Bivariate Data Analysis and Quantitative Research Methods

Overview of Course Progress and Context

  • Academic Timeline and Course Status

    • The current session takes place in Week 99 of the semester, leaving exactly 2 weeks2\text{ weeks} remaining in the course (Week 1010 and Week 1111).

    • Assessment 33 is due in Week 1111.

    • Assessment 11 was completed 9 weeks9\text{ weeks} prior, demonstrating rapid progression despite the extensive material covered since the beginning of the term.

  • Personal Context and Health Considerations

    • Illness was widespread among course participants during Weeks 88 and 99.

    • Emily took time off work due to illness, noting a family member (nephew) contracted croup for the fourth time in the current year and was treated with household-managed steroids.

    • Personal health logs indicated a prior flu illness dating back to 20172017, demonstrating that the current seasonal virus strains were unusually strong.

Quantitative Data Preparation and Software Workflow

  • Lecture Recap: Mateo's Excel and Data Structuring Tutorial

    • Opening and downloading raw data files alongside the corresponding Codebook.

    • Cross-referencing raw dataset columns against numeric or alphanumeric codes.

    • Importing target Excel cells into consolidated comparison tables.

    • Converting raw numerical values into percentages.

    • Generating visual charts and figures during the final 15 minutes15\text{ minutes} of the lecture.

    • Prior lecture content focused on structural definitions of independent variables, dependent variables, and covariates.

    • Visual selection options introduced included clustered column charts and stacked column charts to effectively communicate data stories.

  • Student Progress and File Verification

    • Emily downloaded the dataset in Week 88 and submitted Task 22 on Sunday. Initial feelings of being overwhelmed by Excel subsided after reviewing Mateo's lecture, confirming the spreadsheet manipulation was straightforward.

    • Grace verified that survey response variables (individual survey questions) were formatted across individual columns across the spreadsheet rather than compressed into a single column.

    • Abby downloaded both the Excel spreadsheet and the Codebook, successfully running trial analyses by pairing two survey questions into small contingency tables to calculate percentages.

    • Tanisha downloaded both required files (Excel dataset and Codebook).

  • Methodological Purpose and Application of Codebooks

    • Survey questions are often too lengthy to serve directly as column headers in statistical spreadsheets or software programs.

    • A Codebook functions as an essential explanatory reference document detailing the verbatim survey question corresponding to each column variable/code.

    • Accurate cross-referencing between dataset variable codes and Codebook definitions is mandatory to prevent misinterpretation of variables and inaccurate research narratives.

Principles of Quantitative Analysis: Univariate vs. Bivariate

  • Univariate Analysis

    • Evaluates a single variable in isolation (e.g., examining age distributions, political beliefs, or domestic versus international student status).

    • Instructional step-by-step guidance is contained in Mateo's first Excel tutorial video, which runs for 15 minutes15\text{ minutes}.

  • Bivariate Analysis

    • Evaluates two variables simultaneously to examine potential statistical associations, linkages, or directional relationships.

    • Instructional step-by-step guidance is provided in Mateo's second tutorial video, which runs for 20 minutes20\text{ minutes} in total:

    • The first 10 minutes10\text{ minutes} demonstrate table and chart construction for two variables.

    • The remaining 10 minutes10\text{ minutes} demonstrate optional data analysis techniques utilizing ChatGPT.

Case Study: Evaluating Online Hate Reporting Confidence by Age Brackets

  • Research Question Formulation

    • Is confidence in reporting online hate associated with age?

    • Variable 11 (Demographic / Independent): Age brackets (18–29 years18\text{--}29\text{ years}, 30–49 years30\text{--}49\text{ years}, and 50+ years50+\text{ years}).

    • Variable 22 (Psychological / Dependent): Confidence level in reporting online hate (Low Confidence, Moderate Confidence, High Confidence).

  • Empirical Data Distribution

    • Low Confidence:

    • Age 18–29 years18\text{--}29\text{ years}: 35%35\%

    • Age 30–49 years30\text{--}49\text{ years}: 25%25\%

    • Age 50+ years50+\text{ years}: 5%5\%

    • Moderate Confidence:

    • Age 18–29 years18\text{--}29\text{ years}: 40%40\%

    • Age 30–49 years30\text{--}49\text{ years}: 40%40\%

    • Age 50+ years50+\text{ years}: 35%35\%

    • High Confidence:

    • Shifts across age cohorts by 5%5\% to 10%10\% increments.

