Research Methods Flashcards
Correlation
When co-variables are measured for a relationship
Can be done with large sets of data, easily replicable
Extraneous variables, lower external or internal validity
Aims
The topic of investigation
Starts with ‘to investigate’
alternative hypothesis
Predictive statement, could be directional or non-directional
Random sampling
Every participant had an equal chance of being selected
Potentially unbiased because all members of the target population have an equal chance of selection
A researcher may still end up with a biased sample if some decide not to take part for example.
Directional hypothesis
Statement that predicts exactly what the outcome will be
Field experiement
In a real world setting, IV is manipulated
High ecological validity so generalisable, experimenter effects are reduced as participants are often unaware they’re being studied
Lots of extraneous variables so harder to control, demand characteristics may still be problematic as the way the IV is operationalised may give away the hypothesis
Validity
Accuracy
Opportunist sampling
When readily available people are used in research
Easy to collect sample
Biased as all the same type of people
Sampling frame
Used in random sampling - list of names that the sample is then drawn from
Null hypothesis
Statement that predicts there will be no difference
Independent variable
The thing that is manipulated
Operationalisation
Explaining exactly how the variables could be measured or changed
Quasi experiement
Another name for a natural experiment - In a real world setting, IV is naturally occurring
High ecological validity, is the only way to study certain behaviours and characteristics such as privation
Causal conclusions cannot be drawn from a natural experiment, lower internal validity
Stratified sampling
When the same proportions from the population are used in the sample
Representative
Time-consuming
Non-directional hypothesis
Statement which predicts a difference but not what the difference will be
Volunteer sampling
Advertise for participants and people put themselves forward to be in research
Easy to collect a sample, already have their consent as they’re putting themselves forward
Volunteer bias - all have lots of time or are motivated or interested in psychology
Reliability
Consistency
Systematic sampling
When every nth name is selected
Potentially random
May not be entirely random if the participants in the list are categorised i.e. Alphabetically
Dependent variable
The thing that is measured
Hawthorne effect
Hawthorne effect (added attention of being studied affects participant behaviour)
Demand characteristics
Participants change their behaviour because they think they’ve worked out the aims
Social desirability bias
Try to look good by answering or behaving in a socially acceptable way
Experimenter bias
Effects the result as they know the aims, facial expressions, back-channelling, the way they speak
Interviewer bias
Interviewer affects the responses of the interviewee
Green spoon effect
Formative noises after certain answers which affects the way the participant responds
Order effects
Repeated measures design - order of the conditions effects results e.g. Bordem or learning
Placebo conditions (Hawthorne effect)
Thinks they’re receiving the experimental condition but aren’t to see if their expectations and added attention affect the outcome or if it’s the drug itself
Single blind design (demand characteristics)
Deception - the participant doesn’t know the aims
Double blind design (experimenter bias/green spoon effect)
Experimenter and participants don’t know the Aim of the study
Standardised instructions (experimenter bias)
Written or recorded instructions used
Standardised procedures (experimenter bias)
Set time limits, order everything is conducted in
Random allocation (experimenter bias)
Participant names are drawn out of a hat when assigning which group they should go in
Counterbalancing (order effects)
Half do condition one, half do condition two and then they swap - still repeated measures but done in different orders
Independent groups
Separate groups for each condition
No order effects
Individual Differences
Matched participants
Separate groups matched on various traits e.g. Gender
Reduced Individual Differences
Order effects/demand characteristics and still there will be individual Differences
Repeated measures
Every participants completes all conditions
No individual differences
Order Effects/demand characteristics
Pilot study
Conducted to test the design and ensure variables are operationalised, a small practice run of the study, tests for reliability, ensure ethical issues are dealt with
Questionnaire
Written Questions
Could be closed (set categories to respond to) or open (participants can write anything they like in response to Q)
Easily replicable, closed questions are easy to analyse, open questions are very detailed and give insight into a person
Social desirability bias, unexpected open question answers are hard to analyse, honesty is questionable
Case study
In-depth investigation of a single person, group, event or community
Typically data are gathered from a variety of sources and by using several different methods (e.g. Observations and interviews)
Uses a variety of methods, in-depth analysis of things that are hard to study or are unethical to do so
Not generalisable due to small sample size, lacks validity as you often don’t know if you’re testing what you set out to test
Interview
Spoken Questions
Could be structured (questions pre-set) or non-structured (questions thought up in the spur of the moment)
Easily replicable if structured, structured interviews are easy to do, lots of data if open questions and qualitative
Social desirability bias, unexpected answers that are hard to analyse
Mean
Add up all the results and divide by the number of results there are
Affected by extreme values e.g. Average age of people in a classroom, teacher is an extreme value giving an ‘incorrect’ average
Mode
Most frequent/common
You have to use the mode for qualitative data
Can’t work out if there is only one of the same number for example
Median
Put the results in order and find the middle number
Not affected by extreme values because you’re just finding the middle
Range
Biggest number minus the smallest number
Affected by extreme values
Standard deviation
How far each bit of data is from the average
Not affected by extreme values
Complicated to work it out
Correlation co-efficient
Number between 0 and 1 e.g. 0.85
Nearer to 1 = stronger correlation
Has a plus or minus sign in front of the number e.g. - 0.85
The sign tells you whether the correlation is positive or negative
Content analysis (step-by-step)
Transcribe data
Break down into themes (line by line)
Code themes (so that you can collect all the information more easily)
Combine the themes into larger categories
Collect more data to see if the categories fit
Report - describe the categories discovered and back these up with quotes
Conclusions drawn and new theories produced
Content Analysis (evaluation)
Represents the true complexity of human behaviour, gains access to thoughts and feelings, rich in detail and high in validity
Difficult to detect patterns and draw conclusions, subjective (interpretation of data can be biased), cannot generalise to the population
Triangulation
We cannot generalise from it but if other studies using different research methods have
Found the same then this adds to the validity and reliability
Reflexivity
Because the interpretation of the data is subjective, the researcher needs to reflect on their own views and values and how these may have led to bias
Inductive
Usually there is no hypothesis or theory to test. Instead, data is analysed for themes and a theory emerges from it (bottom-up)
