Research methods

Extraneous variables = external variables that are unwanted and could affect whats being measured

Confounding variables = when the extraneous variables have an effect and mask true result

Aims = general statement of why study is being carried out

Hypothesis = a clear statement, prediction, testable, can be directional/non-directional

External validity = the extent to which you can apply results to other people/places/situations

Inter-rater reliability = extent to which two or more people agree when analyzing behaviour

Internal validity = are we measuring what we intend to within the study

Event sampling = focuses only on actions/events that of particular interest to researcher

Time sampling = observations made only during specified time periods

Point sampling = individual is observed in order to categorise current behvaiour after which a second is observed and so on  

Lab exp - +ves:                                              -ves

• Can be replicated                                       Lacks external validity
• Less risk of extraneous variables                  Demand characteristics
• Good control of variables                             Lacks mundane realism
• Establishes C/E                                         

Field Experiment - natural but IV still manipulated

• No evaluation apprehension                         Lots of extraneous variables
• High external validity                                   Low in internal validity
• Can establish relationships                          Ethical issue - informed consent

Natural Experiment - pre-existing groups, type of Quasi experiment

• Pps not aware                                           No random allocation
• High ecological validity                                IV not directly manipulated
• Look things unethical to manipulate             Difficult to identify what causes behaviour

Case Studies:

• Produces rich detailed data                                     Results cannot be generalised
• Permits study over time                                          No control over extraneous variables
• Study a topic unethical to manipulate                       Researcher bias may be a problems

Interviews

• Good reliability                                                       Hard to analyse
• Non-verbal cues                                                      Social desirability bias
• Lots of personal information                                     Only get info person consciously aware of

Questionnaires

• Easy to analyse (numerical data)                              Social desirability bias
• Large amounts of data short amount of time              Many poorly constructed
• Reasonable validity and highly reliable                      Often used in correlational studies - noC/E

Quasi Exp: Looks like an exp, features are weak, can't randomly allocate pps to conditions

Non-directive interview: Interviewee free to discuss, Interviewer guides, useful in treatment of mental disorders, less relevant to research

Semi-structured: Interviewer knows topics and order, interviewees same qus in same order

Clinical: same questions, but some can be added, flexible, questions asked depend on answers

Fully structured: Set questions,order and answers yes/no/don't know

Xs Interviewer bias

Single blind procedure - interviewer knows hypothesis, interviewee doesnt reduces demand characteristics

Double blind procedure - eliminates investigator effects as someone other than researcher asks questions 

• Open (qualitative) and closed (quantitative)
• Open strength: in depth detailed info, weakness: difficult to analyse
• Closed strength: easy to analyse, weakness:don't provide much information
• Can be too complex and ambiguous
• Psychologists should avoid leading qus and double barrelled items
• Test-retest technique - no of individuals given same questionnaire on different occasions and scores correlated High correlation = reliable
• Validity - comparing scores on a questionnaire to external criteria
• Standardisation: tests administered to large sample so results can be compared to others
• Split-half technique - set of questionnaires, split in two scores from one half compared with the other half 

Natural Observations:

• Pps behave naturally                                       No control over variables - hard to establish C/E
• Provide richer and fuller info                             Training observers difficult and expensive
• Easier to generalise to real life                          Impossible to replicate

Controlled Observations

• Less risk of extraneous variables                       Artificial situation people won't behave naturally
• Can establish c/e                                             Little external validity
• More complete data than experiments as           Investigator effects                                         witnessing some natural behvaiour                     Demand characteristics   

Experimental (or alternative) hypothesis - prediction for an experiment when IV changes

Null hypothesis - Predicts no change and is easier to disprove than prove

Directional hypothesis - one tailed hypothesis - says which direction the change will be

Non-directional hypothesis - two tailed - doesn't state the way the change will go 

Content analysis - analysis of behaviours, written or spoken word into pre-set categories, known   as coding. Can make qualitive data quantitative

• + - Gives opportunity to understand people as rounded indviduals in a social context
• + - Can reduce complex behaviours/info into manageable categories
• + - Suggest interesting hypotheses that could be tested in subsequent research
• - - Social desirability bias
• - - Data may be very unrepresentative making it hard to generalise
• - - Researcher collects a huge amount of material then they could show bias by emphasising only the info that favours their hypothesis 

Independent groups - different pps experience each condition - only experience one condition

• Identical stimulus materials                             Greater extraneous variables
• Demand characteristics less likely                     More pps -> more expensive
• Absence of order effects

Repeated measures - same pp experiences each condition - experience both conditions

• Reduced extraneous variables                          Order effects - can counterbalance
• Fewer pps needed - cheaper                            Demand characteristics more likely                                                                                         May need different stimulus materials

Matched pairs - different pps closely matched experience each condition - each half 1 condition

• Absence of order effects                                   Difficult to find well-matched pairs
• Identical stimulus materials
• Demand characterisitcs less likely

Counterbalance - one group do AB, other BA

Order effects - constant errors introduced to an experiment because of repetition

Random sample

• Every member of population equal chance of being chosen - most representative
• May not all be available to do study, may opt out of study

Volunteer sample

• Pps volunteer when asked or in response to an advert
• May be a volunteer bias - demand characteristics

Opportunity sample

• Most common involves people who are there at the time and fit criteria you are looking for
• Unrepresentative (might be), quite narrow

Stratified sample

• Classifiying into categories then choosing a sample which consists of pps from each category in the same proportions as they are in population

• Psychological consequences
• Get informed consent
• Avoid deception
• Pps should be able to withdraw at any point
• Pps should be debriefed after the study
• Pps have the right to expect confidentiality
• Pps should be protected from harm
• Psychologists should not encroach on pps privacy
• Psychologists should only give advice on matters for which they are qualified
• Colleagues research

• Nominal data - used when categorising something into named groups
• Ranked/ordinal data - data ranked so possible see order of scores in relation to each other
• Interval/ratio data - gives rank order of scores and precise intervals between scores

Measures of central tendency -mean,median,mode

Mean

• More sensitive than median                            Misrepresentative with outlier
• Uses all data

Median

• Not affected by extremes                                Less sensitive than mean

Mode

• Useful when data in categories                        Not useful when several modes

Measures of dispersion - range and standard deviation

Range - difference between highest and lowest scores

Standard Deviation: Average amount all scores deviate from mean

To calculate standard deviation:

• Difference between each score and mean of those scores is calculated + then squared
• Squared deviaitons are added and their mean calculated to give value known as variance
• Square root of the variance gives standard deviation of the scores

Graphs -histograms (interval data only),bar charts, pie charts, freq. polygrams, scattergraphs 

Correlational studies are used to asses strength of relationship between variables, no direct manipulation of IV by experimenter

Correlation coefficients - way of measuring strength of relationship                                                 +1 perfect positive, 0 no correlation, -1 perfect negative

Correlations:

• Quick and easy to collect data                                       Can't establish C/E
• Easy to establish trends between variables                      Problem of extraneous variables
• Can do correlations where exps. unethical                        Ethical issues - studies sensitive

Quantitative:

• Objective                        No details
• Scientific                        Hard to fit into categories
• Patterns                         No feelings/emotions

Qualitative:

• Chronological order                         Subjective - opinions
• Degrees of severity                          Can be exaggerations
• More detailed