Psychology unit 2 research

?
  • Created by: katie1414
  • Created on: 04-04-20 09:13
View mindmap
  • Psychology unit 2: Research work booklet
    • Method
      • Non experimental method
        • Observational methods
          • DESIGN
            • Non participant observation
              • Reasearcher watching or listening in
                • May be hard to gain accurate results as watching from a distance
            • Participant observation
            • Unstructured observation
            • Structured observation
              • Researcher decides what behaviour they are looking for in advance
                • Quantitative data = tally chart
                • easy to analyse quantitative
                  • Checklist assures objective
                    • checklist may miss out key characteristics of behaviour that has not been operationalised = low internal validity
        • Self report techniques
      • Sampling method
        • Random
        • Systematic
        • Stratified
        • Volunteer
        • Snowball
      • Experimental method
        • Field
          • conducted in the real world with direct manipulation over variables
            • Pilliiavin's subway samaritan study
        • Lab
          • highest possible level of control
            • loftus & palmer
            • Able to use technical equipment = replicable & provides reliable results = high validity
              • Complete control over variables
                • Easy to replicate due to control = high reliability
                  • quantitative data easy to analyse
        • researcher manipulates one variable to observe the effect upon another
    • Design
      • Experimental design
        • Matched pairs
          • pair on gender; age; height; other extraneous variables relevant - may have a pilot study to determine
          • pairs in different conditions
          • AD: 1) less order effects each one condition = h int val. ...................... 2) participants are similar - avoids blame on extraneous - results more accurate = h int ex val gene
          • DIS: 1) time consuming, difficult to match ..................... 2) not possible to control all variables so still some individual differences = int rel
        • Repeated measures
          • using same participants twice
            • ADVANTAGE: 1) controls for individual differences as the personalities are consistent = int rel. ..................... 2) need fewer participants can save cost
            • DIS: 1)demand characteristics = low int. val ...................... 2) order effects eg fatigue = low int. val
        • Independant measures
          • DIS: 1)need more participants -difficult to get/cost   ......................2) doesn't control for individual differences
          • AD: 1) less chance of participants knowing study as they have less info = less dc = higher int val ....................... 2) less prone to order effects = higher int val ....................... 2) random allocation = higher int val avoids bias
          • different participants for each condition
            • eg - one group drives no alcohol other group after pint
    • Scientific processes
  • Observational methods
    • DESIGN
      • Non participant observation
        • Reasearcher watching or listening in
          • May be hard to gain accurate results as watching from a distance
      • Participant observation
      • Unstructured observation
      • Structured observation
        • Researcher decides what behaviour they are looking for in advance
          • Quantitative data = tally chart
          • easy to analyse quantitative
            • Checklist assures objective
              • checklist may miss out key characteristics of behaviour that has not been operationalised = low internal validity
  • Quantitative data = tally chart
    • Take notes and record all action
      • Researcher does not decide behaviour in advance.
      • Qualitative data = description of action
    • No checklist means subjective
      • hard to take notes on all action without recording
      • Take notes and record all action
        • Researcher does not decide behaviour in advance.
      • Behavioural categories
      • Event sampling
      • structured: useful for tally chart as researcher needs to break up behaviour into different categories
      • therefore produce accurate results and high internal validity and reliability
      • may miss out other key behaviour that has not been operationalised - unstructured covers more detail. low int val
      • counting number of times a behaviour occurs
      • recording behaviour in a given time
      • Question types
      • closed questions
      • Contingent questions
      • Likert scaling
      • rich & detailed qualitative data
      • quantitative data lacks detail
      • depend on answer to preceding question
      • can be in the form a)b)c)
      • interview: can dig into loads of personal depth
      • improves internal validity as they gather maximum information on target behaviour
      • questionaire: may not be as personal
      • difficult to analyse objectively
      • assesses strength of opinion
      • agree/ strtongly agree...
      • produces quantitative date
      • easy to analyse objectively
      • unbiased as all members of target pop can be selected
      • imp ex eco val objective = imp int val
      • need to have list of target population
      • uses a predetermined system to select participants from a list
      • need to have list of people and set up the system
      • time consuming
      • unbiased as p are selected using objective system
      • imp int val
      • if list is target pop imp eco & ex val
      • classifying the population into categories and then choosing a sample consisting of people from each category, in the same proportions as they are in the population
      • population: all humans
      • eg 60% male 40% female sample: 20 60/100 x 20 40/100 x 20
      • more representative than other methods as it is proportional to representation of society. eg age groupd
      • less bias as based on population numbers
      • time consuming
      • eg friends, location, job
      • quick and easy as you are selecting those you know or are near
      • biased: small target population = lacks external  validity
