Lecture 2: Research Method

Question 1

Variables
= ____
Conceptual variable = ____
Operational definitions = ____
Effective operational definitions rely on ____

One conceptual variable can have ____ definitions; we choose one definition in given study
This choice is a compromise between ____, ____ and _____; there is no ____.

Answer 1

An attribute that assumes different values across people, places and timepoints.

Conceptual variable = an abstract variable that can’t be measured directly (e.g. stress, intelligence)

Operational definitions = things associated with a conceptual variable that can be measured to infer the conceptual variable (e.g. psychological test that defines stress as a participant’s score on a stress questionnaire)

• Effective operational definitions rely on criteria validated by the academic community

One conceptual variable can have many definitions; we choose one definition in given study

This choice is a compromise between validity, cost and convenience; there are no perfect measures

e.g. long personality questionnaires are valid but nobody will not participate.

Question 2

Operational Definition - Psychological Tests
Limitations and Biases in Psychological Tests
____ Effects = ____

Accuracy in Rating Yourself Varies
People with ____ disorder
____ = ____
____ = ____

Accuracy of Rating Others is Better (but still flawed)
____ effect = ____
____ effect = ____

Answer 2

Limitations and Biases in Psychological Tests

Framing Effects = The way a question is worded taps into people’s unconscious bias and influence how people answer.

Accuracy in Rating Yourself Varies

Self-ratings of people with narcissistic personality disorder can be highly inaccurate.

Positive impression management = Exaggerate traits to appear socially desirable.

e.g. Rate oneself high for extroversion, low for anxiety

Malingering = Exaggerate problems for external rewards (e.g., avoiding responsibility, gaining sympathy or benefits)

Accuracy of Rating Others is Better (but still flawed)

Halo effect = Attractive people are rated more positively overall.

Horns effect = Unattractive people are rated more negatively overall.

Question 3

Operational Definition - Psychological Tests
Considerations for a Psychological Test to be Trustworthy
Test-test reliability = ____
Inter-rater reliability = ____
Construct validity = ____
i.e. ____

Answer 3

Test-test reliability = Consistent test result every time we do the test

Inter-rater reliability = Consistent test result every time no matter who is scoring the test

Construct validity = The degree to which a construct (i.e. our test) measures what it claims to be measuring

i.e. Does your test result predict the relevant things in the real world? (e.g. intelligence test results predict GPA, income…)

Question 4

Operational Definition - Psychological Tests
Test Result Analysis
Distribution = ____
The Normal Distribution
Shape: ____, ____
Central tendency: ____
Variability: ____

Skewed Distribution
Shape: ____, ____.
Central tendency: ____

**Central tendency **
Mean = ____
- _____
- _____ to outlier ()
→ ____

Median = ____
- _____ to outlier
→ ____

Mode = ____
- ____

Variability
Range = ____
- ____
- ____ to outlier

Standard Deviation (SD) = High SD means ____
Low SD means ____
(High and low ____)
- ____ to outlier
Outlier = ____

Answer 4

Distribution = a graph of how values are spread out across all the values that a variable can assume

The Normal Distribution

Shape: Symmetrical (no skewness), Limited **kurtosis

Central tendency: Mean = Median = Mode

Variability: 68% of cases between ± 1SD, 95% between ± 2SD

Skewed Distribution

Shape: negatively skewed, positively skewed

Central tendency:

Central tendency

Mean = The average score of the variable within a population

• Most useful measure
• Sensitive to outliers (particularly if the sample is small)→ mean is not representative when the distribution is not normal

Median = The middle score of the variable in the population

• not influenced by outliers→ median is representative when the distribution is not normal

Mode = The most frequently occurring value in a population

• Not used in formal analysis

Variability

Range = The distance between highest and lowest score

• Large ranges theoretically possible, but rare (e.g. Grades could vary from 0 – 100, but 0s and 100s are rare)
• Sensitive to outliers

Standard Deviation (SD) = How much a score in the population
typically deviates (±) from the mean

High SD means a lot of spread around the mean

Low SD means little spread around the mean

(High and low are compared to mean)

• Less sensitive to outliers

Outlier = A score at least 2 SD units away from the mean

Question 5

Research Methods
Hypothesis
= ____
- must be ____ and ____.
- beginning of ____
Hypothesis is not supported by the research → ____ → ____

Answer 5

Hypothesis

= An evidence-based explanation of a phenomenon.

