research methods and experimental design Flashcards
extraneous variables
all variables apart from the independent variable which could affect the results of an experiment e.g. temp of the room
confounding variables
act as an additional independent variable - affects the result of an experiment e.g. age, gender
demand characteristics
bias possibly formed from cues inadvertently provided that suggest to the participants what the outcomes of the investigation may be
participant reactivity
when the participant tries to please the expectations
investigator effects
unwanted influences that the investigator/experimenter communicates to the participants which affects their behaviour and so biases the results
chi^2 test
used when your data is in categories - to see if the data is related
what are independent groups or pairs?
subjects allocated to different groups with different experimental conditions
matched pairs
subjects are matched for at least one variable that could affect the outcomes of the experiment e.g. age, gender
repeated measures
all subjects experience both experimental conditions
counterbalancing
all subjects experience both experimental conditions but in different orders - stops the order of the tasks affecting the outcome
random allocation
subjects once matched are then randomly assigned to the different experimental conditions
nominal data
used to label variables without providing a quantitative value e.g. genotype, gender
ordinal data
classified into categories within a variable to have a natural rank order
sign test
used to measure difference between results
directional hypothesis
a prediction made by a researcher regarding a positive or negative change, relationship or difference between two variables of a population
Analysis for significance
- spearmans correlation (non-parametric)
- pearson’s rho correlation (parametric)
- both pearsons and spearmans test the strength of association between two continuous variables and both can be used with quantitative data
choose pearsons if…
- data is linear
- change between the variables is proportional
- have numeric data
- can only use raw data
choose spearmans if…
- data is monotonic (may have curvy bits)
- change between variables may not be proportional
- have numeric or ordinal data
- data will be ranked
data for non-parametric tests
- nominal or ordinal
- not normally distributed
- outliers/ anomalies
- unequal variances
- small samples
interval level data
data measured in fixed units equal distance between points on the scale
normal distribution
data distributed evenly around the mean
skewed distribution
data does not look normal
