pith. sign in

arxiv: 2607.01747 · v1 · pith:HSESPDCYnew · submitted 2026-07-02 · 💻 cs.HC

Adapting CCDF Plots for Visualizing Ordinal Regression Results

Abhraneel Sarma This is my paper

Pith reviewed 2026-07-03 07:06 UTC · model grok-4.3

classification 💻 cs.HC
keywords ordinal regressioncumulative-link modelsCCDF plotsvisualizationLikert itemsdata analysisHCI
0
0 comments X

The pith

Modified CCDF plots let researchers visualize cumulative-link ordinal regression results while showing the same takeaways as metric analyses of ordinal data.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper argues that cumulative-link ordinal regression models remain underused for ordinal data such as Likert items because their outputs are difficult to visualize and communicate. It proposes adapting complementary cumulative distribution function plots into a modified form that displays the probability structure from these models. This method allows the same comparative takeaways typically reported from linear models or ANOVA to be presented without assuming the data behaves as a continuous metric. A sympathetic reader would care because it could lower the barrier to using statistical approaches that match the ordered nature of the data more closely.

Core claim

Cumulative-link ordinal regression models provide an alternative to linear models for ordinal data with less stringent assumptions, yet their limited adoption stems from challenges in visually representing results. Modified CCDF plots solve this by enabling visualization of model outputs that communicate equivalent key takeaways to those from analyses treating ordinal data as metric.

What carries the argument

The modified Complementary Cumulative Distribution Function (mCCDF) plot, which adapts standard CCDFs to display cumulative probabilities from cumulative-link ordinal regression models.

If this is right

  • Researchers gain a visual method to report ordinal regression results that respects the ordered categories.
  • The same substantive conclusions presented from metric treatments of Likert data can be shown using ordinal models.
  • Adoption of cumulative-link models becomes feasible in visualization and HCI without specialized plotting expertise.
  • Visual communication of model results no longer requires treating ordinal data as continuous.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The approach could be tested on ordinal outcomes beyond Likert scales, such as ranked preferences or severity ratings.
  • Direct user studies comparing interpretation accuracy of mCCDFs versus traditional plots would clarify practical advantages.
  • Similar plot adaptations might apply to other regression families where visualization currently lags behind model sophistication.

Load-bearing premise

That mCCDF plots will accurately and intuitively convey the model outputs without introducing visual misinterpretations or losing critical information from the ordinal regression.

What would settle it

A controlled viewing study in which participants draw different or incorrect conclusions about effect sizes and group differences from mCCDF plots compared with standard metric visualizations of the same data.

Figures

Figures reproduced from arXiv: 2607.01747 by Abhraneel Sarma.

Figure 1
Figure 1. Figure 1: Illustration of how a cumulative probit-link model estimates thresholds of a latent variable (B), which can then be used to estimate [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Prototypical examples of visualizations of ordinal data analyses from the survey. (A) shows mean and error bars representing standard [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Calculation of a mCCDF plot from the latent variable and [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Calculation of a mCCDF plot for an experiment with mul [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
read the original abstract

Cumulative-link ordinal regression models are an alternative approach for analysing ordinal data such as Likert items, which are widely used in Visualization (and other related fields like HCI, psychology etc.). There are many researchers who are strong proponents of this approach, as it makes less stringent assumptions about the data, compared to the more commonly used linear model or ANOVA. Yet, ordinal regression models have seen limited adoption. I posit that one possible reason for this might be due to the difficulty in visually representing the results from such models, and in communicating the key takeaways in an intuitive manner. I propose the use of (modified) Complementary Cumulative Distribution Function (mCCDF) plots to visualize the results of ordinal regression models, and demonstrate how the same takeaways that researchers present from analyses which treat ordinal data as metric can be easily communicated using mCCDFs.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript proposes the use of modified Complementary Cumulative Distribution Function (mCCDF) plots to visualize results from cumulative-link ordinal regression models applied to ordinal data such as Likert items. It posits that visualization difficulty has limited adoption of these models despite their less stringent assumptions compared to linear models or ANOVA, and claims that mCCDF plots enable communication of the same key interpretive takeaways as analyses treating ordinal data as metric.

