COMM 150 midterm - Peterson

Academic (Scholarly, Basic)Research

Type of research:

- Colleges/ Research Universities
- Quality important (Good and cheap)
- (uses theories)
- Subject to ethical concerns
- IRB

• Academic Research: Conducted by universities or independent researchers, typically peer-reviewed and published in journals. It aims to advance knowledge and is publicly accessible. Example: A university study on climate change published in Nature.

Proprietary (Applied) Research

Type of research:

- Industry
- Speed important (fast)
- Competitive
- Examples: Polling, Sampling, Survey Data

• Focus Groups • Industry Research

- proprietary research isn’t worried about theory. they want quick data

• Proprietary Research: Conducted by private companies or organizations for internal use or profit. It is often confidential and not publicly available. Example: A tech company’s internal study on user behavior to improve its algorithm.

Information vs Data

interpreted data; subject to interpretation

vs

collection/measurement/observation/material/artifact of experience

Factual vs Conceptual data

• Factual Data: Objective, verifiable information based on direct observation, measurement, or records.
  Example: The Eiffel Tower is 330 meters tall.

• Conceptual Data: Abstract, interpretative information based on theories, models, or ideas. Example: Democracy is a system of government based on citizen participation.

The Research Process**

1. Conceptualization** (interest, theory, related research)
2. Planning and designing research
3. Methods for conducting research
4. Analysis and interpreting data
5. Reconceptualization

just blowing smoke article?? often on slides

Positivism

- Independence, adherence to the scientific method, quantitative research

- deductive reasoning (general > specific; theory > specific predictions)

- objective, often written in 3rd person

- independent of what we’re researching. deductive.

- a philosophical system that holds that every rationally justifiable assertion can be scientifically verified or is capable of logical or mathematical proof,

- a doctrine contending that sense perceptions are the only admissible basis of human knowledge and precise thought

• ^Positivism argues that knowledge must come from observable, measurable experiences (our senses) and scientific methods. It rejects speculation or things that can’t be tested.

Naturalism

• qualitative, factual or realistic representation

  • “the practice of describing precisely the actual circumstances of human life”

• inductive reasoning; (specific > general; observation > general explanations)

• subjective, often written in 1st person

• some believe that their interpretation is a PART of their research.

• The word Studies- communicates qualitative research

Reasons for naturalistic research

• study people in natural settings vs a lab

• empower marginalized groups

• capture communication that’s about to die out/go away

• to “gain a better understanding”, interpret, report, describe

The word Studies- communicates ___ research

qualitative

Interpretive is related to

naturalism

culture?? (is in the “words to know” list)

CUBA:

• what might happen to it?

critical theory and critical studies

• give voice to the voiceless

• empower marginalized groups

• illuminate lives outside of the mainstream

• ex. study of street kids in Brazil (SAO PAULO)

3 ways of looking at qualitative research: words communicate methods/perspectives***

Critical Studies
Media Studies
Conversational Analysis

Presidential debate video. Issues?

- Margin of error

- Las Vegas research facility

- How results are described

- Independence of measurement

• Evaluation Apprehension

6 Ways of Knowing

ILAOTM - "I Learn About Our True Mind."

1. Intuition - Common sense is the collection of prejudices acquired by the age of 18
2. Logic - Reasoning, more important than intuition
3. Authority - Parents, teachers, religious figures
4. Observation and personal experience

- bad experience with a food chain dictates how you see the world
5. Tradition, custom, and faith

- we’ve always done it this way
6. Magic, superstition, and mysticism

Research Cultures:

• humanities

• physical sciences (life sciences)

• social sciences (human sciences, behavioral sciences)

  • big overlap with psychology (stem, no longer a social science because there’s a lot more grant money to do quantitative research)

Laws

- Commonly believed fundamental ways that things operate
- Newtonian Laws, Law of Gravity, etc.

Theories

• causal statements between two or more variables → supported by hypothesis testing

• after years of research, and tons of evidence and empirical studies, they support the theory

Hypotheses

Predictions about variables based on observation, logic, AND theoretical underpinnings

Research Questions VS Hypotheses

• Research Questions: Open-ended questions that guide an investigation.

  • Example: How does social media usage affect sleep quality?

• Hypotheses: Testable statements that predict a specific relationship between variables based on prior knowledge or theory.

  • Example: Increased social media usage before bedtime reduces sleep quality.