      • all similar types of people as they are those you associate with = lacks population validity
      • researchers advertise and people from the population offer to partake
      • eg cash reward
      • gives a variety of participants as it is advertised to all people = in theory representative of population = higher external validity
      • current participants recruit further participants from people they know
      • eg drug users may contact friends - especially if cash incentive.
      • enables researchers to locate groups of people who are difficult to access eg drug
      • not likley to be cross section = lacks external validity
      • Structured interview: predetermined set of questions
        • method: interview
          • Advantages
            • Comparisons made between different participants
        • Unstructured interview: no specific set of questions: completely dependant on p answer
          • Semi-structured (clinical): some predetermined questions - but new ones developed from p answer
            • lack objectivity compared to structured
            • require more skills from the interviewer. = specialists = expensive
            • more detailed information gathered and gain deeper insights into respondents feelings and thoughts = rich & detailed qualitative data = high internal validity
              • same interviewer may not be best suited for all participants personality types. if use different interviews however = lower int val
          • questions are different for each p so information collected varies between participants = hard to analyse and draw comparisons
      • Advantages
      • easy analysis of data
      • easy ways to check and improve reliability = often good internal reliability
      • online - can reach wider spread of people than face to face large sample = high internal validity = high population validity generalise
      • can be quantified data - if use numbers as answers
      • can be anonymous, so people are more honest..study? = high internal validity
      • anonymity may cause some to not care about answers = confounding reliability
      • social desirability bias = dishonest answers
      • demand characteristics if know aim of study
      • wrong interpretations
      • may not convey full emotional response of participant
      • convey good emotional responses
      • time consuming
      • subjective data
        • use examples of TYPES to produce more points; (non) participant observation, (un)structured, designs. .......................BUT, dont fall into the trap of comparing the 2 - always think big picture
        • Disadavatage
          • of observational methods
            • Advantages
              • Researcher witnesses behaviour firsthand - can make judgements and observe body language used
                • primary data
      • Complete control over variables
        • Easy to replicate due to control = high reliability
          • quantitative data easy to analyse
      • loss of ecological validity as in a lab
        • sample bias / ethical issues
          • highest possible level of control
            • loftus & palmer
            • Able to use technical equipment = replicable & provides reliable results = high validity
          • experimental design (validity)
        • experimental design (validity)
        • less demand characteristics
          • less sample bias
            • conducted in the real world with direct manipulation over variables
              • Pilliiavin's subway samaritan study
          • high ecological validity
        • ethical issue: privacy
          • less control over variables
            • difficult to replicate
          • lack of direct intervention = less researcher bias
            • greater ecological validity
            • no demand characteristics
          • inability to place independant group designs are naturaly occuring. so could be bias in different groups
            • confounding = low internal validity .....results lack accuracy = low ex val & generalise
              • no control over extraneous confounding
            • there is an IV
              • it's naturally occurring and not manipulated
                • unique characteristics of sample = lack pop val
          • compares the relationship between 2 variables
            • Correlations
              • it shows RELATION-SHIP ............not cause and effect
                • people eat more pasta in july
            • - correlation = 0- -1
            • x correlation = 0
            • + correlation = 0-1
            • strong, moderate, weak
            • research can be done in areas that are deemed unethical, as it is only identifying relationships
              • used to investigate trends in behaviour - good base for future research
                • ethical - behaviour not being manipulated
          • does not show causal findings
            • findings may be misinterpreted as causal by public
              • lack ie validity
                • may be sample bias - cant generalise
                  • it shows RELATION-SHIP ............not cause and effect
                    • people eat more pasta in july
          • how to have good validity
            • operationalise variables
            • control extraneous + confounding variables
              • extraneous: ALL variables other than the IV that have the potential to effect the results of the experiment
                • 4 types of extraneous variables....
                  • situational variable
                    • participant variable
            • PILOT STUDY: small, trial version of proposed study - to test the effectiveness and highlight areas for improvements save time and money
            • show clear cause and effect
              • IV + DV
              • DV: what happens as a result of the changed iv
              • Having clear IV + DV produces clear causal findings
                • good external val as can generalise to outside world
                  • good internal val as measured what intended
                • ecological
          • the extent to which a study has accurately tested what it aimed to test... does it accurately answer the hypothesis?
            • internal validity
              • Validity
                • external validity
                  • the extent to which a study can be generalised to the outside world