Significance. If validated, the proposal could meaningfully support wider use of ordinal regression in HCI, visualization, and psychology by providing an intuitive visual bridge to existing metric-style interpretations. The approach aligns with prior uses of cumulative distributions for ordinal data and avoids internal inconsistencies in the stated motivation.

major comments (2)
  1. [Abstract] Abstract: the claim that mCCDF plots 'can easily communicate' the same takeaways as metric analyses is presented without any empirical demonstration, example output, validation data, or comparison to existing visualizations, leaving the central effectiveness claim untested.
  2. [Abstract] Abstract: no construction details, equations, or parameter mappings are supplied for how mCCDF plots are derived from cumulative-link model outputs (e.g., thresholds or link-function probabilities), preventing assessment of whether the visualization preserves ordinal information or risks misinterpretation.
minor comments (1)
  1. The manuscript would benefit from explicit examples or pseudocode showing mCCDF construction from model coefficients.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their review and for identifying areas where the abstract could more clearly signal the manuscript's content. We respond to each major comment below.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the claim that mCCDF plots 'can easily communicate' the same takeaways as metric analyses is presented without any empirical demonstration, example output, validation data, or comparison to existing visualizations, leaving the central effectiveness claim untested.

    Authors: The abstract summarizes the contribution concisely. The full manuscript contains multiple worked examples applying mCCDF plots to cumulative-link model outputs on Likert data, directly showing how the same takeaways (e.g., group differences, effect directions) are conveyed as in metric analyses. These examples constitute the demonstration referenced in the abstract. The paper does not include a formal user study or quantitative validation of interpretability, as its scope is the proposal and illustration of the method rather than an empirical evaluation of visualization effectiveness. We can revise the abstract to explicitly reference the examples provided in the body. revision: partial

  2. Referee: [Abstract] Abstract: no construction details, equations, or parameter mappings are supplied for how mCCDF plots are derived from cumulative-link model outputs (e.g., thresholds or link-function probabilities), preventing assessment of whether the visualization preserves ordinal information or risks misinterpretation.

    Authors: Abstracts are space-constrained and focus on motivation and contribution. The manuscript body supplies the construction details, including the mapping from cumulative-link model thresholds and link-function probabilities to the mCCDF curves, with explicit equations that ensure the plots are derived directly from the ordinal model's cumulative probabilities. This derivation is designed to preserve ordinal information. We can add a short clause to the abstract noting that the plots are constructed from the model's cumulative probabilities. revision: partial

Circularity Check

0 steps flagged

No significant circularity; proposal is self-contained

full rationale

The paper is a methodological proposal for adapting mCCDF plots to visualize cumulative-link ordinal regression outputs. No derivation chain, equations, parameter fitting, or predictions are present that could reduce to inputs by construction. No self-citations, uniqueness theorems, or ansatzes are invoked to support the central claim. The suggestion that mCCDF plots communicate the same takeaways as metric analyses is presented directly as a visualization technique without any load-bearing reductions to fitted quantities or prior author work.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The proposal rests on standard assumptions of cumulative-link models for ordinal data and the mathematical properties of CCDF plots; no new free parameters, axioms, or invented entities are introduced in the abstract.

pith-pipeline@v0.9.1-grok · 5664 in / 1018 out tokens · 18122 ms · 2026-07-03T07:06:30.496748+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

43 extracted references · 28 canonical work pages

  1. [1]

    Bates, M

    D. Bates, M. Mächler, B. Bolker, and S. Walker. Fitting Linear Mixed- Effects Models Using lme4. Journal of Statistical Software , 67:1–48, Oct. 2015. doi: 10.18637/jss.v067.i01 6

    work page doi:10.18637/jss.v067.i01 2015

  2. [2]

    Bradley, G

    D. Bradley, G. Strain, C. Jay, and A. J. Stewart. Magnitude Judge- ments are Influenced by the Relative Positions of Data Points Within Axis Limits. IEEE Transactions on Visualization and Computer Graphics, 31(2):1414–1421, Feb. 2025. doi: 10.1109/TVCG.2024. 3364069 2, 3

    work page doi:10.1109/tvcg.2024 2025

  3. [3]

    Bürkner and M

    P .-C. Bürkner and M. Vuorre. Ordinal Regression Models in Psychol- ogy: A Tutorial. Advances in Methods and Practices in Psychological Science, 2(1):77–101, Mar. 2019. doi: 10.1177/2515245918823199 2

    work page doi:10.1177/2515245918823199 2019

  4. [4]