In short, research questions ask, while hypotheses predict.

Correlation:

A statistical measure that describes the relationship between two variables—how they change together.

It can be:

• positive (both increase or decrease together),

• negative (one increases while the other decreases),

• zero (no relationship).

Correlation is measured with what coefficient?

Pearson’s r: measures the strength and direction of a linear relationship between two numerical variables.

r = 1 → Perfect positive correlation

r = -1 → Perfect negative correlation

r = 0 → No correlation

what is r²?

coefficient of determination:

• how well one variable explain change of the other

• ex. r² = 0.9 means that 90% of the variable A explain the value of variable B, like studying and test performance.

How do we talk about correlation?

- We use words like "increases risk", "co-varies with", "is correlated with", "is associated with", "can be predicted by", "linked to" may", "might", "? (phrasing in terms of a question)"

When do we use correlation?

- All the time
- Extremely useful- they don't require you to randomly assign people to conditions
- All you need is the data to collect

Why is correlation interesting/useful?

- Because it helps you make decisions
- Ex. If chemicals have a chance to be associated with something awful, then you can choose not to do it

- don’t need to perform unethical experiments like assigning people to groups with potentially harmful variables

What are the potential problems with correlation?

- Misinformation
- Tribute causality to correlation

• third variable problem

• ex. hot weather: both ice cream sales and murders go up, but are not related

Why can’t correlation imply causality?

• third variable (ice cream and murders)

• reverse causality - we don’t know which causes which

Causation

A cause and effect relationship in which one variable controls the changes in another variable.

Requires:
- The control of other variables
- Manipulation
- Random assignment to condition

exam: what is a causal statement? what is a correlational statement?

• Causal Statement: Says that one thing directly causes another.

  • Example: Eating more sugar causes weight gain. (Implies a direct effect.)

• Correlational Statement: Says that two things are related but doesn’t prove one causes the other.

  • Example: People who eat more sugar tend to weigh more. (They move together, but other factors might be involved.)

Language used with causation

• Leads to

• Results in

• Affects

• Impacts

• Changes

Theories are built and supported by ___

hypothesis testing

Linked In, Flouride, TikTok, Debate Research, etc. Examples

possible ethics issues with:

• Fluoride in Water

• LinkedIn Study on Weak Ties

• TikTok Addiction

• Facebook Feeds Altered

Was Zimbardo's prison study experimental in nature? Why or why not?

yes, it was experimental because it involved the manipulation of variables and random assignment of participants

What example did the video present that demonstrated how hypothesis testing, through multiple empirical studies, is used to gain a better theoretical understanding of human behavior?

Milgram’s 16 different studies on the same subject (following authority)

What real world questions did zimbardo and milgram tryto answer?

prison abuse, following authority

Zimbardo’s theoretical perspective involved these three fundamental elements:

Dispositional - bad apples,

situational - bad barrel,

systemic - bad barrel makers

What are some ethical concerns with the Zimbardo and Milgram studies?

psychological harm, maybe physical.

people need to live with that guilt for the rest of their life for killing the other person with electric shock.

Zimbardo's Prison Study Ethical Concerns:

• Lack of informed consent (unaware of extreme conditions)

• Psychological harm (emotional distress, abuse)

• Lack of intervention (Zimbardo let abuse continue)

Milgram's Obedience Study Ethical Concerns:

• Deception (participants believed they were shocking real people)

• Emotional distress (many showed extreme anxiety)

• Pressure to continue (felt coerced by authority figure)

How did Zimbardo share his work with the world?

his book “the lucifer effect”, everyday hero courses for kids

Statistical VS Practical significance

• Statistical Significance: The result is unlikely due to chance, but it might not be meaningful in real life.

  • Example: A study finds that a new diet reduces weight by 0.1 lbs—statistically significant but practically useless.

• Practical Significance: The result actually matters in real-world applications, even if it's not statistically strong.

  • ex. A new heart disease drug reduces heart attacks by 15%, but the study only had 50 patients, so p = 0.07 (not statistically significant).
    👉 The effect is still big enough to be life-saving, so doctors might consider using it despite the stats.

Facebook altering feeds - ethical?

not really, but everyone accepted terms of use

Who’s considered vulnerable population?

groups that may have limited ability to give informed consent or are at higher risk of exploitation or harm.

