Research Methods Exam #2

Bivariate Correlation

Examines the relationship between two variables

What does bivariate correlation determine?

Direction & Strength

How does bivariate correlation determine direction?

Positive: both variables increase/decrease together

Negative: one increases while the other decreases

How does bivariate correlation determine strength?

It is measured by the correlation coefficient ( r )

Correlation Coefficient ( r )

Ranges from -1.00 to +1.00

Closer to +1 = stronger relationship

Closer to 0 = weaker relationship

Is r = 0.85 what kind of correlation?

Strong positive

Is r = -0.70 what kind of correlation?

strong negative

Is r = 0.10 what kind of correlation?

very weak

Ice cream sales and drowning rates are correlated. Does this mean ice cream causes drowning?

No! Correlation does not equal causation.

Multivariate Correlation

examines relationships among three or more variables

used to understand more complex patterns than bivariate studies

Multiple Regression

A statistical technique used in multivariate research.

What statistical analysis would be used to predict an outcome from several variables?

Multiple Regression

Advantages of multivariate correlation

• More realistic

• Can control for other variables

Disadvantages of multivariate correlation

• Harder to interpret

• Requires larger samples

population

entire group of interest

Sample

subset of the population

Probability Sampling

everyone has a known chance of selection

Stratified Sampling

Population divided into groups.

Cluster Sampling

Randomly select entire groups

Systematic Sampling

select every nth person

Nonprobability Sampling

where participants are chosen using non-random criteria

Convenience Sampling

Whoever is easy to access

Purposive Sampling

Select participants with specific characteristics.

Snowball Sampling

Participants recruit others.

Sampling Error

Natural random variation between a sample and the full population because you only observed a part of it.

Sampling Bias

Systematic distortion where certain members have a higher chance of being included than others.

Selection bias

A broad error caused by improper participant selection, lack of randomization, or uneven dropout rates. It skews study results and ruins generalizability.

Volunteer Bias (Self-Selection Bias)

A specific type of selection bias where people choose whether to join a study. Volunteers often have stronger opinions, more free time, or a higher personal interest in the topic than non-volunteers.

Unrepresentative Samples

A subgroup whose characteristics do not match the broader target population, causing data analysis to reflect only a skewed portion of reality.

Single-Subject Research

Research focusing on one participant or a very small number of participants.

Researchers repeatedly measure behavior over time.

A-B Design

A = Baseline B = Treatment

A-B-A Design

Baseline - Treatment - Remove treatment (used to see whether behavior returns to baseline)

A-B-A-B Design

Baseline - Treatment - Remove Treatment - Reintroduce treatment (Shows stronger evidence that treatment caused change)

Advantages of Single Subject Research

• Useful for clinical settings

• Detailed individual analysis

Disadvantages of Single Subject Research

• Limited generalizability

• Time intensive

Baseline → Treatment → Baseline → Treatment

A-B-A-B Reversal Design

Survey Research

Collect information through questionnaires or interviews.

What is survey research used to measure?

• Attitudes

• Beliefs

• Behaviors

• Opinions

Reliability

Survey gives consistent results.

Validity

Survey measures what it is supposed to measure.

Correlation

relationship between variables

Survey

method of collecting data

Bivariate

2 variables

Population

everyone of interest

Sample

subset studied

Reliability

consistency

Validity

accuracy

Random sampling

a statistical technique where researchers select a subset of individuals from a larger population such that every member has an exact, equal, and independent chance of being chosen

r = 0.10

Weak effect size

r = 0.30

moderate effect size

r = 0.50

strong effect size

How do I know if the statistic is significant?

p < .05

How do I know if the statistic is not significant?

p ≥ .05

What are the three things that affect correlations?

• Outliers

• Restriction of range

• Curvilinear relationships

Why can't correlations establish causation?

Because they do not establish:

1. Temporal precedence

2. Internal validity

Group research

• Looks at averages

• Combines participant data

Case studies

• Detailed description of one person

• Often generates ideas for future research

• Internal and external validity can be concerns

Stable Baseline Design

• Baseline

• Treatment

• Compare treatment to baseline

Multiple Baseline Design

Treatment introduced at different times across:

• Participants

• Behaviors

• Settings

Reversal (ABA) Design

• Baseline (A)

• Treatment (B)

• Return to baseline (A)

Validity

Single-subject designs can have:

• High internal validity

• High construct validity

• External validity may be limited

High internal validity

You are 100% confident that the independent variable caused the change in the dependent variable, without interference from confounding factors.

High construct validity

Your study’s actual tests, treatments, and questions perfectly reflect the abstract concepts you claim to be studying.

External validity

Generalizability (Real-world application)

Likert Scale

Measures degree of agreement.

Semantic Differential Scale

Measures meaning between opposite adjectives.

Acquiescence

Participant agrees with everything

Fence Sitting

Participant constantly chooses neutral responses.

Socially Desirable Responding

Participants answer in ways that make them look better.

Reverse-Coded Items

Used to detect careless or patterned responding

Representative Sample

• Equal chance of selection

• Generalizes well

• Higher external validity

Biased Sample

• Some people more likely to be selected

• Unknown generalizability

• Lower external validity

Systematic Sampling

• Pick a random starting point

• Select every nth person.

Cluster Sampling

• Randomly choose groups.

• Survey everyone within chosen groups

Multistage Sampling

• Randomly select clusters.

• Randomly select individuals from those clusters

Proportionate Stratified Sampling

Divide into meaningful groups (strata) and sample proportionally from each group

Disproportionate Stratified Sampling (Oversampling)

Purposefully overrepresent a group to ensure adequate representation

Purposive Sampling

Recruit people with specific characteristics.

Quota Sampling

Researcher ensures certain subgroups are included but uses convenience sampling within those groups

Random Sampling: →

Improves external validity

Random Assignment: →

Improves internal validity

What does a positive correlation mean?

As one variable increases, the other variable also increases (or both decrease together).

What does a negative correlation mean?

As one variable increases, the other variable decreases.

What does an r value of +1.00 indicate?

A perfect positive correlation.

What does an r value of -1.00 indicate?

A perfect negative correlation.

What does an r value near 0 indicate?

Little to no relationship between variables.

According to Cohen's guidelines, what is considered a small correlation?

r = .10

According to Cohen's guidelines, what is considered a medium correlation?

r = .30

According to Cohen's guidelines, what is considered a large correlation?

r = .50

True or False: Correlation proves causation.

False

What is the directionality problem?

We cannot determine which variable causes the other.

What is the third-variable problem?

An unmeasured variable may explain the relationship between two variables.

Name three things that can affect correlations.

Outliers, restriction of range, and curvilinear relationships.

Why do researchers use multivariate designs?

To control for potential third variables.

What does "controlling for a variable" mean?

Holding a variable constant while examining relationships among other variables.

In multiple regression, what is the dependent variable?

The outcome being predicted.

In multiple regression, what are predictor variables?

Variables used to predict the outcome.

What does beta (β) represent?

The strength and direction of a predictor's relationship with the outcome.

What does a positive beta value indicate?

A positive relationship between predictor and outcome.

What does a negative beta value indicate?

A negative relationship between predictor and outcome.