Extraneous v confounding variable
• both - any variable that isn’t the IV that a ects the DV
• confounding = varies systematically w/ IV ; extraneous doesn’t
Demand characteristics
• ‘please u’ v ‘screw u’
Randomisation
• use of chance to reduce the e ects of bias from researcher bias
Types of experiment
• lab
• eld
• natural - IV is naturally existing, can be changed by researcher but researcher doesn’t change it
◦lack of control / not replicable
◦high ext v due to natural setting
◦insight into real life situations
• quasi - IV is naturally existing, but cannot be changed by researcher eg age or IQ
◦individual di
◦more ethical
Experimental design
• RM/IG/MP
• Keyword: participant variables
• Counterbalancing & order e ects
Pilot study
• time scale, instructions, sampling methods, ethics
• single bind - pps unaware which condition they are in
• double bind - both pps & researcher unaware
• Control group - baseline which results from the experimental conditions can be compared
Observational techniques
• Covert/overt
• Naturalistic/controlled
• Participant/non-participant
Observational design
• inter observer reliability - total no of agreements / total no of observations x100
• Structured/unstructured
◦predetermined list of behaviours (quantitative) v researcher writes down everything they see
• Behavioural categories (operationalised)
• Time/event sampling
◦recording behaviours at predetermined intervals v tallying when target behaviour occurs
Correlations
• association between co-variables
• NO cause & e ect
• *correlation coe cients
◦> 0 = positive
• Curvilinear relationship - when variables increase tgt until one point (inverted U shape)
, Qualitative data
• subjective!
Secondary data
• aka desk research
Meta analysis
• combines results from many di studies
• Publication bias
Measures of central tendency
• mean (interval)
◦Considers all values
◦Outliers
• median (ordinal)
◦Not a ected by outliers
◦Does not re ect all scores in the dataset
• mode (nominal)
◦Not a ected by outliers
◦Can have more than 1 mode
Measures of dispersion
• range
◦outliers
◦doesn’t indicate distribution pattern across data set
• standard deviation
◦considers all values
◦outlier
Distributions
• normal
◦symmetrical bell pattern
◦68% scores lie within 1sd
◦95% within 2sd
◦99% within 3 sd
• skewed
◦data clusters to one end
• both - any variable that isn’t the IV that a ects the DV
• confounding = varies systematically w/ IV ; extraneous doesn’t
Demand characteristics
• ‘please u’ v ‘screw u’
Randomisation
• use of chance to reduce the e ects of bias from researcher bias
Types of experiment
• lab
• eld
• natural - IV is naturally existing, can be changed by researcher but researcher doesn’t change it
◦lack of control / not replicable
◦high ext v due to natural setting
◦insight into real life situations
• quasi - IV is naturally existing, but cannot be changed by researcher eg age or IQ
◦individual di
◦more ethical
Experimental design
• RM/IG/MP
• Keyword: participant variables
• Counterbalancing & order e ects
Pilot study
• time scale, instructions, sampling methods, ethics
• single bind - pps unaware which condition they are in
• double bind - both pps & researcher unaware
• Control group - baseline which results from the experimental conditions can be compared
Observational techniques
• Covert/overt
• Naturalistic/controlled
• Participant/non-participant
Observational design
• inter observer reliability - total no of agreements / total no of observations x100
• Structured/unstructured
◦predetermined list of behaviours (quantitative) v researcher writes down everything they see
• Behavioural categories (operationalised)
• Time/event sampling
◦recording behaviours at predetermined intervals v tallying when target behaviour occurs
Correlations
• association between co-variables
• NO cause & e ect
• *correlation coe cients
◦> 0 = positive
• Curvilinear relationship - when variables increase tgt until one point (inverted U shape)
, Qualitative data
• subjective!
Secondary data
• aka desk research
Meta analysis
• combines results from many di studies
• Publication bias
Measures of central tendency
• mean (interval)
◦Considers all values
◦Outliers
• median (ordinal)
◦Not a ected by outliers
◦Does not re ect all scores in the dataset
• mode (nominal)
◦Not a ected by outliers
◦Can have more than 1 mode
Measures of dispersion
• range
◦outliers
◦doesn’t indicate distribution pattern across data set
• standard deviation
◦considers all values
◦outlier
Distributions
• normal
◦symmetrical bell pattern
◦68% scores lie within 1sd
◦95% within 2sd
◦99% within 3 sd
• skewed
◦data clusters to one end
