study guide research methods.docx

Concepts for Final Exam

(Max. 75 points)

08/28:

• Controversial research due to ethical considerations

The Tuskegee experiment - shut down in the 70s 40-year study – under the guise of free medical care Studied over 600 black men - secretly studying long-term progression of syphilis (among about 300) Even though they knew penicillin would be an effective cure – they administered other/placebo drugs to observe long-term effects

o Henrietta Lacks - In 1951, a tobacco farmer – Johns Hopkins Hospital (MD)

Cells (two samples) from her tumor collected – and studied as the HeLa

line (survived outside her body) Lending itself to AIDS and cancer research, the polio vaccine, also the first cells to be able to be cloned

o Milgram’s study - Participants were told they were going to look

at the effects of punishment on learning • Experimenter hooked learner up to electrodes. Told learner would be shocked when they got a wrong answer. • Participant was given a

box to enable weak to intense shocks.

o Triplets study - Triplets separated at birth • Grew up in three

different households • Nature vs Nurture?• Impact of parenting? • Found out in the 1980s • Results never published

o The Facebook contagion study- Manipulating feed • Positive/negative stories • Emotional contagion

• The difference between consent and debriefing, their purpose, and when they occur in research

Protect your participants – anonymity (if needed)

Always get consent!

- Informed consent

Debriefing your participants

08/30, 11/11, 11/13, 11/15:

• Secondary research – What is it? What is the process? Why do we need it? When do we use it?

secondary research is another persons research they did on their own from coming up with a research question

What is its relationship to primary research?

Primary can bounce off of it

• What qualifies as credible research?

o What is peer-review? And why should you care whether a research article was peer-

reviewed or not?

• What is a research question (RQ)? What is its significance?

• What are hypotheses? When do you use them instead of RQs?

• Role of AI in secondary research – when to use it and when not to use it

09/04:

• What is validity in a research study? What is reliability in a research study?

VALIDITY

Is about making sure we

are accurately measuring

or assessing

[ACCURACY]

RELIABILITY

Is about making sure we do

that consistently

[CONSISTENCY]

• What are interviews? When to use them?

Interviews are one on one questionaries to can deeper messages to collect data

• Types of interviews – semi-structured & structured (Which is preferred? Why?)

Semi-structured to dig deeper

09/04, 09/06, 09/09:

Sample and Population

• Sample: A subset of a population used to represent the whole.
• Relation to Population: Samples help infer population characteristics when studying the entire group is impractical.
• Why/When Needed: Saves time and resources; used when populations are too large or inaccessible.

Sampling Criteria

• Key Factors: Representativeness, size, randomness, and feasibility.

Types of Sampling

1. Random Sampling: Every population member has an equal chance of selection; reduces bias.
2. Convenience Sampling: Based on ease of access; faster but may introduce bias.

Saturation in Interview Data

• Definition: No new themes emerge from additional data collection.
• Sample Size: Determined by when saturation is reached in qualitative research.

Analyzing Interview Data

1. Transcribe and organize data.
2. Code and identify themes.
3. Interpret findings in relation to research questions.

Theoretical Sampling

• Definition: Sampling guided by emerging findings during data collection.
• Relation to Saturation: Stops when no new insights emerge, ensuring thorough exploration.

09/18:

• “Proof” or “Prove” in research

• What are focus groups?

• History of focus groups

o Lazarsfeld & Merton

o The American Soldier example

• Who is a moderator?

• The three types of moderator techniques: Projective; Probing; Control

• Data analysis techniques: Constant comparison analysis; Keywords-in-context; Micro-

interlocuter analysis

“Proof” or “Prove” in Research

• Caution: The terms "proof" or "prove" are rarely used in research because scientific knowledge is provisional and evolves over time.
• Preferred Language: Researchers typically use terms like "evidence supports," "indicates," or "suggests" to describe findings.

Focus Groups

• Definition: A qualitative research method involving a small, diverse group of participants discussing a specific topic, guided by a moderator.
• Purpose: To explore attitudes, opinions, and behaviors in a group dynamic.

History of Focus Groups

1. Lazarsfeld & Merton:
   • Paul Lazarsfeld and Robert Merton popularized focus groups in the 1940s to study audience reactions to radio and propaganda.
   • Pioneered the structured but flexible group discussion format.
2. The American Soldier Example:
   • Used focus groups during WWII to assess soldiers' morale and attitudes.
   • Helped develop practical applications for military and marketing research.

Who is a Moderator?

• Definition: A trained individual who leads and facilitates a focus group discussion.
• Responsibilities:
  • Encourages participation while keeping the discussion on track.
  • Ensures a comfortable environment for sharing.
  • Balances group dynamics to avoid dominance or silence.

