COMM 88 Midterm

Four Principles of Research

Empirical

Systematic

Intersubjective

Cyclical and Self Correcting

Empirical Research

Based on direct observation or measurement

Systematic Research

Follows a set of rules and procedures

Ensures data can be properly gathered and analyzed

Intersubjective Research

Knowledge is built collectively

Clear definitions and shared understanding

Cyclical and self-correcting

Research builds on previous research

Mistakes are identified and corrected over time (initial study, refinement, replication, advancement)

Unit of Analysis

• Describes the specific type of individual or thing being studied

• Who or what are you actually studying?

• Examples

  • Individuals (students, voters, social media users)

  • Messages (tweets, news articles, advertisements)

  • Organizations (companies, schools, media outlets)

  • Relationships (couples, friendships, teams)

Independent Variable (IV)

• The cause or predictor

• What you manipulate or categorize to observe its effect 

Dependent Variable (DV)

• The effect or outcome

• What you measure to see if it changes due to the independent variable

Third Variable

• Control: remove unwanted influences

• Mediating: explains how X affects Y

• Moderating: explains when X affects Y

Relationships Among Variables (Directionality)

• Positive relationships: both variables increase

• Negative relationships: one increases, the other decreases

Causality

• Variables must be related (X correlated with Y)

  • Ex. time studying increases, GPA increases

• Must establish time order (IV happened before DV or cause precedes effect)

• Must rule out other explanations/causes (think control variables)

Deductive Reasoning

• theory -> hypothesis -> test -> confirm/disconfirm

• Clear predictions

• Tests established theories

• Replicable

Inductive Reasoning

• observations -> patterns -> theory

• Discovers new phenomena

• Flexible

• Rich, detailed understanding 

Abductive Reasoning

• surprising observation -> best possible explanation → test that explanation

• Handles unexpected results

• Creative problem solving

• Generates new hypotheses

Why do research ethics matter?

• Research affects real people’s lives

• Digital age creates new ethical challenges

• Professional responsibility and credibility

• Legal requirements (IRB approval)

Three Pillars of The Belmont Report

Respect for Persons

Beneficence

Justice

Respect for Persons (Belmont Report)

• Core idea: people are autonomous agents who can make their own decisions (and protect those who can’t)

• Key components

  • Informed consent: participants understand what they’re agreeing to

  • Right to withdraw: extra care for vulnerable populations (minors, prisoners, etc.)

Beneficence (Belmont Report)

• Core idea: maximize benefits, minimize harms

• Key components

  • Do good: research should help people/society

  • Do no harm: reduce risks to participants

  • Risk-benefit analysis: benefits must outweigh potential costs

Justice (Belmont Report)

• Core idea: fair distribution of research benefits and burdens

• Key components: 

  • Fair selection: don’t exploit vulnerable groups

  • Equitable distribution: benefits should reach those who bore the risks

  • Access: research findings should help the communities studied

Two Key Aspects of Privacy Protection

Anonymity: the researcher cannot connect responses to specific participants

Confidentiality: the researcher knows who participated but protects their data and identity

Debrief

• Full explanation of true purpose and any deception used

• Help participants process their experience

• Restore well-being and provide resources if needed

• Key point: deception doesn’t excuse harm, benefits must still outweigh risks

Levels of Measurement

Nominal: named variables

Ordinal: named and ordered variables

Interval: named + ordered + proportionate interval between variables

Ratio: named + ordered + proportionate interval between variables + can accommodate absolute zero

Types of Items

Likert Type Items: Measures degree of agreement (strongly disagree to strongly agree)

Semantic Differential Items: Captures the emotional or connotative meaning of a concept

Conceptual Definition

• Abstract description of the concept’s meaning

• Easily understood

• Connect to existing research and theory

• Guide what you should measure

Uses and Gratifications Theory

• Media consumption is an active process through which users seek gratifications such as

  • Acquiring information

  • Entertainment

  • Social interaction

  • Reinforcement of personal identity

Reliability and Validity

Measurement Quality Check

Reliability: Consistency

Validity: Accuracy

Forms of Validity Assessment

• Face validity: does it look like it measures the concept?

• Content validity: Does it cover all aspects of the concept?

• Construct validity: does it relate to other measures as expected?

• Criterion validity: Does it predict relevant outcomes?

