Design Research Dissertation

  TAO YUZE 0366967 

Bachelor fo Design (Honours) in Creative Media

Project Link：https://drive.google.com/drive/folders/16eTJyVGncPYSCTcW68zOmiQvIy8VKWzO?usp=drive_link

Project 1 Draft dissertation

During the draft stage of the dissertation, my biggest obstacle was reviewing the research from the old research methods course, as my memory of the old research had become hazy due to the three semesters that had passed.I decided to start with the data itself and re-examine it. I retrieved and analyzed the data from my previous surveys.

The core task of the early versions was not "to write out all the conclusions," but rather to clearly define the boundaries of the research: Which transitional variables could I control? To what extent could the data support this? Which discussions could only serve as background information and not be presented as my empirical results?

During the project initiation phase, I broke down the "transitional animation" from an abstract narrative problem into several controllable variables more like production parameters: text volume, reading burden, number of characters speaking on screen, hourly frequency (pacing), and interaction type preference. These variables later became the main thread of the paper's methods and results sections.

At the same time, I also realized early on that since the survey data was limited and lacked stronger outcome variables such as fatigue scales, comprehension tests, and skipping behavior, the safest strategy for dissertations was to use "auditable descriptive statistics," using n/% to clearly explain the distribution, rather than making causal inferences or group difference tests that the data could not support. The paper's subsequent explicit use of "frequency-only (n/%)" and emphasis on not performing inferential statistics is a continuation of this early awareness of boundaries.

The so-called "expanded thinking on limited data" at this stage is mainly reflected in two things: First, use literature review to explain "why these variables are important", but at the same time clarify that these constructs (such as fatigue, coherence, reading fluency) were not directly collected by the questionnaire, so they can only be used as background framework; Second, clearly write down the types of data that may need to be added in the future (such as comprehension/recall tests and behavioral logs), and make clear "what I can say now and what I cannot say".

Iteration of  Dissertation

Once the formal writing began, the focus of the iteration shifted from "correct topic selection" to "readable and verifiable paper." I gradually developed a more rigorous structure for the article: each chapter's content was clearer, and every key conclusion had its source in a table.

The most significant change occurred in the expression of methods and results: I fixed the statistical scope to "use the valid n of each item as the denominator," avoiding directly dividing everything by the total sample size N=82; for multiple-choice questions, I explicitly stated that "percentages do not need to be summed to 100%." ​​These rules may seem trivial, but they determine whether readers can recalculate and avoid misinterpretation.

Furthermore, to avoid the confusion of "many tables, but each with a different denominator," I adopted a presentation strategy of "inserting a single consolidated summary table": Chapter 4 inserts the entire consolidated survey table in only one location, and subsequent text only references the entries from that table, without repeatedly pasting scattered smaller tables. The paper explicitly states this in the results section, emphasizing that it's for the purpose of maintaining denominator consistency and auditability.

This stage also involves an important trade-off: removing qualitative parts that weren't actually implemented to avoid readers thinking "it was done but not written down." The paper clearly indicates in the methods section that qualitative visual analysis has been removed and states that if it were to be done later, it would be recorded and reported separately. This makes the research boundaries more honest and more in line with the standards for undergraduate theses.

 Final dissertation

The dissertation process was ultimately fraught with difficulties. Just when I thought everything was settled, I received feedback from my supervisor—a comprehensive document of evaluation and criticism. It revealed many shortcomings in my work; the original version, in fact, had fatal flaws, forcing me to drastically revise my paper.

The major revisions in the final draft stage didn't primarily involve adding more content, but rather two types of fixes: one was correcting the argumentative structure, and the other was adjusting the delivery standards. In other words, it needed to simultaneously satisfy the requirements of "logical coherence and sound conclusions" and "read like a final draft for the professor."

At the argumentation level, I standardized the key sentence structures of the paper into more robust expressions: the conclusions were no longer written as "general laws," but rather as "baseline recommendations + applicable premises + deviation conditions." This approach is particularly suitable for dissertations that "can only use existing questionnaire data," as it maximizes the use of distributional information while avoiding exaggerating descriptive data into causal relationships. The paper repeatedly emphasizes "descriptive only / avoid over-claiming" in the methods and results sections to keep the conclusions within the limits allowed by the evidence.