    Cabouat, T

    A.-F. Cabouat, T. He, P . Isenberg, and T. Isenberg. PREVis: Perceived Readability Evaluation for Visualizations. IEEE Transactions on Visu- alization and Computer Graphics , 31(1):1083–1093, Jan. 2025. doi: 10.1109/TVCG.2024.3456318 2, 3, 4

    work page doi:10.1109/tvcg.2024.3456318 2025

  5. [5]

    Carifio and R

    J. Carifio and R. J. Perla. Ten Common Misunderstandings, Miscon- ceptions, Persistent Myths and Urban Legends about Likert Scales and Likert Response Formats and their Antidotes. Journal of Social Sci- ences, 3(3):106–116, Mar. 2007. doi: 10.3844/jssp.2007.106.116 2

    work page doi:10.3844/jssp.2007.106.116 2007

  6. [6]

    Cushman, L

    F. Cushman, L. Y oung, and M. Hauser. The Role of Conscious Rea- soning and Intuition in Moral Judgment: Testing Three Principles of Harm. Psychological Science, 17(12):1082–1089, Dec. 2006. doi: 10 .1111/j.1467-9280.2006.01834.x 1, 6

    work page arXiv 2006

  7. [7]

    Dragicevic, Y

    P . Dragicevic, Y . Jansen, A. Sarma, M. Kay, and F. Chevalier. Increas- ing the Transparency of Research Papers with Explorable Multiverse Analyses. In Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems , pp. 1–15. ACM, Glasgow Scotland Uk, May 2019. doi: 10.1145/3290605.3300295 2, 3, 4

    work page doi:10.1145/3290605.3300295 2019

  8. [8]

    Gabry, R

    J. Gabry, R. ˇCešnovar, A. Johnson, and S. Bronder. Cmdstanr: R Interface to ’CmdStan’, 2025. 6

    2025

  9. [9]

    L. W. Ge, Y . Cui, and M. Kay. Calvi: Critical thinking assessment for literacy in visualizations. In Proceedings of the 2023 CHI Confer- ence on Human Factors in Computing Systems , CHI ’23. Association for Computing Machinery, New Y ork, NY , USA, 2023. doi: 10.1145/ 3544548.3581406 4

    work page arXiv 2023

  10. [10]

    Gigerenzer

    G. Gigerenzer. Mindless statistics. The Journal of Socio-Economics , 33(5):587–606, Nov. 2004. doi: 10.1016/j.socec.2004.09.033 2

    work page doi:10.1016/j.socec.2004.09.033 2004

  11. [11]

    Gigerenzer

    G. Gigerenzer. Statistical Rituals: The Replication Delusion and How We Got There. Advances in Methods and Practices in Psychological Science, 1(2):198–218, June 2018. doi: 10.1177/2515245918771329 2

    work page doi:10.1177/2515245918771329 2018

  12. [12]

    Helske, S

    J. Helske, S. Helske, M. Cooper, A. Ynnerman, and L. Besançon. Can Visualization Alleviate Dichotomous Thinking? Effects of Visual Representations on the Cliff Effect. IEEE Transactions on Visualiza- tion and Computer Graphics , 27(8):3397–3409, Aug. 2021. doi: 10. 1109/TVCG.2021.3073466 2, 3

    work page arXiv 2021

  13. [13]

    T. M. Liddell and J. K. Kruschke. Analyzing ordinal data with metric models: What could possibly go wrong? Journal of Experimental Social Psychology, 79:328–348, Nov. 2018. doi: 10.1016/j.jesp.2018 .08.009 1, 4

    work page doi:10.1016/j.jesp.2018 2018

  14. [14]

    R. Likert. A technique for the measurement of attitudes. Archives of Psychology, 22 140:55–55, 1932. 2

    1932

  15. [15]

    Lisnic, Z

    M. Lisnic, Z. Cutler, M. Kogan, and A. Lex. Visualization Guardrails: Designing Interventions Against Cherry-Picking in Interactive Data Explorers. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems , pp. 1–19. ACM, Y okohama Japan, Apr. 2025. doi: 10.1145/3706598.3713385 2, 3

    work page doi:10.1145/3706598.3713385 2025

  16. [16]

    McElreath

    R. McElreath. Statistical Rethinking: A Bayesian Course with Exam- ples in R and STAN . Chapman and Hall/CRC, New Y ork, 2 ed., Mar