The Reliability-Validity Relationship

• A reliable measure may or may not be valid

• An unreliable measure cannot be valid

• Reliability is a necessary but not sufficient condition for validity

Systematic Error

• Consistent deviation from the population

• Caused by flawed sampling methods

• Doesn’t decrease with larger samples

• Cannot be fixed with statistics

Random Error

• Differences between the sample and the population due to chance

• Random, not systematic

• Decreases with a larger sample size

• Can be estimated statistically (confidence interval or margin of error)

Theory

• An attempt to explain some aspect of social life

• A systematic explanation of how and why things work the way they do

Probability Sampling

• Sample should be a miniature version of the target population

• Allows you to generalize results to that population

• The Gold Standard: everyone in population has a known (and equal) chance of being included in sample - random selection

Simple Random Sampling

Every individual in the population has an equal chance of being selected

Select elements randomly from sampling frame one at a time and independently

• High representative - no systematic bias

• Each selection is independent of other selections

• Easiest for statistical inference

Systematic Sampling

• From a list of the population, select every “kth” element

  • Formula: k = population size/desired sample size

• Easier to implement than simple random sampling

• Ensures spread across the list

• Works well for large lists

Stratified Sampling

• 1. Divide the population into subsets (strata) of a particular variable

  • Usually stratify for demographic variables (sex, race, political party)

• 2. Select randomly from each strata to get the right proportions of the population

• 3. Use random sampling within each stratum

• Ensures accurate representation

• Reduces sampling error & increases repetitiveness

Cluster Sampling

•  select groups (clusters) first, then individuals within clusters

  • 1. Identify clusters

  • 2. Randomly select clusters

  • 3. Randomly sample clusters

• More cost effective

• Logistically simpler

• Good for interviews/observations

Ecological Fallacy

• Occurs when relationships between properties of groups (cities, countries, classrooms) are used to make inferences about individuals within those groups

Sampling Bias

• Systematically over- or under-representing certain segments of the population

Convenience Sampling

select participants who are easily accessible

Deliberate Sampling

• How it works: intentionally target specific types of individuals

• Depends on knowledge and judgment of researcher

• May provide broadly representative sample

• Types

  • Homogenous: similar participants to understand specific experience

  • Maximum variation: diverse participants to understand the range of experiences

  • Expert: people with specialized knowledge

Quota Sampling

• How it works: set targets for different types of participants, but don’t use random selection

• Identify characteristics that are relevant to the study being conducted

• Ensure representative numbers of people whose absence would distort the results

Snowball Sampling

• How it works: participants recruit other participants from their network

• When to use

  • Hidden or hard-to-reach populations

  • Studying social networks

  • Trust is important for participation

Theoretical Sampling

Observations selected to “elaborate or refine categories in an emerging theory”

Types of Validity

Internal

External

Ecological

Measurement

Construct

Content

Criterion

Internal Validity

Confidence that the relationships observed in a study are accurate and genuine (they truly exist)

External Validity

• Addresses whether the results of a study can be generalized to a broader population

Ecological Validity

• Subtype of external validity

• How closely the study setting, materials, and procedures match real-world conditions

Measurement Validity

• Extent to which a measure accurately reflects the concept it is intended

• Helps achieve strong internal validity to assess

Construct Validity

• Evaluates whether a measurement tool truly captures the theoretical construct it claims to measure

• It is assessed by examining how the measure relates to other variables as expected by theory

Convergent Construct Validity

• Demonstrates when a measure is highly correlated with other measures that assess the same or similar constructs

Divergent Construct Validity

• Divergent validity is established when a measure does not correlate with measures of unrelated constructs

Content Validity

• Whether the measure covers all aspects of the concept being studied

• Careful match between the elements of the construct and what is included in the measure

Criterion Validity

• Assesses whether a measure is related to an outcome or an established benchmark (the criterion) that it should theoretically be related to

• The criterion can be measured at the same time (concurrent validity) or in the future (predictive validity) 

Between Subjects Design

Participants assigned to one condition, and comparisons made between the groups 

• Studies where seeing one condition ruins the other conditions

• When you want to avoid practice fatigue effects

• When conditions are very different

Within Subjects Design

• all participants exposed to every condition; compare each participant to themselves

• When you need more statistical power

• When individual differences are large

• When conditions are similar enough to compare

Factorial Designs

• Multiple IVs manipulated simultaneously (2x2)

• Can test main effects and interaction effects

Threats to Validity: observer effect

• The presence of an observer can influence participants’ behavior

Threats to Validity: History Effect

• Historical events during a study can affect participants' behavior and skew results

Threats to Validity: Interparticipant Bias

• Selection bias

  • Occurs when certain groups are over or underrepresented in a study

• Researcher Bias

  • Occurs when the researcher’s expectations influence the study’s outcomes

• Sensitization

  • Refers to when participants’ awareness of a study’s purpose influences their behavior