Research Methods

• Experimental method is the use of random allocation of participants and the manipulation of variables = establish cause and effect 
• Independent variable - manipulated by the researcher (change)

  Dependent variable - measured by the researcher
• Extraneous variable - variables other than the IV that affects the DV

  1. participant variables such as age and intelligence
  2. situational variables such as the environment / temperature and noise
  3. Experimenter variables concerning the personality and appearance of the researcher

  Confounding variables - uncontrolled extraneous variables that affect the results (negative)
• Operationalisation of the variables - precisely define the variables into measurable factors (easy to identify)
• Demand characteristics are features that allow the participants to work out the aim of the study = changing their behaviour to match the aim of the study

  • includes the ‘screw you effect’ that makes people purposely give wrong answers OR the other way and give all right answers to please the researcher
  • acting unnaturally out of fear and nervousness and due to social desirability bias (wanting to be liked and give the answers expected from them)
• Investigator effects are the researchers features that affect the participants response

  • age / gender / tone of voice / ethnicity / unconsciously bias
• LAB performed in controlled settings with standardised procedures = randomly allocated to conditions of the experiment (eg Baddeley encoding of memory)

  +) high degree of control meaning the IV and the DV are precisely operationalised

  +) replication by other researchers

  +) cause and effect can be established due to being controlled

  -) experimenter bias as their expectations can affect the results = influence the p’s

  -) low ecological validity = DC’s
• FIELD occur in real world settings with the researcher manipulating the IV with the rest being controlled (Bickman’s study of obedience = uniform)

  +) high ecological validity = generalise findings to real life

  +) fewer demand characteristics as they don’t know they are taking part

  -) ethical issues as if they do not know they are being observed = right to withdraw and consent

  -) less replicable due to natural setting = environment changes each trial
• QUASI theI IV is naturally occurring and can’t be directly manipulated by the researcher (it already exists = can’t randomly allocate ps to conditions)

  • typically in artificial settings

  +) more convenient and ethical as no direct manipulation of the IV and allocation to conditions

  +) often carried out in lab conditions = scientific and thorough = replicable

  -) confounding variables affect the DV = not randomly allocated
• NATURAL where the IV is pre existing like a quasi BUT it has changed over time even if the researcher wasn’t interested

  +) more practical and ethical

  +) high external validity as they study real life issues as they happen = effect of natural disasters on stress levels

  -) naturally occurring event of interest might happen rarely = little opportunity to study it = limits the scope for generalising findings

  -) p’s may not be randomly allocated to conditions = less aware of the IV and the DV
• Observational technique

  covert = p’s are unaware that they are being observed

  +) fewer SDB and DC = more valid data removes the problem of participant reactivity

  -) ethical issues
• Observational technique

  overt = p’s are aware they are being observed (EG - Zimbardo)

  +) more ethical

  -) knowledge of being observed can influence the behaviour
• Observational technique

  Participant observation involves the observer being actively involved in the situation being studied = Zimbaro

  +) increased insight of difficult to acceess behaviours = increases validity

  -) highly unethical if covert

  -) loss of objectivity
• Observational technique

  Non participant involves the observer NOT being actively involved = Ainsworth’s

  +) more objective and less likely to be bias

  -) loss of valuable insight into behaviours if psychologist is too far removed from the p’s
• Observational Design

  Naturalistic observation = natural occurring events

  +) high external validity as they are generalised to everyday

  -) replication can be difficult as environment varies

  SHAFFER AND EMERSON
• Observational Design

  Controlled observation as the environment is manipulated in some way to observe target behaviour

  +) replication is easier as more controlled

  -) not readily applied to everyday

  MILGRAM OR AINSWORTH
• Behavioural categories by dividing target behaviours into subsets of behaviours through coding systems = should reflect what is being studied 
• Time sampling - counting behaviour in a set time frame = for example every 30 seconds within 10 minute time frame

  +) reduces amount of time spent

  +) effective in reducing number of observations that have to be made

  -) behaviour may be missed if random time samples are taken across the day (every 30 seconds within a 10 minute time frame)
• Event sampling - counting number of times behaviour occurs in a target group (10 minutes)

  +) higher inter rater reliability due to being operationalised

  +) limits behaviour being observed = reduces chances of being missed

  -) problems with behaviour categories if they are too ambiguous or not defined
• Inter - observer reliability = when two independent observers have the same behaviour categories and code the behaviour in the same way

  = should come up with similar scores and can use this for a statistical test (typically go for a correlation of +0.8)
• p>0.05 being = / or more than means there is a 95% confidence rate and a 5% down to chance or luck = REJECT null hypothesis 
• p<0.05 = being less than meaning there is a 95% confidence level and a less than 5% due to luck or chance = ACCEPT null hypothesis
• If results are NOT significant then we accept the null as this assumes that there will be no significance = results are more than 5% due to chance 
• If results are significant then we reject the null as this assumes no significance = less than 5% chance 
• Type 1 error - when a difference or relationship is wrongly accepted as a real one (says pregnant when not actually pregnant)
• Type 2 errors - wrongly accepted as being insignificant (not a real difference or relationship) = null is wrongly rejected (when it says not pregnant but you are)
• normal distribution is something that tends to be around the mean (symmetrical pattern of frequency data = bell shaped curve)
• Positive (R) skewed where data skews to the right (concentrated to the right) The tail of graph points to the right

  • contain mainly LOW scores with some high outliers = really hard test produces a lot of low scores
• Negative (L) skewed where data clusters towards the left of the graph (tail points to the left)

  • mainly HIGH scores to one end with low outliers = really easy test so most people get high scores
• Nominal data

  Independent measures

  test of difference

  = CHI square
• Nominal data

  Repeated measures

  test of difference

  = SIGN test
• Nominal data

  Independent measures

  Test of relationship

  = CHI square
• Ordinal data

  Independent measures

  Test of difference

  = MANN WHITNEY
• Ordinal data

  Repeated measure

  Test of difference

  = WILCOXON
• Ordinal data

  Test for relationship

  = SPEARMANS RHO
• Interval data

  Independent measures

  Test of difference

  = UNRELATED T TEST
• Interval data

  Repeated measures

  Test of difference

  = RELATED T TEST
• Interval data

  Test for correlation / relationship

  = PEARSON'S R
• What is counterbalancing?


  technique used to deal with order effects when using a repeated measures design
• Features of science

  1. replicability - being able to repeat a study to check the validity of the results
  2. objectivity - observations made without bias = not based on viewpoint
  3. falsification - verification and validation of the process where the scientific theory is tested to see if false