Experiments Flashcards
Evaluate lab
S- reliability, can standards use procedure. More controls, less EVs, high internal validity
W- low external (ecological) validity. Artificial setting, artificial task leads to less natural behaviour. More effort and time into setting up
Evaluate field
S- natural setting provoked natural behaviour. High internal and external validity (ecological)
W- less control, more EVs. Lower internal reliability
Evaluate quasi
S- IV naturally occurring means high ecological validity. Hard to guess aims. Less researcher bias
W- less controls, more EVs. Lower reliability, time consuming/less useful have to wait for IV to naturally occur
Evaluate repeated measures
S- no participant variables. Use less participants, easier to obtain sample
W- order effects, demand characteristics
Evaluate independent measures
S- no order effects, less demand characteristics
W- participant variables, need more participants, harder to obtain sample
Evaluate matched participants design
S- no order effects, eliminates important participant variables, less demand characteristics
W- may have some participant variables. Process of matching is very time consuming
Participant variables and how to fix
Differences in participants, eg. Age, intelligence, motivation, expertise/experience, gender
Fix by using repeated measures or matched pairs. Use random allocation
Situational variables and how to fix
How the situation may affect the results, eg. Order effects, if they experience both conditions may be better or worse second time due to building skill or boredom
Fix by using counterbalancing ,ABBA method. Use independent measures or matched pairs
Environmental factors and how to fix
How the environment may affect results, eg. Time of day, temperature and noise.
Fix by imposing controls on the experiment.
Demand characteristics and how to fix
Cues in the experiment that communicates what is expected of participants which may affect results.
Fix by don’t tell participants aim of the study. Use single blind or double blind procedure
Single v double blind
Single: participant doesn’t know aims
Double: participants and researcher doesn’t know aims of the study
Alternative hypothesis
One tailed(directional), two tailed(non-directional) There will be a significant difference between (one condition) as opposed to (second condition)
Null hypothesis
Statement of no relation between IV and DV. There will be no significant difference between … any difference will be due to chance factors
Operationalisation
Process of making variables physically measurable or testable
Eg. Healthy defined as lower than 30 BMI unhealthy defined as higher than 30 BMI
Evaluate self selecting
S- quick and easy. Good for ethical issues- consent. Can request certain characteristics
W- not representative, demand characteristics, may get smaller sample, can be more expensive
Evaluate opportunity
S- quick and easy
W- not representative, small sample, researcher bias
Evaluate random
S- avoids bias, includes chance, representative
W- may not be willing to take part, equal chance to chose an outlier
Evaluate snowball
S- quick and easy, likely to get large sample size
W- not representative, people who they suggest are likely similar
Evaluate quantitative data
S- objective, reliable, easy to analyse and distinguish cause and effect, finding averages and making comparisons
W- low validity, lack of details- don’t know why
Evaluate qualitative data
S- high validity, in depth, detailed, quality, know why
W- subjective, low reliability, hard to analyse or distinguish cause and effect
Evaluate mean
S- most accurate, representative of all the data
W- time consuming, have to involve any anomalies, decimal results
Evaluate median
S- discounts anomalies
W- doesn’t take all data into account, can be unrepresentative, decimal results
Evaluate mode
S- easy to calculate, whole value, can do with qualitative data
W- doesn’t take into account all data, unrepresentative, might be no mode or multiple modes
Measures of dispersion
Range, variance, standard deviation
Evaluate range
S- quick and easy, shows spread between minimum and maximum
W- takes anomalies into account. Not representative of all data, doesn’t show if spread is even
Evaluate variance
S- representative- takes all data into account. Less likely to be affected by anomalies
W- time consuming and difficult. Not in same units as original measure
Evaluate standard deviation
S- representative, takes all data into account. Same units as original measure
W- time consuming and more difficult to calculate. Takes into account anomalies
How to calculate standard deviation
- Work out the mean
- For each number: subtract the mean and square the result
- Work out mean of new set of numbers
- Square root it
Under ethical guideline respect
Informed constant
Right to withdraw
Confidentiality
Under ethical guideline responsibility
Protection from harm debrief
Under ethical guideline integrity
Deception
Researcher effects
Researcher’s behaviour influencing results
Self select
Participants volunteer to take part
Eg. Through posters, adverts which contain contact details if P want to take part
Opportunity sampling
Thise readily available at a time and place of researcher choosing
Random sampling
Each member of target pop has equal chance being selected
Snowball sampling
Participants asked to contact other people to take part in their research
One tailed vs two tailed hypothesis
One tailed (directional) ‘less’ ‘more’
Eg. Men will be perceived as significantly older when having a beard compared to clean shaven
Two tailed (non directional)
Eg. There will be a significant difference in how old men are percieved to be when having a beard or being clean shaven
Nominal data
Data in categories and counted
Frequency/headcount
Nominal data eval
S- can find mode
W- cant find mean/median
Ordinal data
Data collected (often whole number scores or rating scale etc) can be ranked from highest to lowest
Interval data
Data can be ranked and we know exact gap between each piece of data
Because they use universal measurements
Eg. Time, temp
Criterion validity
How well one measure predicts an outcome for another measure
Useful for predicting behaviour in another situation
Eg. Job applicant takes performance test in interview which should predict how well the employee will perform on the job
Face validity
Does it seem to measure what it’s meant to measure at face value
- relevance to the topic, suitable for the purpose
Structured observation
Used predetermined checklist of behavioural categories
Unstructured observation
Does not use predetermined checklist of behavioural categories (make notes on what you observe in detail)
Controlled observation
Experimentor controls some variables (often lab)
Naturalistic observation
Conducted in environment where behaviour would naturally occur
External: Inter-rater reliability
Extent to which two raters are agreed in their observations consistently
- ensuring categories operationalised and interpreted the same
External: Test retest reliability
Measure external consistency of a test over time
- giving same test on different occasions to see if similar results gained
Internal: Split half reliability
Internal consistency of a test
Extent to which all parts of the test contribute equally what is being measured
- compare one half of results in a test to another half
Construct validity
Must be measuring something that exists and is measurable
Concurrent validity
Comparing
Coding scheme
Categories of behaviour for an observation
Event vs time sampling
Event- every occurrence of behaviour is observed and recorded
Time- behaviour observed and recorded at specific time intervals
Researcher bias/effect
Bias- researcher interprets/conducts in a way which influences data to reflect wanted outcome (behaviour which causes the effect)
Effect- researcher unintentionally influences outcome eg. Communicating aims to participants through showing reactions to certain data