Three Moderator Techniques

1. Projective Techniques:
   • Indirect methods like storytelling, role-playing, or drawing to uncover hidden thoughts or feelings.
   • Example: Asking participants to describe a brand as if it were a person.
2. Probing Techniques:
   • Follow-up questions to delve deeper into participants' responses.
   • Example: "Can you explain why you feel that way?"
3. Control Techniques:
   • Strategies to manage the discussion, maintain focus, and avoid tangents.
   • Example: Politely redirecting dominant participants to ensure everyone contributes.

Data Analysis Techniques

1. Constant Comparison Analysis:
   • Systematically compares themes, codes, or concepts across data to identify patterns and relationships.
2. Keywords-in-Context (KWIC):
   • Examines the context around frequently used words or phrases to understand their meaning.
3. Micro-Interlocutor Analysis:
   • Analyzes interactions between participants, such as tone, body language, or interruptions, to understand group dynamics and underlying meanings.

09/20:

• What are surveys? When to use them?

• What is a variable?

• Types of variables:

o Categorical and continuous

o Independent and dependent

• When to use what kinds of variables?

Surveys

• Definition: Surveys are research tools used to collect data from individuals through structured questionnaires or interviews.
• When to Use:
  • To gather large-scale data on opinions, behaviors, or characteristics.
  • When standardization is important for comparison.
  • For descriptive, exploratory, or explanatory research.

Variable

• Definition: A variable is any characteristic, factor, or attribute that can take on different values across individuals or groups.
• Example: Age, gender, income, or test scores.

Types of Variables

1. Categorical Variables (Qualitative):
   • Represent discrete groups or categories.
   • Examples: Gender (male, female), ethnicity, education level.
   • Subtypes:
     • Nominal: No inherent order (e.g., colors: red, blue).
     • Ordinal: Ordered categories (e.g., satisfaction levels: low, medium, high).
2. Continuous Variables (Quantitative):
   • Represent measurable quantities that can take on a range of values.
   • Examples: Height, weight, income, test scores.
3. Independent Variables:
   • The variable manipulated or categorized to observe its effect.
   • Example: Type of teaching method in an educational study.
4. Dependent Variables:
   • The outcome or effect being measured.
   • Example: Student test scores resulting from different teaching methods.

When to Use Different Kinds of Variables

• Categorical Variables: When analyzing groups or categories (e.g., demographic studies, frequency counts).
• Continuous Variables: When measuring trends, correlations, or relationships involving numeric data.
• Independent Variables: To test the impact or influence on outcomes.
• Dependent Variables: To observe and measure the effects of changes in independent variables.

09/23, 09/27, 09/30, 10/02:

• Programming a survey on Qualtrics

o What purpose do blocks serve when programming a survey into Qualtrics

o What is the purpose of attention checks?

§ Where is the best placement for these checks?

o What is the purpose of adding collaborators to the survey on Qualtrics

o Why is it important to label your question numbers with variable names in your Qualtrics

survey?

09/25:

• Inclusivity in surveys

• Gender measurement and consideration of intersectionality

Inclusivity in Surveys

Inclusivity ensures that surveys are designed to capture the diverse perspectives and experiences of all participants, fostering fairness and reducing bias in results. Key practices include:

1. Language: Use clear, respectful, and neutral language. Avoid jargon or terminology that may alienate participants.
2. Accessibility: Design surveys to accommodate people with disabilities (e.g., screen-reader compatibility, large fonts).
3. Representation: Include diverse response options that reflect participants' identities and experiences.
4. Sampling: Ensure the sample includes people from various backgrounds and demographics to avoid underrepresentation.

Gender Measurement

• Challenges: Gender is complex, non-binary, and often intersects with other identities like race, culture, or socioeconomic status.
• Best Practices:
  1. Avoid binary options ("male" and "female")—provide inclusive choices like "non-binary," "genderqueer," or "prefer not to say."
  2. Allow open-ended responses when feasible.
  3. Clearly explain why gender data is collected to build trust and transparency.

Intersectionality in Surveys

• Definition: Intersectionality acknowledges how overlapping identities (e.g., race, gender, class, sexuality) shape individual experiences.
• Considerations:
  1. Include multiple demographic questions to capture the complexity of participants' identities.
  2. Analyze data in subgroups to identify how intersecting identities influence outcomes.
  3. Avoid tokenism—ensure questions reflect genuine interest in participants' lived experiences, not just compliance.

10/07, 10/21:

• What are type I and type II errors?

o How to avoid type I and type II errors?

• What does a p-value mean in survey analysis?