  17. [17]

    doi: 10.1201/9780429029608 1, 6

    work page doi:10.1201/9780429029608

  18. [18]

    K. A. Nadib, M. Kogan, A. Lex, and M. Lisnic. Guardrail Selection in Line Charts to Contextualize Persuasive Visualizations. Computer Graphics F orum, 2026. 2

    2026

  19. [19]

    Nakano and T

    K. Nakano and T. Narumi. Avatars, Should We Look at Them Di- rectly or Through a Mirror?: Effects of Avatar Display Method on Sense of Embodiment and Gaze. IEEE Transactions on Visualization and Computer Graphics , 31(5):2912–2922, May 2025. doi: 10.1109/ TVCG.2025.3549545 2

    work page arXiv 2025

  20. [20]

    Pinheiro, D

    J. Pinheiro, D. Bates, S. DebRoy, D. Sarkar, S. Heisterkamp, B. V an Willigen, and R. Maintainer. nlme : Linear and nonlinear mixed effects models. r package version 3.1-103. http://cran.r- project.org/web/packages/nlme/index.html, 2017. 6

    2017

  21. [21]

    Rogha, S

    M. Rogha, S. Sah, A. Karduni, D. Markant, and W. Dou. The Im- pact of Elicitation and Contrasting Narratives on Engagement, Recall and Attitude Change With News Articles Containing Data Visualiza- tion. IEEE Transactions on Visualization and Computer Graphics , 30(7):4375–4389, July 2024. doi: 10.1109/TVCG.2024.3355884 2, 3

    work page doi:10.1109/tvcg.2024.3355884 2024

  22. [22]

    Sarma, S

    A. Sarma, S. Guo, J. Hoffswell, R. Rossi, F. Du, E. Koh, and M. Kay. Evaluating the Use of Uncertainty Visualisations for Imputations of Data Missing At Random in Scatterplots. IEEE Transactions on Visu- alization and Computer Graphics , 29(1):602–612, Jan. 2023. doi: 10. 1109/TVCG.2022.3209348 2, 3

    work page arXiv 2023

  23. [23]

    Sarma, M

    A. Sarma, M. Hedayati, and M. Kay. More Forecasts, More (Decision) Problems: How Uncertainty Representations for Multiple Forecasts Impact Decision Making. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems, CHI ’25, pp. 1–14. Associ- ation for Computing Machinery, New Y ork, NY , USA, Apr. 2025. doi: 10.1145/3706598.3713725 4

    work page doi:10.1145/3706598.3713725 2025

  24. [24]

    Sarma, K

    A. Sarma, K. Hwang, J. Hullman, and M. Kay. Milliways: Taming multiverses through principled evaluation of data analysis paths. In Proceedings of the 2024 CHI Conference on Human Factors in Com- puting Systems, CHI ’24. Association for Computing Machinery, New Y ork, NY , USA, 2024. doi: 10.1145/3613904.3642375 4

    work page doi:10.1145/3613904.3642375 2024

  25. [25]

    Saske, L

    A. Saske, L. Koesten, T. Möller, J. Staudner, and S. Kritzinger. A Multidimensional Assessment Method for Visualization Understand- ing (MdamV). IEEE Transactions on Visualization and Computer Graphics, 32(3):2695–2708, Mar. 2026. doi: 10.1109/TVCG.2026. 3653265 2

    work page doi:10.1109/tvcg.2026 2026

  26. [26]

    L. J. Savage. The F oundations of Statistics. Dover Publications, New Y ork, 2d rev. ed ed., 1972. 4

    1972

  27. [27]

    H. Song, A. Cho, C. X. Bearfield, and J. Stasko. Visualizing Trust: How Chart Embellishments Influence Perceptions of Credi- bility. IEEE Transactions on Visualization and Computer Graphics , 32(1):1306–1316, Jan. 2026. doi: 10.1109/TVCG.2025.3634785 2

    work page doi:10.1109/tvcg.2025.3634785 2026

  28. [28]

    South, D

    L. South, D. Saffo, O. Vitek, C. Dunne, and M. A. Borkin. Effec- tive use of likert scales in visualization evaluations: A systematic review. Computer Graphics F orum, 41(3):43–55, 2022. doi: 10.1111/ cgf.14521 2, 3

    2022

  29. [29]

