Variables

• Reliability= the consistency of the results, indicates whether results are trustworthy enough to draw firm conclusions. Found out through repetition of studies
• Validity= we are measuring what we set out to measure, affects whether we can generalise the results
• Extraneous variable= any variable that isn't the IV, but could affect the DV. Need to control, as they can reduce our validity
• Extraneous variables: Participant variables= connected to participants, ie age, gender, intelligence
• Extraneous variables: Situational variables= connected to research situation, ie temperature, time of day, researcher's behaviour
• Extraneous variables become confounding variables if they are not controlled for, and therefore affect one of the conditions of your IV
• Control participant variables by:

  • randomly assigning people to each condition- should hopefully "spread out" participant variables
  • Repeated measures/ matched pairs design
• Situational variables: demand characteristics

  • is a change to participant's behaviour as a result of being in the study- behaviour is no longer natural, so validity is compromised.
  • cues in the research environment that lead pps to work out the aim
  • control by: single blind technique- pps don't know which condition they're in, they are unable to guess how the researcher is expecting them to behave
• Situational variables: investigator effects

  • investigator behaves in a way to increase the chance of their hypothesis being correct- through the way instructions are given or gestures etc
  • Control by: double blind- both researcher and pps unaware of the condition each is in (reduces both demand characteristics & investigator effects)
  • Control by: standardisation- all instructions given to pps are standardised (made identical using a script etc) reducing investigator effects
• Situational variables: order effects

  • only relevant when repeated measures design used
  • pps get better between condition 1 & 2 (practice), or get worse (fatigue).
  • Control by: counterbalancing- may get half of the pps to do condition 1 first and the other half do condition 2 first
  • if they do get better/ worse, the two different groups will cancel each other out
  • Control by: randomisation- changing order the conditions are completed every time
• Situational variables: observer effect

  • when a pp is aware they're being observed, which therefore affects their behaviour & makes it less natural
  • Control by: double blind- neither pp or researcher know which condition the pp is in- reduces the likelihood of behaviour changes
• Assessing/ measuring reliability:

  • Inter-observer reliability- how closely do the ratings of the two people administering the test correlate, look for strong positive correlation between observers.
  • Test-retest reliability- whether the same test administered on 2 different occasions gives highly positively correlated scores
• Improving reliability:

  • Pilot studies- a small scale test of methods/ procedures being used on a larger scale. Allows you to check the proposed method of measurement for issues that might compromise reliability
  • Average of multiple results- find an average from more than one measurement from each pp to reduce the impact of anomalous results
• Internal validity is the extent to which:

  1. Findings can be attributed to the effect of the IV, ie minimised extraneous variables rather than demand characteristics
  2. The DV accurately measures the behaviour we are interested in- down to operationalisation of the variables, ie the way you have decided to measure your variable
• External validity= the extent to which results can be generalised outside of the experiment
• Ecological validity- to other settings/ real life & influenced by how realistic the setting or task is to the behaviour being measured
• Temporal validity- to other time frames & influenced by whether aspects such as specific cultural pressures are considered
• Population validity- to other people & influenced by how representative and unbiased the sample is
• Assessing validity:

  1. Face validity- does it look correct/ valid in the eyes of experts
  2. Concurrent validity- involves assessing how closely the scores on one test correlate with another already established assessment of the same behaviour
• Improving internal validity:

  Controlling extraneous variables (situational variables):

  1. Participant variables- control by randomisation
  2. Investigator effects- control by double blind/ standardisation
  3. Demand characteristics- control by single blind
  4. Order effects- control by counterbalancing
  5. Choosing experimental methods carefully; environments which allow you to control the variables can improve internal validity, but may compromise external validity.
  6. Using face validity or concurrent validity assessments to make alterations (amending questions or putting controls in place).
• Reliability refers to how much we can depend on a measurement- is all about consistency. We can only draw firm conclusions if results are reliable, also reliability does not mean accuracy.
• Observation:

  • Form of measurement- a researcher is recording what people/ animals are doing.
  • Assessing reliability: assessed by the observer. Repeating the observations for a second time. Looking for strong positive correlation.
  • Improving reliability: operationalising categories more, observers practising using the categories.
• Self-report:

  • Research method where the individual tells a researcher what they think/ feel.
  • Assessing reliability: test-retest reliability, measure should be reliable.
  • Improving reliability: reduce ambiguity- test items need to be re-examined and possibly re-written.
• Experiment:

  • The dependent variable in an experiment is often measured using a rating scale or behavioural categories.
  • Assessing reliability: testing if the method used to measure the DV is consistent, eg test-retest.
  • Improving reliability: standardisation- same procedures exactly each time when repeating experiments.
• Validity refers to the truthfulness or legitimacy- does the results measure what they claim to measure (ie are they a true reflection of the behaviour or are they down to the set up of the study). Validity is important as it affects our ability to generalise the results.