SSRM FINAL

Passive Observation

Researcher does not intervene, only observes.

Participant Observation

Researcher becomes involved in the setting.

Disguised vs Undisguised Observation

Whether participants know they are being observed.

Structured Observation

A controlled environment with specific behaviours targeted.

Features of Naturalistic Observation

• Data can be qualitative (descriptive) or quantitative (measured).

• Observations occur in real-world settings.

• May use video/audio recordings, field notes, or checklists.

Time Sampling

Observing behaviors at specific time intervals.

Event Sampling

Observing specific behaviors regardless of timing.

Reactivity

Participants may alter behavior if they know they are being observed.

Key Informants

People within the environment who provide insights to researchers.

Advantages & Limitations

• Advantages: High ecological validity, useful for studying complex behaviors.

• Limitations: Observer bias, lack of control, ethical concerns regarding privacy.

Intervals

Equal intervals but no true zero (e.g., temperature in Celsius).

Ratio

Equal intervals with a true zero (e.g., height, weight).

Nominal

Categories without numerical meaning (e.g., gender, political party).

Ordinal

Ordered categories but unequal intervals (e.g., race rankings).

Simple Random Sampling (Randomized)

Every individual has an equal chance.

Stratified Random Sampling (Randomized)

Population divided into subgroups, then randomly sampled.

Cluster Sampling (Randomized)

Groups (clusters) randomly selected instead of individuals.

Convenience Sampling (Non-Randomized)

Selection based on availability.

Purposive Sampling (Non-Randomized)

Selection based on specific characteristics.

Snowball Sampling (Non-Randomized)

Participants recruit others.

Reliability

Consistency

1. Internal Consistency

Whether test items measure the same construct.

2. Test-Retest Reliability

Consistency over time.

3. Interrator Reliability

Agreement between observers.

Numerical Standard for Good Reliability:

Cronbach's alpha > 0.7 for internal consistency.

Construct Validity:

Measures what it intends to measure.

Face Validity

Appears to measure correctly.

Concurrent Validity

Correlates with established measures.

Convergent/Discriminant Validity

Correlates with similar constructs but not unrelated ones.

Internal Validity

Causal relationships; must control for confounds.

Threats

History, maturation, testing effects, regression to the mean, selection bias.

External Validity

Generalizability.

Threats

Artificial settings, sample bias, interaction effects.

Improving External Validity

Using real-world settings, replication, meta-analysis.

Mundane Realism:

How much an experiment resembles real life.

Experimental Realism

How much participants engage with the study.

Face-to-Face Interviews

High response rate, expensive

Telephone Interview

Efficient but lower response rates

Online/Computer-based Interview

Cheap, but response bias possible

Types of interviews

Structured

Set questions, standardized responses.

Semi-Structured

Some set questions, room for flexibility.

Unstructured

Open-ended, Flexible response

Survey designs over time

Panel

Same participants over time

Trend

Different participants, same questions over time

Cohort

Specific subgroups studied over time

Double-Barreled Question

Asking two things at once

Question Format

Close-Ended

Likert, multiple choice, semantic differential scales

Open-Ended

Qualitative Responses

Archival Data

• Includes government records, newspapers, social media.

• Advantages: Cost-effective, access to large datasets.

• Limitations: Missing data, lack of control.

Case Studies

• In-depth analysis of an individual or small group.

• Useful for rare conditions or unique phenomena.

Descriptive Statistics

Summarize data (mean, median, mode, standard deviation).

Inferential Statistics

Draw conclusions (t-tests, chi-square, ANOVA).

Recognizing statistical categories

• If a statistic describes a sample → Descriptive.

• If a statistic makes a prediction about a population → Inferential.