range - (highest score - the lowest score) + 1. Easy to calculate, but only takes into account the 2 most extreme values, so it may be unrepresentative of the ata as a whole.

Standard deviation - tells us how far scores deviate from the mean, by calculating the difference between the mean and each score. Differences are added up and divided by the number of scores (giving the variance). The standard deviation is the square root of the variance.

There may be anomalous results that make this distorted.

Presentation and display of quantitative data -

• Summarising data in a table: show descriptive statistics, not just raw scores.
• Bar charts: frequency of each variable is represented by height of bars, there will be gaps between the bars.
• Scattergrams: show correlations/associations between variables.

Distributions - Normal distribution: symmetrical spread of frequency data; mean, median and mode located at highest peak. Skewed distribution: spread of frequency data is not symmetrical, the data clusters to one end.

Positive skew: most distribution is concentrated towards left of graph, giving a long tail on the right. (Mode remains at peak, then the median, the mean is dragged to the right). Negative skew: opposite of positive skew.

Calculation of percentages - e.g. (number of participants who spoke more after a drink / total number of participants) x 100.

Converting percentage to decimal -move decimal point 2 places to the left, e.g. 37% = 37.0 = 0.37.

Converting decimal to fraction -work out number of decimal places (numbers after decimal point), number of decimals = number of zeros, e.g. if there are 2 divide by 100.

Using ratios -e.g. part to whole ratios - number who spoke more: total number, part to part - number who spoke more: number who spoke more in different condition.

Estimate results -use rounded figures, not actual figures.

Mathemetical symbols - 

• = (equal to),
• > (greater than),
• < (less than),
• >> (much greater than),
• << (much less than),
• ∝ (proportional to), 
• ≈ (weak approximation).

Probability -e.g. the probabilkity is 5% is written as P = 0.05.

Use an appropriate number of significant figures - e.g. 432,765 rounded to 2 significant figures is 430,000.

Statistical testing - determines whether or not a hypothesis should be accepted.

Concept of significance -have to show that it didn't occur by chance, do this by using a statistical test.

The sign test -analyses difference in scores between related items. Have to have done these things, 1) Look for difference, 2) use repeated measures design, 3) have data organised into categories (nominal data).

The concept of probability -accepted level of probability is 5%, a more stringent significance level is 1% which is used to give an even less possability the result was by chance.

The critical value -comparison of calculated value to  critical value to decide significance. The critical values will be given in a table.

Criteria - testing for difference, nominal data, repeated measures.

Steps - 

1) convert to nominal data, e.g. by working out which participants produced highere word count after drink and which produced lower word count. Subtract score for water from score for energy drink. If answer is negative put negative sign, if positive put positive sign.

2) Add up total number of pluses and minuses

3) S = total number of the less frequent sign

4) Compare value of S with the critical value.

Peer review - assessment of scientific work by other specialists in the same filed. Aims to allocate research funding, validate the quality and relevance of research, and to suggest ammendments or improvements.

Evaluation

Anonymity - peer remains anonymous to give more honest appraisal. But, may use this to critisise rival researchers.

Publication bias - editors prefer to publish positive results, meaning some things may be ignored or disregarded. This creates a false impression of current state of psychology.

Burying ground-breaking research - can be critical of research that contradicts their own, and more favourable to what matches it.

Attachment research into the role of the father - there can be equal care from mother and father, which may promote more flexible working arrangements. Which means parents are better equipped to maximise their income and contribute more effectively to the economy.

Development of treatment for mental illnesses - absence from work costs the economy an estimated £15 billion a year. A third of all absences are caused by mental illnesses. . But there is now more help for people with mental disorders, meaning they can return to work faster and this can bemefit the economy.