At the delivery level, the final major revisions made the paper more like a submittable finished product: chapter titles, table of contents, figure and table labeling, and layout were unified; the results section provided more complete explanations of "how to read tables, valid n, missing, and multi-select"; and "unanalyzable variables" were explicitly listed as out of scope (e.g., subtitle language, reading fluency, fatigue/coherence scales, behavioral logs, etc.) to prevent readers from misunderstanding that the paper had omitted necessary analyses.

Project 2 dissertation Visual Design

This part made me feel a bit more relaxed; it felt more like my familiar area of ​​expertise.

Since my article is purely data analysis with few images and visual icons, I decided to use a more minimalist and restrained design language. I sought out many Bauhaus and International Style typographical works for reference.

My visual design process generally followed this sequence: first, establishing the "reader experience goals"; second, creating a "reusable layout system"; and finally, ensuring a proper correspondence between graphics and content.

First, I determined that the reading experience must be "long-text friendly," therefore choosing low-saturation colors and ample white space to prevent visual noise from distracting from the main text. Second, I established a fixed grid and hierarchy: chapter pages used strong contrast (graphic pages vs. title pages) to create rhythm, while text pages used consistent margins, line spacing, and heading levels to ensure continuous reading. Third, I created a reusable graphics library: contour textures, wave grids, checkerboard modules, and circular/semi-circular geometric shapes for chapter separation, transition pages, and information block decoration; these are not purely decorative but rather contribute to the visual "rhythm" of the entire paper, implicitly corresponding to the "pacing" of the research topic.

In terms of color scheme, I mainly used low-saturation beige/grayish-green, creating an overall calming feel suitable for long texts. The chapter covers and double-page spreads create a sense of breathing space (e.g., large areas of white space and large titles on the cover and chapter pages). I added recurring contour/topographical line textures, wavy grids, geometric checkerboard patterns, and striped modules to make the entire book feel like a unified system, rather than a temporary patchwork on each page (e.g., the cover and the chapter pages of Chapter 2, Chapter 3, and Chapter 4 are visually highly consistent). Chapter pages are often left-right splits, with one side featuring strong visual graphics and the other side displaying large-font titles and brief descriptions. The main text maintains relatively wide margins and a stable grid, emphasizing readability and information hierarchy. Conceptual framework diagrams use boxes and arrows to clearly map "variables → information," aligning with the academic paper's emphasis on "auditability and verifiability" (the framework diagram on the Chapter Structure page is a prime example).

Final Project Article

After completing the dissertation, my article wasn't simply a "shortening of the dissertation," but rather a redefinition: focusing the issue on a more specific, easily disseminated, and practically applicable direction—the hourly frequency of events within the first hour—and discussing the "baseline optimal frequency" based on this.

In the article, I proactively made two trade-offs: First, I withdrew the discussion of multiple variables, retaining only material directly related to "frequency/rhythm," ensuring readers' attention from the title to the conclusion. Second, I expressed the "optimal frequency" in a more conservative and verifiable form: avoiding causal claims and explanations beyond the questionnaire's capabilities, instead deriving a suggested framework of "default baseline + when to deviate" from the distribution itself (e.g., mode and central tendency). This approach aligns with the dissertation's "frequency-only, avoid over-claiming" principle, but the article's goal is clearer: to provide readers with a usable rhythm reference within a shorter space.

Looking back at the entire process, the maturation of dissertations occurred more in terms of "whether the research is verifiable and does not overstep boundaries": standardizing statistical methods, merging summary tables, clarifying "out of scope," and presenting conclusions as baseline recommendations rather than general laws.

The maturation of articles, on the other hand, occurred in terms of "whether the expression is focused and reusable": it uses dissertations as the foundation for data and methods, narrowing the problem to a more specific direction, and outputting more directly quotable conclusions with a more compact structure.