Examples of Vulnerable Populations:

1. Children & Minors (can’t legally consent)

2. Prisoners (coercion risk)

3. Pregnant Women & Fetuses (potential harm)

4. Elderly (cognitive decline, dependency)

5. People with Disabilities (intellectual or physical limitations)

6. Low-Income or Illiterate Individuals (may not fully understand risks)

7. Patients with Mental Illness (decision-making capacity concerns)

8. Undocumented Immigrants & Refugees (fear of legal consequences)

9. Animals?

Internal VS external ethics*

Internal:

• norms of science (share knowledge openly with community, be objective, don’t seek personal gain)

• transparency, honesty, lying slows down the entire research process

• disclose any conflicts of interest

External:

• society,

• protecting others like vulnerable population in studies (subjects/participants),

• outside organizations that impose laws and regulations (IRB-Institutional review board)

Fabrication

making up data or results and recording or reporting them.

Falsification

manipulating research materials, equipment, or processes, or changing or omitting data or results such that the research is not accurately represented in the research record.

Plagiarism

the appropriation of another person's ideas, processes, results, or words without giving credit

citation bias

positive studies are cited 3 times more than negative

Vioxx case

Ethics Concerns

• Suppressed Data: Merck allegedly hid evidence that Vioxx increased the risk of heart attacks and strokes.

• Informed Consent Issues: Patients may not have been fully informed of the risks.

• Conflict of Interest: Some researchers had ties to Merck, raising concerns about bias.

2. Study Significance

• Vioxx was meant to be a safer alternative to NSAIDs (like ibuprofen) for pain relief, but safety concerns over cardiovascular risks outweighed benefits.

3. Key Studies: 090 and VIGOR

• VIGOR Study (2000): Compared Vioxx to naproxen (another painkiller). Found fewer stomach issues but a higher risk of heart attacks in Vioxx users.

• 090 Study: Another trial that showed increased cardiovascular risks, but Merck allegedly downplayed the findings.

4. Replication & Peer-Reviewed Journals

• Studies showing Vioxx’s risks were replicated, confirming the cardiovascular dangers.

• Some peer-reviewed articles were later found to be ghostwritten or influenced by Merck, raising credibility concerns.

5. Role of Drug Reps

• Pharmaceutical sales reps were trained to downplay risks when promoting Vioxx to doctors.

• Internal Merck emails showed a strategy to avoid discussing safety concerns.

6. Placebo/Control Groups

• In trials, Vioxx was compared to naproxen (instead of a placebo), which may have masked cardiovascular risks, since naproxen has heart-protective effects.

7. Independent & Dependent Variables

• Independent Variable (IV): Whether a patient took Vioxx or relafen.

• Dependent Variable (DV): Cardiovascular events (heart attacks, strokes), gastrointestinal issues, and pain relief effectiveness.

What’s a necessity for causality in experimental studies?

A manipulated IV

• ex. i manipulate what type of cox2 inhibitor a subject receives (vioxx, relafen)

• needs different LEVELS

What does “operationalize a DV mean?” Operational definitions

• defining it in a way that can be measured “what is a cardiovascular event?”

  • define love for a study comparing dating apps (tinder/hinge)

• deciding on measurement scale

• They need to be very detailed and specific, such that everyone would agree on what you are defining.

• variables must accurately reflect the abstract concepts they are supposed to measure or manipulate

  • example: If you're studying stress (construct), your DV might be cortisol levels, which is a valid measure of stress

• generally included in the methods section

Significant difference between variables in Vioxx study?

• relafen was sig. different from viox

• placebo was sig. different from viox

• placebo and relafen were not sig. different

Most studies are done on men… how do we compensate for that?

Adding a MEASURED second IV - sex/gender.

What is a measured IV?

Means we split up the data based on demographic (sex, age, ethnicity, hometown) because it can’t be manipulated.

What are marginal means?

Marginal means are the average values of one factor, ignoring other factors. They help summarize data when there are multiple variables involved.

• used to find potential main effects

• M/F x Vioxx/Relafen/Placebo: relafen is better for men, placebo for women

  • ^ interaction or interaction effects

  • bars and lines help identify interactions

Interaction effects

• interaction effects show how two IVs work together in ways that are not just additive.

• whenever you say “it depends”, it’s an interaction

• When the affect of the drugs on cardiac events is dependent on whether the participant is male or female.

