Psychology Research Methods

• Operationalizing


  To decide how a variable is going to be studied. E.g. concentration will be measured used a spot-the-difference task
• Laboratory Experiment


  An experiment carried out in controlled conditions where the independent variable is controlled or manipulated
• Laboratory Experiment


  • Allows for precise control of extraneous variables, which increases reliability
  • Allows us to engineer and create situations that rarely occur, so we can study behaviour
  • Participants know they are taking part in a study, therefore consent can be gained, the right to withdraw given and the participants can be debriefed at the end
• Laboratory Experiment


  • High level of control may create an artificial situation and prevent natural behaviour from being exhibited; this may also include demand characteristics (where participants guess the aim of the study and change their behaviour), which lowers validity
  • People may behave differently in the lab, so it may be difficult to generalise to other settings, therefore having low ecological validity
  • It is sometimes necessary to deceive and cause participants a small amount of harm to improve the validity and reliability of the study
• Field Experiment


  A formal experiment conducted outside the laboratory, in a natural setting
• Field Experiment


  • Has high ecological validity as it is conducted in a natural setting, so can be more easily generalised to other settings
  • Participants may not be aware they are being studied, so they are unlikely to show demand characteristics
• Field Experiment


  • It is not possible to precisely control extraneous variables that might affect the DV. This will make them more difficult to replicate
  • Ethical issues may arise as it is often impossible to have the informed consent of those taking part
• Quasi Experiment


  Natural (Quazi) experiments are conducted in the everyday environment of the participants, but here the experimenter has no control over the IV as it occurs naturally in real life
• Quasi Experiment


  • Allows research where the IV cannot be manipulated for ethical or practical reasons
  • They are often carried out in a natural environment and are therefore high in ecological validity
• Quasi Experiment


  • There may be other differences which cannot be separated from the IV, making it impossible to infer cause and effect
  • The researcher has to wait for the conditions to occur naturally so they may not be available at the the time of the research
• Independent Measures Design


  Compares the performance of different participants in each of the experimental conditions
• Independent Measures Design


  • Overcomes problems caused by order effects, such as fatigue
• Independent Measures Design


  • More participants are needed than with repeated measures design
  • Individual differences may affect performance, you might be measuring that difference rather than what you think you are measuring. (this will affect the validity of your results)
• Repeated Measures Design


  Compared the performance of the participants in one condition with their own performance in another condition
• Repeated Measures Design


  • Fewer people are needed as they take part in both conditions
  • The same participants means that individual differences (participant variables) are reduced
• Repeated Measures Design


  • There may be order effects. (the order of the conditions may have an effect on the participant's behaviour)
• Matched Pairs Design


  Participants in each condition are matched to each other on important variables (such as gender, age, intelligence etc), to match them with someone as similar to them as possible
• Matched Pairs Design


  • Reduces individual differences (participant variables)
  • No order effects
• Matched Pairs Design


  • Time consuming and some variables might need to be measured by the researcher before the experiment. (e.g. Bandura measuring levels of aggression)
  • Impossible to match people exactly. (unless they are identical twins)
  • If one participant drops out you lose 2 participants' data
• Self-report methods


  Data is gathered by asking the participant. Examples: questionnaires, interviews and psychometric tests
• Self-report


  • Can find out things that cannot be easily discovered by experimentation or observation (attitudes, beliefs, opinions)
  • A number of variables can be examined simultaneously
  • Relatively cheap and easy to carry out
  • Can gather both quantitative and qualitative data
• Self-report


  • Some people may refuse to answer some of the questions, so a random sample is difficult to obtain
  • People may not respond truthfully: social desirability bias (where participants present an image of themselves that is socially desirable), demand characteristics or leading questions. This can be minimised using 'filler' questions
  • Open-ended questionnaires rely on the researcher's interpretation, thus it is difficult to score them objectively and analyse
• Questionnaires


  A set of written questions that are answered in writing
• Questionnaires


  • Can be easily interpreted: data can be collected from a large number of people quickly and cheaply
  • People may feel more willing to give away personal information than they would face to face
• Questionnaires


  • People may produce 'response set' answers by ticking the same answer box to all questions
  • The sample may be biased because they can only be completed by literate people with time to fill them in
• Interviews


  In an interview questions are asked face to face or over the phone. The questions may be: Predetermined (structured), Developed as the interview progresses with some set questions (semi-structured), A conversation which is allowed to flow freely (unstructured)
• Interviews


  • Detailed information can be gained from each respondent, especially if it is unstructured
  • Unstructured and semi-structured interviews can access information that may not be revealed by structured questions as the conversation is free to flow
• Interviews


  • Data collected can be easily influenced by the interviewer as the respondent may pick up on their expectations (interviewer bias), or display social desirability bias
  • It required the interviewed to be trained and skilled at getting information from people which they feel uncomfortable about giving
• Psychometric tests


  Types of questionnaires that give a standardised measurement of a psychological variable. (i.e. measure something in a controlled way - e.g. IQ tests)
• Psychometric tests


  • It provides lots of quantitative data which is easy to analyse statistically
  • It is easy to administer, therefore a very large sample can be used
• Psychometric tests


  • They often contain culture bias (e.g. IQ tests) and designer bias in favour of the creator's viewpoint. (thus 'intelligence' ends up being defined by the creator of the questionnaire, but may be inaccurate)
  • Labelling somebody as having a particular trait can encourage them to behave in that way
• Fixed / Closed Question


  The participant can only pick an answer from those provided in the questionnaire/ interview etc. ( yes / no)
• Closed Questions


  • Very easy to measure and quantify
  • No ambiguity about the chosen response
• Closed Questions


  • Preferred response might not be available
  • Researcher doesn't know the participant's reasons for their choices
• Rating Scale


  Participants rate their opinions/ beliefs on a rating scale. Example: On a rating scale from 1 to ten 10, circle the number that indicates how happy you are. ( 0 is miserable and 10 very happy)
• Rating Scale


  • Shows strength of opinion, this more detailed than a yes / no answer
  • Gives quantitative data, so it is easy to analyse statistically
• Rating Scale


  • Response bias / social desirability bias / screw you effect
  • The researcher doesn't know the reason for the participant's answers
• Likert Scale


  A scale on which participants indicate like / dislike or agree / disagree. Example: 'I believe in the existence of the supernatural' Strongly Agree / Agree / Not Sure / Disagree / Strongly disagree
• Likert Scale


  • Shows strength of opinion, this more detailed than a yes / no answer
  • Gives quantitative data, so it is easy to analyse statistically
• Likert Scale


  • Response bias / social desirability bias / screw you effect
  • The researcher doesn't know the reason for the participant's answers