                    • can the findings be safely used in the world?
                • ecological
                  • the extent to which a study is REALISTIC to the outside world
                • historical
                  • the extent to which a study can be GENERALISED to TIME PERIOD of outside world
          • show clear cause and effect
            • IV + DV
            • DV: what happens as a result of the changed iv
            • Having clear IV + DV produces clear causal findings
              • good external val as can generalise to outside world
                • good internal val as measured what intended
              • ecological
          • randomisation
            • IMPROVING VALIDITY IN EXPERIMENTAL DESIGN
              • counter balancing
                • repeated measures design
                  • researcher changes order of the tasks or uses ABBA technique
                    • For example, in an experiment testing the effect of alcohol on reaction times
                      • If there was only one group, A, who tested without alcohol then with, the researcher may expect that the results of the second condition are due to practice rather than the iv, alcohol.
                        • However, if they counterbalanced and introduced a group B who, opposite to A, first tested with alcohol then without, the researcher would be able to compare the data from both groups and draw more accurate conclusions.
                          • Therefore, this improves validity because although both groups will experience order effects, they will occur equally and so balance and "cancel out".
              • order effects: practice, boredom, fatigue influence experiment. Occur when participant completes a task more than once
            • researcher changes the order of tasks for each participant in a random fashion
              • for example, if the participants were to rate the attractiveness of different pictures the order of pictures would be random and different for each participant
                • Participants tend to give an average score to the first picture as they conserve higher scores incase there is a more attractive subsequent picture.
                  • Therefore, the random order of tasks allows the researcher to identify anomalous results more clearly and can produce more accurate conclusions
          • researcher changes order of the tasks or uses ABBA technique
            • For example, in an experiment testing the effect of alcohol on reaction times
              • If there was only one group, A, who tested without alcohol then with, the researcher may expect that the results of the second condition are due to practice rather than the iv, alcohol.
                • However, if they counterbalanced and introduced a group B who, opposite to A, first tested with alcohol then without, the researcher would be able to compare the data from both groups and draw more accurate conclusions.
                  • Therefore, this improves validity because although both groups will experience order effects, they will occur equally and so balance and "cancel out".
          • SINGLE BLIND DESIGN: participant is not aware of the research aims or which condition they are in. The researcher is aware. avoids dc
            • DOUBLE BLIND DESIGN: researcher & participant not aware of research aim or condition. avoids dc and re
          • the extent to which a study is consistent within itself
            • Untitled
              • Untitled
                • Untitled
                  • Untitled
                    • Untitled
              • Untitled
                • Untitled
                  • Untitled
                    • Untitled
                      • Untitled
            • internal reliability
              • Reliability
                • external reliability
              • SPLIT HALF METHOD
                • what is it? where participants do a test and the first half is as important as the second half in answering the hypothesis
          • the extent to which a measure is consistent when assessed over time or across different individuals.
          • can be safely & reliably applied to different occasions
          • GOOD: measure extent to which all questions contribute to hypothesis (what is being measured) so can see where to improve
            • can form correlations of each half. INTER ITEM CORRELATION - assesses the extent to which items on a scale all measure the same thing
              • MORE THAN 0.5
              • eg if one half has low correlation the questions arent good
            • SPLIT HALF METHOD
              • what is it? where participants do a test and the first half is as important as the second half in answering the hypothesis
          • BAD: only effective when only one thing is being measured. not effective with subscales
            • for example, Minnesota multiphasic  personality inventory > test involving subscales measuring behaviour eg depression
          • GOOD
            • 0.2 - 0.5
              • can form correlations of each half. INTER ITEM CORRELATION - assesses the extent to which items on a scale all measure the same thing
                • MORE THAN 0.5
                • eg if one half has low correlation the questions arent good
            • questions contribute but not isomorphic :)
          • HIGH
            • too specific: all questions measure identical aspect of measure. limited information produced.
          • TEST RETEST
          • if the score is consistent over time = high ext rel
          • BAD: need to use the same participants (hard/withdrawal) and researcher to keep internal validity
          • participant may have changed personality
          • GOOD: very effective in measuring ext rel
          • GOOD: useful in assessing behaviour and limits subjectivity so limits researcher bias.
          • BAD: still reliant on subjective opinion not scientific or objective method
          • IMPROVE BY:
          • operationalising categories: divide measure into categories and score each with a quantitative number scale. eg behaviour
          • standardisation: conditions same for all participants eg instructions

          Comments

          No comments have yet been made

          Similar Psychology resources:

          See all Psychology resources »See all Data analysis resources »