Research Methods ALL Flashcards
Aim
An intention of an investigation to see if or to investigate whether
Hypothesis
Statement of expected outcome. A prediction and what you find. Null and alternative hypothesis and operationalised
Null
IV does not effect DV
Alternative
IV effects DV
Directional
Specific about what the effect will be (sufficient background evidence)
Non directional
IV will affect DV but not specifically (when no previous research or to avoid bias)
Extraneous variables
Could effect DV but should be controlled
Confounding variables
Already have affected the DV
Operationalised
Telling how you will conduct and measure to make hypothesis clear and testable
What is the point of an experiment
Keep variables constant and manipulate one variable to se effect on DV, establishing a causal relationship
Summarise lab experiments
- High control of extraneous variables
- Controls IV and measures DV
- Good as increase confidence that IV affects DV and more reliability due to high control
- Bad because artificial situation so lacks ecological validity and lacks mundane realism
Summarise field experiments
- Real world, so little control over extraneous variables
- P’s do not know
- Good as no demand characteristics and high validity as P’s act naturally
- Bad as loses control of extraneous variables so less causal
- Harder to replicate
Natural
IV is a situation or environmental factor
Quasi
IV is an individual difference
What is good and bad about Natural and Quasi
- Study variables that we cannot in other ways due to ethics
- Cannot manipulate the IV so may not be causal as no control
Summarise repeated measures
- Ps take part in both so effect in DV due to IV and not pts variables
- Controls individual differences as score 1 and 2 compared so higher internal validity
- Less pts
- Demand characteristics
- Practice and order effects
How can we deal with p and e effects
- Counterbalancing ABBA
- Assign randomly
- Randomisation
Summarise matched pairs
- Each person in one condition matched with someone in the other on a factor such as age
- controls panda effects and individual differences
- practically difficult
- hard to decide which factors are more important
- how to measure the variable we are matching
- Hard to find a match in large sample sizes
Summarise independent measures
- Divides randomly into 2 groups, P only completes once
- Reduces demand characteristics
- Avoids panda effects
- More ps needed
- Individual differences
Randomisation
Use of chance to control for effect of bias when designing materials and order of conditions
Standardisation
All procedures standardised so all ps subject to same environment
Demand characteristics
- Features or cues which help ps work out what is expected
- May respond according to what they think is being investigated so invalid results
Social desirability bias
- ps behave in way to present the best behaviour
- researcher should focus on experimental realism, so task is engaging and ps forget they are being observed
Investigator effects
- Any effect of investigators behaviour on the research outcome
Pilot study
- Small scale trial to identify potential issues and modify design and save time and money
What must a sample be
Representative to generalise
Random sample
- Each person has same chance of being picked
- Unbias selection and probability means it will usually be representative
- Chance of being unrepresentative so cannot generalise
Systematic sample
- Taking every nth person so avoids bias as no control
- probability means it will usually be representative
- Chance of being unrepresentative so cannot generalise
- Not as objective as random as researcher may decide on who is listed before selection and the nth number
Stratified sample
- Sampling frame divided into groups
- Number from each group taken that is representative
- Ensures sample representative and objective as after stratas, left to chance
- Time consuming
- Researcher may not identify all key characteristics
Opportunity sample
- Made up of ps available at the time
- Easy and convenient and less time
- Biased as small, and only certain people say yes
- Researcher may show bias when selecting ps
Volunteer sample
- Ps volunteer to take part
- Access tow idea population
- Only a certain type of person
Questionnaires
- Self report
- Open and closed questions
- Collect large data quickly, see what people are thinking and good reliability
- Influenced by motivation, social bias and does not reflect thinking
What makes a good questionnaire?
- Clarity so understand Q
- Unbiased
- Quant for closed and vice versa
- Enough responses for useful data, too many = boredom
- May need filler question s to avoid demand characteristics
- More challenging ones at the end to build trust
- Pilot study
Structured Interview
Face to face with Qs
- Pre determined questions asked in fixed order
- Easy to replicate
- Interviewee cannot elaborate and so may miss useful information
Unstructured interview
Convo
- Topic discussed and interaction free flowing
- Interviewee expands
- More flexibility but risk of social bias
Natural observation
Observed in natural setting
- Investigator does not interfere
