Comprehensive Guide to Scientific Method, Research Types, and Causation in Psychology

Scientific method

A systematic process of gaining knowledge through empirical observation analysis and replication

Empirical

Based on data collected through observation or measurement

Why is the scientific method self correcting

Because findings are tested repeatedly through replication and can be revised

Confirmation bias

The tendency to notice information that confirms existing beliefs

Why is personal experience unreliable

Because it involves bias and confounding variables making conclusions inaccurate

Method of tenacity

Believing something because it has always been believed

Method of authority

Believing something because an expert says it is true

Empiricism

Knowledge gained through sensory experience

Reason

Using logic to draw conclusions

Three goals of science

Description prediction and explanation

Description

Observing and recording behavior

Prediction

Using relationships between variables to forecast outcomes

Explanation

Determining cause and effect relationships

Hypothesis

A testable statement about the relationship between variables

What is causation

When one variable directly produces a change in another

Three requirements for causation

Covariation temporal order elimination of alternative explanations

Induction

Developing theory from data

Deduction

Testing theory using data

Theory

A set of statements explaining a variety of occurrences

Function of theory

Organizes data and generates predictions

Parsimony

Explaining phenomena with the simplest explanation

Testability

A theory must be able to be tested and potentially proven wrong

Observational research

Watching and recording behavior without manipulating variables

Why no causation in observational research

Cannot rule out alternative explanations

Naturalistic observation

Observing behavior in a real environment

Participant observation

Researcher joins the group being studied

Structured observation

Behavior is observed in a controlled setting

Case study

In depth analysis of one individual or event

Survey

Collecting self reported data from participants

Meta analysis

Combining results from multiple studies

Observer bias

Researchers see what they expect

How reduce observer bias

Use operational definitions training and blind observers

Reactivity

Participants change behavior when being observed

How reduce reactivity

Use hidden measures or allow participants to get used to observer

Correlation

A relationship between two variables

Correlation coefficient

A number from negative one to positive one showing strength and direction

Positive correlation

Variables increase together

Negative correlation

One increases while the other decreases

Zero correlation

No relationship

Why does correlation not equal causation

Because of directionality and third variables

Directionality problem

Unclear whether one variable causes the other

Third variable problem

A hidden variable affects both variables

Why can no correlation still be misleading

Because relationships may be nonlinear or restricted

Experiment

A controlled test where a variable is manipulated to observe its effect

Independent variable

The variable manipulated

Dependent variable

The variable measured

Control variables

Variables held constant

Control group

Group not exposed to the independent variable

Why compare groups

To isolate the effect of the independent variable

Random assignment

Assigning participants randomly to conditions

Why use random assignment

To evenly distribute confounding variables

Confounding variable

An outside factor that affects results and creates alternative explanations

Why are confounds a problem

They prevent clear cause and effect conclusions

Construct validity

How accurately a study measures what it intends to measure

How improve construct validity

Use clear operational definitions and consistent procedures

Internal validity

Confidence that the independent variable caused the effect

History threat

Outside events influence results

Maturation threat

Participants change over time

Testing threat

Prior testing affects performance

Instrumentation threat

Measurement tools change

Attrition threat

Participants drop out

Selection threat

Groups differ before the study

External validity

Ability to generalize results to real world situations

How improve external validity

Use representative participants variables and settings

Reliability

Consistency of measurement

Measurement reliability

Consistency across trials

Statistical reliability

Results unlikely due to chance

Experimental reliability

Ability to replicate findings

Test reliability

Same results across repeated tests

Random sampling

Selecting participants randomly from a population

Purpose of random sampling

To improve generalizability

Random assignment vs sampling

Sampling selects participants assignment places them into groups

Nominal scale

Categorical with no order

Ordinal scale

Ranked but unequal intervals

Interval scale

Equal intervals no true zero

Ratio scale

Equal intervals with a true zero

Which scale allows three times greater

Ratio scale

Ranking preferences uses

Ordinal scale

Categorizing responses uses

Nominal scale

Between subjects design

Different participants in each condition

Within subjects design

Same participants in all conditions

Carry over effects

Previous condition affects later performance

Order effects

Performance changes due to order of conditions

How reduce order effects

Randomization or counterbalancing

Counterbalancing

Varying order of conditions across participants

Complete counterbalancing

All possible orders are used

Problem with complete counterbalancing

Too many combinations

Incomplete counterbalancing

Only some orders are used but balanced

Latin Square design

Each condition appears once in each position

Advantage of Latin Square

Controls order effects efficiently

Mixed design

Includes both between and within subjects variables

Main effect

Overall effect of one independent variable

Interaction

Effect of one variable depends on another

Two by four by three design IVs

Three independent variables

Total levels in two by four by three

Nine levels total

How many groups in a two by four by three design

Twenty four groups total

Complete randomization

Participants randomly assigned to conditions

Block randomization

Conditions presented in repeated randomized blocks

Advantage of block randomization

Balances conditions across time

What is an example of a history threat

A major outside event occurs during the study and influences participants behavior