• Must be clear and testable
• Serves as a starting point for a research

Hypothesis is not supported by the research → Revise approach → Construct another hypothesis

Question 6

Research methods
Research approaches
The research approach we use depends upon ____ and ____.
1. Descriptive Research
= ____ → (infer causation?)
Case studies = ____
Benefits: ____, ____, ____
Limitations: ____

Surveys = ____
Benefits: ____, ____
external validity = ____

Naturalistic observation = ____
Benefits: ____, ____
Limitations: ____, ____, ____.

Answer 6

1. Descriptive Research

= Observation of variables → can’t infer causation

Case studies = An intensive examination of one individual to provide proof of existence to rare phenomena.

Benefits: Inspire new hypothesis; common in medicine; basis of Freud’s theories

Limitations: Low generalizability

Surveys = Record data on many variables in a large population
via questionnaires or interviews.

Benefits: Highly generalizable; valuable in social psychology

(external validity = whether the results of a study can be generalized to other contexts)

Naturalistic observation = Observation of an animal/human in its natural setting without direct intervention.

Benefits: Highly generalizable (observation in animal/human’s natural environment); reducing observer effect

Limitations:

Poorly controlled (hard to isolate causes);

Limited range of variables can be assessed (can’t access mental processes);

Difficult to study infrequent behaviours and thoughts.

Question 7

Research methods
Research approaches
2. Correlational Research (‼️an ____, not a ____)
= ____ → (infer causation?)
Correlation coefficient (____) = ____ measure of ____
- ranges from ____ to ____.
the ____ the ____, the stronger the ____.
- only suitable for ____

Correlation relationships of different strength
____, ____, ____
____, ____, ____

Why can’t correlation infer causation?
1. ____: ____
2. ____
3. ____: ____

Answer 7

= measure of the strength of correlation between two variables → can’t infer causation

Correlation coefficient (r) = statistical measure of the strength of correlation between two variables

• absolute r values range from 0 – 1

The higher the absolute r value (positive or negative), the stronger the relationship!

• Only suitable for linear relationships

Correlation relationships of different strength

Why can’t correlation infer causation

1. Non-linearity: curvilinear relationships cannot be accurately described by the correlation equation (r = 0)
2. The Third Factor
3. Spurious associations: correlations emerge by chance

Question 8

Research methods
Research approaches
3. Experimental Research
= ____ → (infer causality)
Independent variable/IV = ____ ()
Dependent variable/DV = ____ ()
All other variables are ____ if possible.
Why does experiment infer causality? ____

Elements of a True experiment
1. ____
Groups/Conditions =
Control group =

2. ____ = ____
   → ____
   Quasi-experiment = ____ (____!)
   - Used when ____

Study design
Between-subject Design = ____, ____
Control group = ____
Within-subject Design = ____
Control group = ____
Mixed Design (Between and Within Factors) =
In within-subject designs and mixed designs, subjects are compared ____
→ allow us to control ____

Answer 8

1. Experimental Research

= a controlled environment wherein we study the relationship between a defined set of variables by controlling all the rest → can infer causation

Independent variable/IV = Variable manipulated (Cause)
Dependent variable/DV = Variable Measured (Effect)

All other variables are kept constant if possible.

Why does experiment infer causality? Because all variables are kept constant but the IV, any change in the DV must be due to the IV.

Elements of a True Experiment

1. Manipulation of an IV

Groups/Conditions = experimental groups and control group

Control group = the group that does not receive the manipulation, used as a reference point to compare the DV in other groups.

1. Random assignment (makes a true experiment) = randomly place participants into different groups, so that each person has an equal chance of being in any group.

