Research methods

• Aim - what the researcher intends to investigate
• Hypothesis - what the researcher believes to be true. Directional states the change, non-directional states there will be a difference
• Operationalise - making hypothesis clearly defined and measurable
• Extraneous variables - 'nuisance' variables. Don't vary systematically with IV
• Confounding variables - change systematically with IV to cause change in DV
• Demand characteristics - behaving as expected when aim of study is worked out
• Investigator effects - their behaviour eg. giving cues, changing outcome
• Randomisation - using chance to avoid bias
• Standardisation - same instructions for all participants
• Control groups - allow comparison and causation to be found
• Single blind/double blind - participant or participant & researcher don't know aim of study

Independent groups - one do condition A, one do condition B

• No order effects so no practising or getting bored
• Participant variables reduce validity

Repeated measures - same participants do A and B

• Participant variables controlled
• Easy to guess aims so get demand characteristics

Matched pairs - pairing participants based on participant variables

• No order effects
  
• Matching participants is not perfect
  

Lab experiment - controlled environment where IV is manipulated

• Can be replicated because standardised procedure is used
• Lacks generalisability as it's artificial

Field experiment - natural setting where IV is manipulated

• Participants don't know they are being studied so greater external validity
  
• Ethical issues as consent is not given
  

Natural experiment - natural setting where IV is already existing

• Only ethical option eg. studying effects of institutionalisation
  
• Participants not randomly allocated so participant variables

Quasi experiment - controlled environement where IV already exists

• Comparisons can be made between different types of people
  
• Change in DV is not always due to IV
  

Opportunity sample - people who are available in that area at that time

• Quick and convenient
• Biased as all from one area

Volunteer sample - people select themselves through adverts

• Willing participants so will engage properly
• Biased as all participants could have the same traits

Random sample - everyone has equal chance of being selected from lottery method

• Unbiased 
• Not necessarily representative of population

Stratified sample - selecting people according to frequency in population

• Representative
• Can't represent all ways in which people are different
  

Conflict

• between rights of participant and aims of research
• benefits and negatives are weighed up to decide whether study can go ahead

Informed consent

• people choose whether they want to take part
• don't give away too much info by getting them to agree to being decieved, asking a similar group or getting consent after

Deception 

• deliberately misleading people
• details of true aim, other conditions and use of data should be given after study

Protection from harm

• people shouldn't be at more risk than they would everyday
• right to withdraw given, reassurance afterwards and counselling provided

Confidentiality

• personal information should not be shared
• protect personal details, use initials or code names and data not shared with other researchers

Illustrates strength and direction of association between two variables

Uses scattergram

Postive - variables rise/fall together, Negative - one falls the other rises

No manipulation of IV so cause and effect is not established, just relationship between variables

Useful starting point for research to assess strength/direction of relationship to make hypotheses

Economical - no need for controlled conditions or manipulation

No cause and effect so intervening variables may explain the relationship

Flawed methods of research - low in reliability and validity

Naturalistic - where target behaviour would normally occur

• High external validity so very generalisable
• Low control over EV's

Controlled - control/manipulation of variables

• Can be replicated - uses standardised procedures
  
• Low external validity as setting changes behaviour

Covert - unaware of being studied

• Less demand characteristics 
  
• Unethical as right to privacy ignored
  

Overt - aware of being studied

• Ethical as consent is given
  
• Demand characteristics

Participant - researcher becomes part of group being studied

• Greater insight into behaviour so higher validity
• Loss of objectivity by identifying too strongly with group

Non-participant - researcher remains separate from group

• More objective so less bias
• Loss of insight reduces validity

Behavioural categories - target behaviour broken into set of observable categories

• Difficult to make clear and overlapping is likely
• Dustbin categories - all behaviours should be in the list

Time sampling - observations made at regular intervals

• Reduces observations - more structured and systematic
• Unrepresentative - not all behaviour will be observed

Event sampling - target behaviour recorded every time it occurs

•  Infrequent behaviour can be recorded 
• Complex behaviour oversimplified - affects validity

Pre-set list of questions participants respond to, used to assess DV

• Can be distrubuted to lots of people to get lots of data
• Easier to speak about personal information
• Untruthful responses
• Response bias eg. always agreeing

Closed questions

• Easy to analyse
• Restricted answer - cannot expand on true opinions

Open questions

• Unrestricted answer - more detail and insight
• Hard to analyse

Questions should never use technical vocabulary, be double-barrelled or be leading questions

Face to face interaction

Structured interview - pre-determined questions asked in set order

• Easy to replicate - standardised format
• No elaboration of anwers

Unstructured interview - no set questions, just general topic to be discussed

• Greater flexibility so more insight
  
• Difficult to replicate

Should be in quiet room so interviewee will open up, neutral starting questions to build rapport, answers kept confidential

Quantative data - numerical data

• Easy to analyse
• Oversimplifies behaviour into specific categories

Qualitative data - non-numerical data

• Allows detail and unexpected recordings
  
• Hard to analyse

Primary data - first hand data collected for investigation

• Only necessary data relevant to aims
  
• Very time consuming

Secondary data - collected by someone other than the researcher eg. articles, books, etc.

• Inexpensive and little effort 
  
• Outdated/incomplete data
  

Meta analysis - combining data from lots of studies

• High validity - large sample
• Negative/non-significant studies may be ignored

Measures of central tendency - mean, median, mode

Measures of dispersion - range, standard deviation

Standard deviation - measure of average spread around the mean

Used to analyse difference in scores between related items.  Can be used with nominal data.

1. Score from B subtracted from A to produce difference sign (+ or -)

2. Total no. of pluses and total no. minuses should be calculated 

3.  Same score in A and B should be taken away from N value

4.  S value is total of less frequent sign

Critical value - if S is equal to or less than this value, S is significant and hypothesis is retained

All aspects of investigation should be scrutinised by experts in the field, who are unknown to the researcher.

Allows good allocation of research funding, good quality of research and improvements can be suggested.

• Protects quality of published research to minimise fraudulent research
• Reviewers use anonymous status to criticise rival researchers
• 'Headline grabbing' findings most likely to be published
• Reviewers more critical of work that contradicts their own