• Interaction of drug type (IV1) and gender(IV2) on cardiac events (DV)

Main effects VS Interactional effects

Main: Variation in one variable (regardless of) variation in others

• main effect: looking for 1 iv ONLY and differnces in the level of one regardless of other variables

• JUST 1IV and 1 DV

Interactional: Variation of one variable dependent on variation of others

• interactions: differences in variavles depend on changes in other variables; when we say “people” we don’t care about the gender of the participant

• multiple(!!) IVs in the response

Issues with Pizza experiment?

• Chef John (experimenter) was also a participant

• Independence of measurements

• Measurement Scale unclear

• Order they ate the pizzas

• Maybe the taste is dependent on what pizza they had first! (Internal Validity)

• Then… many limitations (External Validity) • Participants, Pizza Style, Oven, Toppings, Water, etc.

4 measurement scales?

NOIR:

Here’s a simple breakdown of the four scales of measurement:

1. Nominal Scale

• Labels or categories only, no meaningful order.

2. Ordinal Scale

• Definition: natural order but no fixed distance between categories. The order matters, but we don't know how far apart the ranks are. “ranking”

• Example: (very dissatisfied, neutral, very satisfied); Ranking, birth order

3. Interval Scale

• Definition: equal intervals between values, but no true zero point. “rating” something

• Example: The difference between 20°C and 30°C is the same as 30°C and 40°C, but 0°C doesn’t mean "no temperature."; SAT

4. Ratio Scale

• Definition: true zero point, meaning absence of the variable is meaningful.

• Example: A weight of 0 kg means no weight, and a person weighing 20 kg weighs twice as much as a person weighing 10 kg.

Examples of scales with pizza experiment

Nominal: Which Pizza dough is the best?

Ordinal: Rank order the dough.. Best to worst

Interval: How about a rating of deliciousness on 1-10 (10 being the most)

Ratio: How many calories does the dough have?

Conceptualization leads to ___

predictions

Differs by condition means…

the outcome changes depending on the specific situation or treatment applied.

• pizza experiment: If we get differences in deliciousness scores, we cannot attribute them to the water in the dough, because the ambiance, ovens, etc. differ by condition.

• ^This is a clear example of a confounding variable.

Confounding vs Extraneous

• Extraneous variables = noise, add extra variation but don’t directly confuse the IV-DV relationship

  • impacts each condition in the exact same way = HOMOgenuosly

  • doesn’t necessarily mess with your results

• Confounding variables affect the study’s validity by creating a false link between IV and DV.

  • only hits one condition - we can no longer attribute the differences to the water

  • messes with your results HETEROgenously

PETERSON: extraneous - an unwanted variable that doesn’t necesarily mess with your results if it hits variables homogeneously

• extraneous variable becomes a confounding variable when it impacts only one variable. because then you can’t attribute the changes to your manipulated variable.

Types of validity:

IEECC

1. Internal Validity
2. External Validity
3. Ecological validity
4. Construct validity
5. Conceptual validity

Internal Validity*

- Causality
Internal validity is about whether the study proves a cause-and-effect relationship between the independent variable (IV) and the dependent variable (DV), without interference from other factors.
- In the research study done such that the findings are accurate based on the manipulation, subjects, etc.?

External Validity:*

- The generalizability of the study
- How well does your study generalize to other people? Extent to which results provides an accurate description of what typically happens in the real world

Internal VS External validity

• Internal / within study: Ensures that nothing else besides the IV is affecting the DV. are you measuring hat you intend to measure?

• External/outside of study: Ensures that the findings apply beyond the specific study context. generalizability

Ecological validity

- Does it simulate people in their own environments?
- Ex. Zimbardo- picking up subjects in police cars to go to jail

- ex. providing snacks for a tv watching study to make people feel like they are in a living room
- Naturalistic studies high in ecological validity

Construct validity:

how well you operationalize your variables
How well do IV's and DV's represent the construct that you want to study?
Does the study accurately measure the theoretical concept (construct) it claims to?

• Threats: Poorly defined constructs, bad measurement tools, experimenter bias, etc.