→ ensures any differences in the DV are more likely caused by the IV, not by pre-existing differences between participants.

Quasi-experiment = experiment without random assignment to groups (not a true experiment)

• Used when random assignment is not ethical or practical

1. Experimental Research

= a controlled environment wherein we study the relationship between a defined set of variables by controlling all the rest → can infer causation

Independent variable/IV = Variable manipulated (Cause)
Dependent variable/DV = Variable Measured (Effect)

All other variables are kept constant if possible.

Why does experiment infer causality? Because all variables are kept constant but the IV, any change in the DV must be due to the IV.

Elements of a True Experiment

1. Manipulation of an IV

Groups/Conditions = experimental groups and control group

Control group = the group that does not receive the manipulation, used as a reference point to compare the DV in other groups.

1. Random assignment (makes a true experiment) = randomly place participants into different groups, so that each person has an equal chance of being in any group.

→ ensures any differences in the DV are more likely caused by the IV, not by pre-existing differences between participants.

Quasi-experiment = experiment without random assignment to groups (not a true experiment)

• Used when random assignment is not ethical or practical

Study Design

Between Subject design = 2+ groups, each given a different treatment

Control = no treatment group

Within Subject design = One group observed 2+ times (e.g. before and after treatment)

Control = group before treatment

Mixed Design (Between and Within Factors) = 2+ groups, each
observed 2+ times

In within-subject designs and mixed designs, subjects are compared to themselves

→ allow us to control for confounding variables

Question 9

Concerns with Experimental Design
1. Sampling Bias
To generalize results of any one study to other contexts, we need to make sure our study’s sample is ____.
The W.E.I.R.D. Problem = ____
⚠️____
Addressing the WEIRD Problem
1. ____
2. ____
3. ____
4. ____

2. confounding variables
   = ____, making it _____
3. Observer/Hawthorne effect
   = ____
4. The Power of Expectation
   Demand Characteristics = ____
   Limited by ____ and/or ____.

Placebo Effect = ____
Limited by ____
→ If a treatment truly works, ____

Single-blinding/participant blinding = ____
- Mandatory in ____ and ____ study

More on Placebo
- Placebos can work even if ____
- Associated w/____
- ____ over the past few decades in America

5. Experimenter effect
   = ____

Double-blinding = ____
→ Eliminate both the ____ from participants and the ____ from the researcher

Answer 9

1. Sampling Bias

To generalize results of any one study to other contexts, we need to make sure our study’s sample is representative of that context.

The W.E.I.R.D. Problem = 80% of studies focus on Western, Educated individuals living in Industrialized, Rich + Democratic countries (while Non-WEIRD communities may differ in many ways)

⚠️Does not mean all WEIRD studies are useless

Addressing the WEIRD Problem

• More diverse samples
• More settings
• Acknowledgment of additional interpretations
• Replication of prior work in different contexts

1. Confounding Variables

= **an uncontrollable **variable that influences IV (no effect when there is) and/or DV (has effect when there actually isn’t), making it hard to tell if the results are truly caused by the IV.

1. Observer/Hawthorne Effect

= When people alter their natural behaviour simply because they know they’re being watched.

1. The Power of Expectation

Demand Characteristics = When participants guess the purpose of the study and change their behaviour to fit what they think the researcher expects.

Limited by opacity (vague) **and/or deception (misleading)

Placebo Effect = a real **change in behaviour simply because they believe they are receiving a treatment—even if it’s inactive.

Limited by including a group given a placebo with no active gradients

→ If a treatment truly works, it will affect behaviour more
than either a placebo or no treatment

Single-blinding/participant blinding = Groups do not know which treatment they are getting

• Mandatory in drug and alcohol study
• Placebos can work even if you know they’re placebos
• Associated w/neurophysiological changes
• Increased over the past few decades in America

1. Experimenter effect

= Researchers unconsciously behave in a way that gets the result they expected

Double-blinding = both the participants and the researchers interacting with them don’t know who is in the experimental or control group.

→ Eliminate both the placebo effect (from participants) and the experimenter effect (from the researcher)