Research Design and Ethics Notes

Scales in Surveys

• Scales should primarily use positive numbers, such as 1 to 10 or 1 to 5.

• For agreement scales, use options ranging from "strongly disagree" to "strongly agree."

• Avoid simple "agree or disagree" scales as they may not capture the full range of opinions.

Question Order

• The order of questions can influence responses.

• Asking about relationship satisfaction before life satisfaction may skew the latter if the relationship is problematic.

• Start with general questions before moving to specific ones to reduce bias.

Random Sampling

• Randomization is crucial when surveying a subset of a population.

• The selected sample must represent the entire group to ensure the data is representative.

• Every participant should have an equal chance of being chosen for the study.

Research Designs

Case Studies

• Considered an older method.

Correlational Studies

• Examine the relationship between two or more variables.

• Determine if a relationship exists without determining causation.

• Example: Examining the relationship between variables x and y.

• Not concerned with whether x impacts y or vice versa, or if a third variable influences the relationship.

• Strength: Provides information about relationships and associations between variables.

• Flexible and can be applied to practical or unethical situations.

• Weakness: Does not allow for scientific control of variables.

• Experimenters cannot control extraneous variables.

• Correlation does not equal causation.

• Example: Length of marriage correlates with hair loss in men, but marriage doesn't necessarily cause hair loss; a third factor, such as age, may be involved.

• Example: Parenthood is associated with happiness. This is not always factual or true because it doesn't mean that all parents are happy.

• Example: Adolescents who feel loved by their parents are less likely to smoke, use alcohol, or drugs. We cannot conclude that not feeling loved by parents directly causes these behaviors.

Types of Correlation

• Positive Correlation: Both variables increase or decrease together. As one axis goes up, another axis goes up.
  *Visual representation: Both go up. This graph shows a perfect positive correlation.

• Negative Correlation: As one variable increases, the other decreases. The axis coming down
  *Visual representation: Both axes are going down. This graph shows a perfect negative correlation.

• No Relationship: No discernible linear relationship between variables.
  *Visual representation: No line in the graph can be formed

Questions Answered by Correlational Studies

• How closely are personality traits related within identical twins?

• How well do intelligence tests predict academic achievement?

• How closely is stress related to illness?

Examples:

• The more children use media, the less happy they are (negative correlation).

• The less sexual content teens see on TV, the less likely they are to have sex (positive correlation).

• The longer children are breastfed, the greater their academic achievement (positive correlation).

• The more income rises, the fewer psychiatric symptoms children experience (negative correlation).

Experimental Studies

• Involve manipulating variables to determine causation.

Within-Subject Design

• Variables are tested within the same subject.

• All participants are exposed to all conditions.

• Seeks to identify individual factors.

• Requires fewer participants.

• Minimizes individual differences.

• Con: Potential for carryover effects, where one condition influences behavior in another.

• Individual differences include fatigue and boredom.

• Practice effect: Performance improves with repeated tasks.

• E.g., In memory tests, scores reported lower initially, but it gets higher as a participant gets used to it.

Between-Subject Design

• Looks at differences between two or more groups.

• Different groups are exposed to different conditions.

• Two different tests are being designed to see what are similarities, differences, and correlations.

• Faster but potentially more biased.

• Susceptible to situational and individual differences, such as motivation and intelligence.

Experimental Design Elements

• Treatment or experimental group: Receives the treatment or test.

• Control group: Does not receive the treatment; used for comparison.

• Participants are randomly assigned to either group.

Quasi-Experiment

• Similar to a regular experiment but lacks random assignment of participants.

• Independent variable is present, but certain factors (age, gender) cannot be manipulated.

Examples

• Does longer periods of breastfeeding lead to increased intelligence in children?

• One study randomly assigned newborns to a control group (normal pediatric care) or an experimental group (promoted breastfeeding).

• A British study assigned premature infants to formula or breast milk, finding breast milk more effective for developing intelligence.

• Breastfeeding example is a between-subject design with a control group.

• Unlike correlational studies, experiments manipulate factors to determine impact or causation.

Evaluating New Drug Treatment

• Use a random sample of participants.

• Implement a double-blind procedure, where neither the participants nor the researchers know who receives the actual drug versus a placebo.

Placebo Effect

• Improvement in condition due to belief in treatment, rather than the treatment itself.

• Example: Participants told they were consuming a high-calorie milkshake experienced a steeper decline in hunger hormones.

Preventing Placebo Effect

• Use a double-blind procedure.

• Randomize participants into control and experimental groups.

Independent vs Dependent Variables

• Independent Variable: The factor being manipulated.

• Dependent Variable: The outcome being measured.

Research Design Recap:

• Descriptive: observe and record behavior

• Correlation: look at a relationship between 2 factors to see how well one predicts the other

• Experiment: manipulate factors to determine effect

The "WEIRD" Problem

• WEIRD = Western, Educated, Industrialized, Rich, and Democratic societies.

• Using WEIRD samples limits the generalizability of findings.

  • Creates discrimination and consequences

  • Has no representing culture or mentality.

  • Not at all diverse.

• Most psychology research is conducted on participants from WEIRD countries.

• This can lead to bias and non-representative results.

• Important to weigh practical specifics when considering sample diversity.

• Truth vs Opinion.

  • Facts

    • Can be verified

    • Unbiased information

  • Opinions

    • Bias

    • Consist of judgement

Research Ethics

Animal Research

• Ethical considerations in animal testing.

• Balancing potential benefits (curing cancer) with animal welfare.

• Some have opinions whether or not this is right to do so.

• Some prioritize human life more than animal life.

• Should have proper guidelines to developing.
  *

Human Research

• Institutional Review Board (IRB) to ensure ethical standards.

• Informed consent from participants is required, or from guardian if it is a child.

• Milgram Experiment investigated obedience to authority and personal conscience, but was considered unethical due to the stress induced on participants.

• Ordinary people are likely to follow orders.

• Stanford Prison Experiment showed how quickly people can abuse power, but was also highly controversial due to psychological abuse of participants and potential coaching from researchers.

• The movie based on this experiment was shown during the presentation.
  *Some say this movie was more of a simulation.

• Debrief participants immediately after study.

• Deception comes into play.

Ethical Guidelines

• Collect informed consent.

• Protect participants from harm and discomfort.

• Maintain confidentiality.

• Debrief participants after the experiment.

Key Terms

• Variables: Anything that can vary (age, intelligence, TV exposure).

• Independent Variable: The factor being manipulated in an experiment.

• Dependent Variable: The outcome being measured in an experiment.

• Random Assignment: Minimizes preexisting differences between groups.

• Representativeness: Surveying a group that represents a country's adult population.

• Negative Correlation: The more childbirth training classes women take, the less pain medication they require.

• Visual Representation: Scatterplot

• Double-Blind Procedure: Neither the person assigned or researchers knows what they are receiving

• Random Sampling: A sample of the population (e.g., pulling a random sample from a hat).