    Spiegelhalter

    D. Spiegelhalter. Risk and Uncertainty Communication. Annual Re- view of Statistics and Its Application , 4(V olume 4, 2017):31–60, Mar

    2017

  30. [30]

    doi: 10.1146/annurev-statistics-010814-020148 4

    work page doi:10.1146/annurev-statistics-010814-020148

  31. [31]

    S. S. Stevens. On the Theory of Scales of Measurement. Science, 103(2684):677–680, June 1946. doi: 10.1126/science.103.2684.677 1

    work page doi:10.1126/science.103.2684.677 1946

  32. [32]

    R. C. Team. R: A Language and Environment for Statistical Comput- ing. R Foundation for Statistical Computing, 2024. 6

    2024

  33. [33]

    Victor Syiem and E

    B. Victor Syiem and E. V elloso. Better Assumptions, Stronger Con- clusions: The Case for Ordinal Regression in HCI. In Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems , pp. 1–21. ACM, Barcelona Spain, Apr. 2026. doi: 10.1145/3772318. 3790821 1, 2, 3

    work page doi:10.1145/3772318 2026

  34. [34]

    H. W. Wang, K. Lin, A. Cohen, R. Kennedy, Z. Zwald, C. Nobre, and C. X. Bearfield. Do Y ou “Trust” This Visualization? An Inventory to Measure Trust in Visualizations. IEEE Transactions on Visualization and Computer Graphics , 32(3):2515–2528, Mar. 2026. doi: 10.1109/ TVCG.2025.3646847 2, 3

    work page arXiv 2026

  35. [35]

    Wilkinson and C

    G. Wilkinson and C. Rogers. Symbolic description of factorial mod- els for analysis of variance. Journal of the Royal Statistical Society: Series C (Applied Statistics) , 22(3):392–399, 1973. doi: 10.2307/ 2346786 6

    1973

  36. [36]

    Y ang, M

    F. Y ang, M. Cai, C. Mortenson, H. Fakhari, A. D. Lokmanoglu, J. Hullman, S. Franconeri, N. Diakopoulos, E. C. Nisbet, and M. Kay. Swaying the Public? Impacts of Election Forecast Visualizations on Emotion, Trust, and Intention in the 2022 U.S. Midterms. IEEE Trans- actions on Visualization and Computer Graphics , 30(1):23–33, Jan

    2022

  37. [37]

    doi: 10.1109/TVCG.2023.3327356 2, 3, 4

    work page doi:10.1109/tvcg.2023.3327356 2023

  38. [38]

    Y ang, M

    F. Y ang, M. Hedayati, and M. Kay. Subjective Probability Correc- tion for Uncertainty Representations. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems , CHI ’23, pp. 1–17. Association for Computing Machinery, New Y ork, NY , USA, Apr. 2023. doi: 10.1145/3544548.3580998 2

    work page doi:10.1145/3544548.3580998 2023

  39. [39]

    Sarma, K

    F. Y ang, C. R. Mortenson, E. Nisbet, N. Diakopoulos, and M. Kay. In Dice We Trust: Uncertainty Displays for Maintaining Trust in Elec- tion Forecasts Over Time. In Proceedings of the CHI Conference on Human Factors in Computing Systems , pp. 1–24. ACM, Honolulu HI USA, May 2024. doi: 10.1145/3613904.3642371 2, 3, 4 A Supplementary Materials All supplemen...

    work page doi:10.1145/3613904.3642371 2024

  40. [40]

    The action principle : Harm caused by action is morally worse than equivalent harm caused by omission

  41. [41]

    The contact principle: Using physical contact to cause harm to a victim is morally worse than causing equivalent harm to a victim without using physical contact

  42. [42]

    The above scenario applies the action principle as the actor has to perform an action to create the outcome (as opposed to being passive and letting events unfold)

    The intention principle : Harm intended as the means to a goal is morally worse than equivalent harm foreseen as the side effect of a goal. The above scenario applies the action principle as the actor has to perform an action to create the outcome (as opposed to being passive and letting events unfold). However, the other two princi- ples do not apply as ...

  43. [43]

    For more details on the dataset, please refer to [ 6, 16]

    Each block included 15 test scenarios and 1 control scenario.” The dataset contains responses from 331 participants, and includes details on participants’ age, gender, and level of education. For more details on the dataset, please refer to [ 6, 16]. B.2 Analysis The mCCDF visualizations described in this paper are based on the results of the analysis out...