Type I and Type II Errors

• Type I Error (False Positive): Rejecting a true null hypothesis. It means concluding there is an effect or difference when none exists.
  • Example: A medical test incorrectly indicates a person has a disease.
• Type II Error (False Negative): Failing to reject a false null hypothesis. It means missing an actual effect or difference.
  • Example: A medical test fails to detect a disease that is present.

How to Avoid Type I and Type II Errors

1. Type I Error:
   • Use a smaller significance level (α\alphaα), like 0.01 instead of 0.05.
   • Perform multiple testing corrections if many tests are conducted.
2. Type II Error:
   • Increase sample size to enhance statistical power.
   • Choose an appropriate significance level and effect size.
   • Ensure proper study design and sufficient variability in the data.

P-Value in Survey Analysis

• Definition: The p-value is the probability of observing results as extreme as the data, assuming the null hypothesis is true.
• Interpretation:
  • A small p-value (e.g., <0.05< 0.05<0.05) suggests evidence against the null hypothesis.
  • A large p-value (e.g., >0.05> 0.05>0.05) suggests insufficient evidence to reject the null hypothesis.
• Key Point: A p-value does not measure the effect size or prove causality; it only assesses statistical significance.

Would you like a more detailed example or explanation for any of these concepts?

10/14, 10/16:

• In R: What are packages? What is a library?

• What does a correlation test tell us? What are the two types of correlation?

• When to use a t-test?

• Understanding the results of a correlation test or a t-test.

Step 1: Check the p-value

- You only say it is significant correlation or a

significant difference if the p-value is less than

0.05.

For a correlation:

Step 2: Check cor value only if p-value is less

than 0.05.

And then check to see if the cor value is ‘+’ or

‘–’ and if it is ‘+’ you can say it is a significant

positive correlation and if is ‘–’ then you can

say it a significant negative correlation.

For a t-test:

Step 3: Check the mean values only if p-value

is less than 0.05 and report that there is

significant difference and state which is one is

greater/lesser than the other.

In R: Packages and Libraries

• Packages:
  • Collections of functions, data, and code developed to extend R’s capabilities.
  • Examples: ggplot2 for visualization, dplyr for data manipulation.
  • Installed using install.packages("package_name").
• Library:
  • Refers to the location where installed packages are stored.
  • Activating a package for use in a session is done with library(package_name).

Correlation Test

• Purpose: Measures the strength and direction of the relationship between two variables.
• Types of Correlation:
  • Pearson’s Correlation: For linear relationships between continuous variables.
  • Spearman’s Correlation: For non-linear relationships or ranked/ordinal data.
• Results Interpretation:
  • Correlation Coefficient (rrr): Ranges from -1 (strong negative) to +1 (strong positive).
  • P-Value: Indicates if the correlation is statistically significant (p<0.05p < 0.05p<0.05).

When to Use a T-Test

• Purpose: Compares the means of two groups to determine if they are significantly different.
• Types of T-Test:
  • Independent T-Test: For comparing two unrelated groups (e.g., male vs. female).
  • Paired T-Test: For comparing two related groups (e.g., before-and-after measurements).
  • Assumptions: Normal distribution, independent observations, and equal variances (in some cases).

Understanding Correlation Test Results

• Correlation Coefficient (rrr):
  • Close to 0: Weak or no relationship.
  • Positive (r>0r > 0r>0): Variables increase together.
  • Negative (r<0r < 0r<0): One variable increases as the other decreases.
• P-Value:
  • p<0.05p < 0.05p<0.05: Significant correlation.
  • p>0.05p > 0.05p>0.05: Insufficient evidence for a correlation.

Understanding T-Test Results

• T-Statistic: Measures the size of the difference relative to the variation in the sample data.
• P-Value:
  • p<0.05p < 0.05p<0.05: Significant difference between group means.
  • p>0.05p > 0.05p>0.05: No significant difference.

Would you like guidance on running these tests in R or interpreting specific output?

10/23:

• When to conduct experiments? What method can establish causality?

• The three criteria for causation

• What is internal and external validity?

• Differences and preference between pretests & posttests vs posttest-only design

• What is a control condition?

• Triangulation

When to Conduct Experiments

• Purpose: Conduct experiments when you need to test cause-and-effect relationships between variables.
• Establishing Causality: Only experimental methods with random assignment can establish causality by isolating the effect of the independent variable on the dependent variable.

Three Criteria for Causation

1. Correlation: The variables must be associated.
2. Temporal Precedence: The cause must precede the effect.
3. No Confounding Variables: Alternative explanations must be ruled out, often achieved with random assignment and controls.

