Research Methods Year 1- Experimental

Hypothesis

Directional- predicts nature of the relationship e.g scores in B will be higher than A

Non-Directional- predicts there will be a difference/ relationship e.g scores in B will be different to A

Null- there will not be a difference between conditions

Variables

Dependent- what the experimenter measures e.g height

Independent- what the experiment will manipulate/ change

Laboratory- highly controlled conditions and participants are aware they're taking part
+ minimised extraneous variables (potential to affect results) so high internal validity
+ replicable
- artificial conditions so lacks mundane realism
- demand characteristics

Field study- experiments carried out in natural settings but still controlled
+ high ecological validity
+ reduced demand characteristics
- less control over extraneous variables

Natural experiment- researcher doesn't control IV or DV. Pre-existing groups 
+ high mundane realism
+ reduced demand characteristics
- hard to replicate
- difficult to make causal relationships

Quasi- IV is naturally occuring e.g sex and cannot be made. 
+ allows for comparisons between different types of people
- Cannot establish cause and effect

Participants must leave a study in the same psychological state as they enter

Informed consent
ppts given as much information about the study as possible before so they can decide if they want to take part. Not everyone is able to give informed consent so have to be careful studying vulnerable peopl e.g children

Deception
deliberatelt misleading or witholding info should be avoided, as decieved ppts cannot give fully informed consent

Right to withdraw
should be made aware before and after the study that they can withdraw at any point.

Protection from harm
No physical/ psychological harm should occur

Confidentiality
Information from ppts must remain confidential so names should be changed

Independent groups design
One group of participants for each condition. Ppts are randomly allocated
+ no risk of practise effects as ppts only do one condition
- no control of individual differences e.g how good memory is

Repeated measures
The same person does a number of conditions. 
+ economical as only requires one 
+ ppt variables are controlled
- practise effects as ppts get better second time round boredom effects. ppts get bored second time round 

Matched pairs design
Ppts are paired based on similar characteristics e.g age but split into two seperate groups to make them easier to compare
+ combats ppt variables as similar in each group
+ no order effects
- matching is never completely successful and is difficult
COMBAT ORDER EFFECTS BY COUNTERBALANCING (ABBA)

VALIDITY= whether what was intended to be measured has actually been measured

Internal validity(valid within the research setting)

e.g participant awareness which is an awareness of being studied. This affects internal validity as people may show demand characteristics (pleasing experimenter) and investigator effects (behaviour of investigator affects the participants) 

How to deal with this: 

• Single blind design- only the investigator knows e.g who's recieving medicine
  
  
• Double blind design - no-one knows 

HIGH internal validity means differences found between groups were directly linked to the DV not other confounding variables

External Validity- whether the results can be applied to other settings/ ability to generalise

Generalise to:

• Different settings (ecological validity= mundane realism and generalisablity)
  dealt with by having a naturalistic setting

• Different people (population validity= can ge generalised to whole target population)
  dealt with by having a diverse sample

RELIABILITY= results are reliable if they are consistent

Internal reliability- whether what happens to one participant is consitent with the other participants.

e.g inconsistent time of day (morning/ afternoon) so not consistent.
dealt with by standardising procedures. everyone treated the same

Inter-observer reliability= two researchers produce the same outcome. it can be improved by training researchers.

Improve internal reliability by Split-half method seperate test items in half and compare.High correlation means high internal reliability

External reliability- reliability over time/ population and location

Improved by Test-retest method  same test on same ppts should produce same results.

and Replication  if a different group get same results it has high external validity

OPPORTUNITY- selecting anyone available. 
+ most straightforward. quick and easy
- not representative. only certain groups tested

RANDOM- everyone in population has chance of being picked
+ Unbiased selection 
- Time consuming and impractical

VOLUNTEER- advertisements attract participants
+ Easy of formation
- Biased as ppts likely to be highly motivated as they signed up

SYSTEMATIC- taking the nth person from a list to create a sample
+ Unbiased selection
- Does not guaruntee representativenes 

STRATIFIED- small scale production of population.dividing pop into characteristics thensampled
+ Representative as it's re-creation
-Time consuming 

NOMINAL= data is in categories 

e.g favourite food

ORDINAL= data is ordered some way

e.g order of favourite food items

INTERVAL= data is measured using units of equal intervals

RATIO= there is a true zero point, most measures of physical quantities e.g grams

Provide a single value to represent a set of numbers

Mean- add altogether, divide by total number
+ uses all the data
- diffucult to calculate

Median- middle number 
+ easy to calculate and unaffected by outliers
- doesn't work well with small sets of data

Mode- most common number
+ easy to calculate
- lacks sensitivity

Examines the variability within the data

Range-difference between the highest and lowest numbers
+ quick and easy to calculate
- provides no data about spread of values within the range

Standard Deviation- spread of data around the mean
+ very sensitive and less affected by outliers
- may hide some characteristics e.g extreme values

Normal Distribution

Symetrical bell shaped curve. 95% of the data lies within 2 standard deviations of the mean. The mean, median and mode are all equal. An example of data which would be normally distributed is height. Being very tall or very small is unusual.

Skewed distributions

If one tail is longer than the other/ asymmetric

Negatively skewed= long left tail. very few lower scores 

Positively skewed= long right tail. very few higher scores

Sign test- Looking for consistent differences between two sets of data

• Place data in table and decide whether condition A is larger/ smaller than B
• + sign if larger and - sign if smaller
• No difference then ignore
• count up how many +s and -s
• S= lower number of two
• look at critical values and compare