Final Project Proposal

Data Quality Assurance + Visualization

Published

March 12, 2026

PSYC 859 Spring 2026

Due dates

  • Proposal: 3/26/2026
  • Final product: 4/27/2026

Proposal guidelines

Please submit a brief proposal (4–6 pages) describing a final data quality assurance (QA) and visualization project. Ideally, this project should extend from the data QA project earlier in the semester and focus especially on visualization and substantive insights into the data.

The project should showcase what you have learned about the principles of effective visualization, such as Tufte’s principles of analytical design, Cleveland’s elements of effective graphics, and Bertin’s three stages of reading and three objectives of graphs.

Finally, the project is an opportunity to highlight your critical thinking about figure design by providing a written rationale for why you chose to approach a visualization in particular ways to maximize its communication to the intended audience (you, your lab, a conference, social media, etc.).

Key ingredients of the proposal

  1. Background and rationale for project (1–2 paragraphs).

    • Feel free to adapt the data QA description, tilting toward the substantive questions that can be answered (or at least interrogated) through graphical and nongraphical EDA.
    • Focus especially on the EDA pipeline discussed in class, including formulating research questions, validating your dataset, detecting anomalies, developing parsimonious models, and assessing relationships among variables.
    • What are your hypotheses?

  2. Description of data structure: only necessary if the final project departs substantially from the data QA project. See the data QA assignment for details if you wish to include this section.

  3. Proposed file system structure for project: only necessary if the final project departs from the structure proposed in the data QA project. See the data QA assignment for details if you wish to include this section.

  4. Data quality assurance plan: this should extend from the data QA project, incorporating the feedback on that project and now considering how EDA (especially graphical) techniques can help identify problems in the dataset (e.g., outliers, unexpected relationships that suggest incorrect coding of variables).

    1. What are the potential sources of artifacts or errors in the data?

    2. What checks can be conducted to mitigate these problems using algorithmic methods (i.e., automated code that walks through a dataset diagnosing and reporting problems)?

    3. How will missing data be identified and resolved? Resolution of missing values includes:

      1. When possible, finding the missing data and entering it.

      2. Documenting the missingness and, when possible, having a mechanism for quieting your QA script so that it doesn’t perpetually yell at you.

    4. Are there QA steps that can only be conducted by a trained human (e.g., ECG artifact correction)? If so, briefly document why these cannot be automated/codified.

    5. What graphical methods can you employ that would help identify artifacts or problematic data points?

  5. Data QA + visualization pipeline proposal: Building on the above, describe key aspects of the data processing and visualization pipeline for the project. What graphical idioms do you anticipate employing for each substantive question? Document key anticipated steps in your pipeline, including:

    1. Import of original data

    2. Basic tidying of data, if necessary

    3. Basic QA checks on each dataset. What are the expected outputs that will document data quality?

    4. Dataset merging, including fidelity checks on merge

    5. Data manipulation of merged data for analysis and visualization

    6. Visualization of data quality, such as missingness by variable, univariate densities, bivariate scatterplots, or time series/index plots that would provide information about accuracy of the data

    7. Exploratory visualization to develop an understanding of relationships among key variables in the dataset

    8. Graphical exploration of your hypotheses about the data. What sorts of plots would allow you to identify the evidence in favor of your hypotheses?

  6. What functions or specialized visualization tools do you anticipate developing or using to accomplish this pipeline? Document any crucial project objectives that will require a custom analysis, visualization, algorithm, or function.

Preview of final project expectations

As a preview of the final project, anticipate that you will need to:

  1. Visualize the same data in at least three (somewhat) different ways.
  2. Create 10+ EDA graphics.
  3. Demonstrate the use of an exploratory analysis, such as principal component analysis, multidimensional scaling, or clustering.
  4. Create at least 2 graphics for others (publication quality).
  5. Create at least one plot with multiple layers that result either from distinct datasets, or from different levels of aggregation (e.g., basic geoms versus stat marks).
  6. Create and interpret at least one generalized pairs plot.