Internal and External Validity

• Internal Validity:
  • Refers to the extent to which an experiment establishes a causal relationship.
  • Threatened by biases, confounding variables, or poor design.
• External Validity:
  • Refers to the generalizability of findings to real-world settings.
  • Threatened by artificial settings, non-representative samples, or unrealistic conditions.

Pretest-Posttest vs. Posttest-Only Design

1. Pretest-Posttest Design:
   • Measures the dependent variable before and after the treatment.
   • Pros: Allows assessment of change and ensures groups are equivalent before treatment.
   • Cons: May introduce testing effects (participants alter responses based on the pretest).
2. Posttest-Only Design:
   • Measures the dependent variable only after the treatment.
   • Pros: Avoids testing effects; simpler.
   • Cons: Assumes random assignment ensures group equivalence.

Preference: Pretest-posttest is preferred when baseline measures are critical, while posttest-only works well with robust randomization.

Control Condition

• Definition: A group in an experiment that does not receive the treatment or independent variable manipulation.
• Purpose: Provides a baseline to compare the treatment effects and rule out confounding factors.

Triangulation

• Definition: The use of multiple methods, data sources, or theories to cross-check and validate findings.
• Types:
  1. Methodological: Combining qualitative and quantitative approaches.
  2. Data: Using data from diverse sources.
  3. Theoretical: Applying different theoretical perspectives.
  4. Investigator: Involving multiple researchers to reduce bias.

10/28 - 11/06:

• What is content analysis? When does it make sense of use it?

• Types of data and sources for content analysis

• Types of content – manifest and latent

• Unit of analysis

• Role of coders

• What is a codebook and a codesheet/codefile?

Content Analysis

• Definition: A systematic research method for analyzing and interpreting the content of textual, visual, or audio data to identify patterns, themes, or meanings.
• When to Use:
  • To analyze communication (e.g., media, documents, speeches).
  • To study trends, cultural patterns, or recurring themes.
  • When working with qualitative or unstructured data that needs to be quantified.

Types of Data and Sources for Content Analysis

1. Data Types:
   • Textual (e.g., books, articles, transcripts).
   • Visual (e.g., photographs, videos, advertisements).
   • Audio (e.g., interviews, podcasts).
2. Sources:
   • Media: News articles, social media posts, advertisements.
   • Organizational Documents: Reports, meeting notes, policies.
   • Cultural Artifacts: Movies, songs, art.

Types of Content

1. Manifest Content:
   • Observable, explicit content (e.g., the frequency of a word in a text).
   • Easier to identify and measure.
2. Latent Content:
   • Underlying, implicit meanings (e.g., the tone, emotion, or themes in a text).
   • Requires interpretation and context awareness.

Unit of Analysis

• Definition: The specific element of the content being analyzed.
• Examples:
  • Words, phrases, or sentences.
  • Themes or concepts.
  • Entire documents or sections (e.g., articles, chapters).

Role of Coders

• Definition: Coders analyze the content and assign codes to data based on predefined criteria.
• Responsibilities:
  • Follow coding guidelines consistently.
  • Maintain reliability and minimize bias.
  • Work collaboratively to resolve discrepancies (if multiple coders are involved).

Codebook and Codesheet/Codefile

1. Codebook:
   • A document detailing the coding system, including codes, definitions, and rules.
   • Ensures consistency and replicability.
2. Codesheet/Codefile:
   • A record where coded data is stored.
   • Can be a physical sheet or a digital file used for analysis.
   • Example: A spreadsheet where rows represent data units and columns represent codes.

Other:

• Why does PR need research?

• Role of AI in PR research

Why Does PR Need Research?

1. Informed Decision-Making:
   • Research identifies target audiences, their preferences, and communication channels.
   • Helps craft effective messages and strategies.
2. Measuring Effectiveness:
   • Evaluates campaign success through metrics like brand awareness, audience engagement, and sentiment analysis.
3. Crisis Management:
   • Assesses public perception and sentiment during crises to guide response strategies.
4. Staying Competitive:
   • Tracks industry trends and competitor activities to identify opportunities and threats.
5. Building Credibility:
   • Research-backed claims enhance trust and reputation.

Role of AI in PR Research

1. Data Collection and Analysis:
   • AI tools gather data from social media, news, and other platforms.
   • Perform sentiment analysis, trend detection, and audience segmentation efficiently.
2. Real-Time Monitoring:
   • Tracks brand mentions, public sentiment, and crisis indicators in real time.
3. Predictive Analytics:
   • Forecasts trends and outcomes, helping PR teams anticipate public reactions.
4. Content Personalization:
   • Analyzes audience preferences to tailor messages for better engagement.
5. Automating Routine Tasks:
   • Streamlines reporting, media analysis, and data visualization, freeing up time for strategic planning.