research methods advanced week 2 = advanced issues in research

advanced issues in research

hypothesis

a testahle specific claim or prediction on what you expect to observe gicven a set of circumstances. statment about the relationship between variables

directional vs non directional

one tailed (directional) states direction of effect and two tailed (non directional) is more ambigious.

correlation

means that two variables vary together, and how they vary together why they vary together

correlation vs causation

causation is definitive that A changes B, correlation is less in that it suggests how variables change together. correlations used surveys, causation use experiments

features of experiments

IV manipulation, holds other variabels constant, measures change in DV

extraneous variable

anything other than the independent variable that could affect the dependent variable

confounding variable

type of extranous variable that affects the DV and the IV in systematic manner. its uncontrolled and obscure causal effects being researched

within and between subject designs

withinhas all participant tested in each condition, between allocates participants to different conditions and they arent tested across conditions

effect size in experiments

to detect a smaller effect size in a population a larger sample size is needed.

internal validity

extend on which the changes observed have been casued by our manipulation not external factors. how confident the cause anf effect relationship cannot be explained by other factors

between subjects validity threats

maturation effects (participants behaviour changes over time naturally), history effects ( something in their personal life happens or changes causing changes), testing effects ( merely having been tested before changes how they do on the post test)

threats to internal validity

regression towards the mean- an extreme score is likley to become more average( selection effects)

control groups

control groups minimise influence of maturation/testing/history effects. they don’t remove the effects but assume they are likely to control in all conditions and participants are randomly allocated. this means we can account for influence of them and see effects despite them (can be passive, active and wait list groups)

Elimiating confounding variables from participant characteristics

use a within-participant design, use random assignment to spread participant difference across the groups or use a matched pairs design

eliminating confounding variables due to experimenters or procedures

standardise prodedures, randomise the orders of conditions and ensure conditions are entirely similar excluding manipulations

differential attrition

when people leave one condition or treatment more than any other- the data becomes biased to those who compleate it.

order effects

threat of validity in within-subjects design. include practice, fatigue and habituation. participants beform better on tasks in later conditions from repetition, worse from fatigue or become desensitised to stimuli.

reducing order effects

counterbalance the participants by having groups compleate the trials in different orders. all participants do all trials but it is split in regards to which order which participants complete it (latin square design)

counterbalancing doesn’t remove order effects

there will still be other effects of order such as the fatigue or practice. however it will be balanced and minimised.

threats to internal validity- participant reactivity

control group being aware of its status (active control), (single blinding) participant behaves differently if they are aware they are being tested and try to understand what the researchers want from them. experimenter effects (double blinding) expectations can affect participants behaviour

blinding and alternative

single blinding- particpants dont know what group they are in

double blinding- participants and researchers dont know which group participants are in

deceotion- present participants with a cover story and lie about the purposes of study ETHICAL ISSUE

ceiling and floor effects

ceiling means the task is too easy or stimulus is too extreme, floor effects are if the task is too hard and stimulus is too weal. these may disguise real effects

increasing internal validity

use appropriate stimulus and measures, appropriate design (within vs between), randomisation, control group, experimental control, counterbalancing, blinding, standardisation

external validity (ecological valitity)

external validity see to which extent the results have real- world relevance. generalisability- how much our finginds can be gerenalised to other groups of people in other circumstances