  • Analytical Interpretations from Seminar Discussion

    • Emily observed that as age increases, general confidence in reporting online hate increases.

    • Grace observed that younger individuals (18–29 years18\text{--}29\text{ years}) represent the smallest proportion within the high confidence category.

    • Abby observed that moderate confidence remains relatively steady across all age cohorts (40%40\%, 40%40\%, and 35%35\%), demonstrating only a 5%5\% variance between groups.

  • Methodological Rules for Presenting Bivariate Findings

    • Precision in Category Framing: Vague terms such as "more confidence" should be replaced with explicit sub-group definitions (e.g., "the high confidence category").

    • Directional Specificity: Researchers must state both association and directionality (e.g., "As age increases, the proportion of respondents reporting high confidence also increases").

    • Contextualizing Literature Counter-Trends: While this dataset reflects increased confidence with age, external academic literature often demonstrates the inverse relationship—that older demographics exhibit lower confidence in reporting online hate due to lower technological literacy.

Selection Criteria for Data Visualizations

  • Comparative Visual Options in Excel

    • Tables: Ideal for displaying exact numerical percentages and specific category counts for direct lookup.

    • Clustered Bar Charts: Useful for side-by-side category comparisons, but can obscure visual trends when percentage differences are minimal (e.g., 5%5\% to 10%10\% shifts).

    • Stacked Bar Charts: Optimal for illustrating cumulative distributions, proportions of a whole, and directional trend shifts across continuous or ordinal demographic groups.

  • Applied Visual Selection Rationale

    • Abby and Emily selected stacked bar charts for bivariate analysis because cumulative bars highlight subtle percentage increases across age cohorts far more effectively than clustered bar charts.

    • Tanisha selected tables to prioritize exact numerical precision across confidence levels without visual interpretation errors.

    • General Guideline: Use raw tables when discrete percentages are required; use stacked or clustered charts when communicating visual patterns, relationships, or structural trends.

Six-Step Workflow Strategy for Assessment 3

  • Assessment Parameters

    • Final Submission Target: Week 11$.\n * Maximum Word Count: 1,600\text{ words}.\n * Analytical Scope: Students must analyze a targeted subset of variables rather than the entire dataset. Comprehensive analysis of the full dataset is beyond the required scope and represents fourth-year or postgraduate level work.\n\n* **Step-by-Step Execution Model**\n * **Step 1: Data Access**: Download and store the Excel dataset and Codebook files.\n * **Step 2:UnivariateVideoReview∗∗:FollowMateo′s: Univariate Video Review**: Follow Mateo's15\text{-minute} tutorial on single-variable table and chart construction in Excel.\n * **Step 3:BivariateVideoReview∗∗:Followthefirst: Bivariate Video Review**: Follow the first10\text{ minutes} of Mateo's bivariate analysis tutorial video.\n * **Step 4: Codebook Review and Topic Framing**: Examine the Codebook to identify broader demographic clusters (e.g., age, education, gender) or specific thematic groups to establish a workable research angle.\n * **Step 5: Exploratory Data Analysis**: Generate trial charts in Excel to identify visually compelling patterns, trends, or unexpected data distributions.\n * **Step 6:NarrativeDrafting∗∗:Writethe: Narrative Drafting**: Write the1,600\text{-word} report, framing clear verbal explanations around selected data visuals.\n\n# Seminar Agenda and Preparation for Week 10\n\n* **Prerequisites for Week 10 Seminar**\n * Download dataset and Codebook files.\n * Complete watching Mateo's univariate video (15\text{ minutes})andbivariatevideo(first) and bivariate video (first10\text{ minutes}).\n * Bring multiple candidate research questions and initial exploratory Excel charts to class.\n\n* **Week 10 Content Structure**\n * Introduction to Computational Social Science methods (standalone topic provided for methodological breadth; non-examinable for Assessment 3).\n * Interactive Assessment Workshop: Group feedback and refinement of student research questions and exploratory data visualizations to finalize report outlines prior to Week 11$.

Questions & Discussion

  • Task 2 vs. Assessment 3 Scope

    • Question: Does the research question for Assessment 33 need to match the research question used in Task 22?

    • Answer: No. Task 22 was a qualitative research assessment that is complete and submitted. Assessment 33 is an independent quantitative research report based entirely on the provided survey dataset.