Conceptual validity:

it should be consistent within the industry and theory; (water making someone a better driver doesn’t make sense)

The study or the hypotheses naturally follow what's known in the field

How to do factorial design

• potential 3 main effects if there’s 3 IVs

• 4 potential interactions: 3 2-way interaction, 1 3-way interaction

Main effect or interaction?
H1: People will disclose more to a peer than a younger person -

main effect

Main effect or interaction?
H2: Women will disclose more information to a doctor that will a man

interaction:
participant gender vs authority figure

Main Effect or interaction?
H3: A patient will disclose more to a female dr than they will to a male dr but will disclose less info to a female peer thana male peer.

• 2-way interaction without the gender of the participant

Main effect or interaction?
H5: For male Drs, females disclose less than males.

3 way interaction- we are talking about all 3 IVs.

• authority gender, participant gender, authority figure

reliability

The consistency or repeatability of the study.
Would you get the same results if you conducted the experiment again?

Internal consistency

- Different questions that test the same construct should give consistent results.

• measures how well different items in a test or survey measure the same concept.

• A test with high internal consistency is more trustworthy because all parts of it are working together to measure the same concept.

Temporal consistency

- evaluates reliability across time

- Test-retest: Same people, different times.

Threats to Validity

Possible problems in a study's design

1. Passage of Time ⌛⌛

2. Statistical, Nature of Data 📊📊

3. Mortality ☠☠

4. Hawthorne Effect 🔍🔍

5. Sleeper Effect 👵🏻

6. Participant Expectancies 😇😇👿👿

7. Researcher Personal Attribute Effect 🥸🥸

8. Researcher Unintentional Expectancy Effect 😏😏

9. Researcher Observational Biases 🥱🥱

10. Selection bias: 🧑‍🤝‍🧑🧑‍🤝‍🧑

11. Non-response bias

12. Self-selection bias

2. Statistical, Nature of Data

- regression toward the mean

- ceiling effect/floor effect (test is too easy or too hard) - so all groups score the same

• the tendency for extreme or unusual scores to fall back (regress) toward their average.

Mortality

- people leave the study in the middle (the cockroach dies in the middle of the study)

- homogeneous attrition (people leave approximately the same amount over all conditions)

- heterogeneous attrition/confounding variable

- people leave more in one condition than others - potential confounding variable

Hawthorne Effect

- the idea that people perform better when they are being observed.

Sleeper Effect

- the idea that some changes may occur over long periods of time (after the study is done)

participant expectancies

• demand characteristics of the study

  • participants guess the nature of the study and want to help the researcher get the appropriate results 😇😇🔫🔫

• participant reactance

  • want to discomfirm, jack iup the study👿👿

• evaluation apprehension

  • they don’t want to appear in a certain way

  • cover stories or unobtrusive observations

  • “who is here a cheater?”

Researcher Personal Attribute Effect

- a threat to validity
- the idea that researchers may elicit different responses when they act friendly, angry, mustached, ethnicity, etc.

Researcher Unintentional Expectancy Effect

- body language- smile when subjects confirm their hypotheses

researcher observational biases

when a researcher's expectations, beliefs, or prior knowledge affect how they observe, record, or interpret data in a study.

- observer drift: gets bored or tired

- observer bias: see what we wanna see

- halo effect: researcher gives a better rating based on previous performance

ways to avoid experimenter bias

• double blind procedures: experimenters keep participants’ assignment to condition a secret

• multiple observers, randomly assigned to condition

• using other tools and tech (videos, recording, etc.) - to make sure that everything you do is the same BUT your manipulated IV

selection bias

randomization of the selection of individuals is not achieved

Non-response bias

Bias introduced into survey results because individuals refuse to participate.

Self-selection bias

a type of bias that can arise when study participants choose their own treatment conditions, rather than being randomly assigned.

r* The earliest explanations for human behavior (and for the physical world as well) appear to have been ___

metaphysical

r*___, the belief that natural phenomena are alive and influence behavior

animism

r* This second category of metaphysical explanations includes___ .

mythology and religion

r* A third very old category of metaphysical systems is ___

astrology

r* One of the earliest systems of thought to compete with metaphysical systems was ___

philosophy

r * Psychology probably owes its current emphasis on systematic observation to its roots in the physical sciences, especially___

physiology

r* Fundamental principles that are more or less accepted on faith are often referred to as ___

canons

r* This is the doctrine that the universe is orderly—the idea that all events have meaningful, systematic causes.

determinism.

r* artifact vs confounding variable

• A confounding variable hides the true effect by influencing both IV and DV.

• An artifact makes results only true under specific conditions, so they might not apply to the real world.