Pith. sign in

REVIEW 3 major objections 6 minor 36 references

Toward Inclusive AI-Driven Development: Exploring Gender Differences in Code Generation Tool Interactions

T0 review · 3 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This proposal hypothesizes that developers' gender changes how they interact with AI code-generation tools, affecting cognitive load, outcomes, and tool use, and plans a 54-person counterbalanced experiment to test it.

desk verdict A careful protocol, but it needs to control for prior tool familiarity before it can test gender differences. read the letter →

arxiv 2507.14770 v1 pith:BFJZ7LDN submitted 2025-07-19 cs.SE

Manaal Basha , Ivan Beschastnikh , Gema Rodriguez-Perez , Cleidson R. B. de Souza This is my paper
classification cs.SE
keywords CodeGenerationToolsGenderDifferencesCognitiveLoadTaskPerformanceToolUtilizationWindsurfHuman-ComputerInteractionRegisteredStudy
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This is a study proposal, not a results paper. Its pith is a hypothesis and a design to test it: developers' interactions with code-generation tools (CGTs) such as Windsurf differ by gender, and those differences show up in cognitive load, task outcomes, and how the tool is used. The authors argue the question matters because CGTs are trained largely on code from a male-dominated industry, so any gender-linked mismatch would point to a fairness gap in widely used developer tools. To test the claim, they plan a counterbalanced experiment with 54 computer science students, half women and half men, each completing two Python tasks, one assisted by Windsurf and one by internet search alone. The comparison to internet search is the benchmark: it asks whether gender differences are unique to AI assistance or already present in ordinary developer practices.

What carries the argument

The machinery is a mixed-subjects experiment with gender as the between-subjects variable and tool condition (Windsurf vs internet-only) as the within-subjects variable, so each participant serves as their own control. The named measurement instruments are a 15-item cognitive load questionnaire that separates intrinsic, extraneous, and germane load; Code Watcher, a custom VS Code extension that records prompts, suggestion acceptances, modifications, and flags AI-generated code; and unit-test suites that score core and advanced correctness. Statistical analysis uses ordinal regression (proportional odds model) for Likert-scale loads, Poisson or negative-binomial regression for prompt counts, robust linear regression for task time, and beta regression for proportions, with random intercepts for repeated measures. The internet condition functions as a baseline so that any gender effect found in the CGT condition can be read as specific to AI assistance rather than to programming in general.

What would settle it

Run the planned 54-person experiment as specified. If the preregistered analyses show no statistically significant gender effect on cognitive load, task outcomes, or tool utilization in either the Windsurf or internet condition, the central hypothesis is falsified. A more targeted check: collect a post-study self-report of VS Code and Windsurf familiarity; if familiarity differs by gender and correlates with every outcome metric, the gender effect cannot be distinguished from a tool-familiarity effect.

Watch

Extended reading notes

Core claim

The paper's central claim, stated as its hypothesis, is that gender affects how developers interact with code-generation tools, with measurable consequences for cognitive load, task outcomes, and tool utilization. It will test this by asking 27 women and 27 men to complete two medium-to-hard Python tasks, one using Windsurf and one using only internet resources, with task order and condition order counterbalanced across four groups. The planned outcomes are the three cognitive-load types (intrinsic, extraneous, germane) from a 9-point survey; completion time and code correctness from unit tests; and interaction metrics collected by a VS Code extension, including suggestion acceptance rate, number of prompts, modification of accepted suggestions, and the proportion of final code generated by the CGT. The authors expect statistically significant gender differences in each of these areas (H1-H3) and a difference in tool utilization between the CGT and internet conditions (H4).

Load-bearing premise

The load-bearing premise is that any gender difference observed in the experiment can be attributed to gender-related cognitive or behavioral traits rather than to unequal prior experience with VS Code, Windsurf, or the specific task types, since familiarity is not treated as a measured covariate.

Editorial extensions

If this is right

  • If H1-H3 hold, code-generation tool designers would have concrete, metric-level evidence of gender-linked differences in mental load, success, and usage patterns, not just anecdotal reports.
  • If H4 holds, the tool-vs-internet comparison would show whether CGTs widen, narrow, or leave unchanged gender gaps that exist with traditional search-based programming.
  • A null result for all hypotheses would itself be informative, suggesting that at least among senior CS students, Windsurf's interaction style does not track gender.
  • The per-metric analysis (intrinsic vs extraneous vs germane load) would point at which part of the interface, whether task complexity, UI design, or schema-building, drives any disparity, guiding targeted fixes.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • My inference: the same measurement stack (Code Watcher plus load survey) could be run on other code-generation tools with minimal changes; a positive result on Windsurf would be stronger if it replicated across interfaces.
  • My inference: the design's gender comparison could be extended to include self-reported prior exposure to VS Code and Windsurf as a covariate; without that, any observed gender effect is hard to separate from unequal familiarity, a threat the paper itself acknowledges.
  • My inference: the study's metrics, especially suggestion acceptance rate and modification rate, are natural audit signals for inclusive tool design and could be used by teams to check whether their internal AI assistants serve all developers evenly.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. This manuscript is a registered-report-style proposal rather than a completed empirical study: it motivates and designs a controlled experiment to test whether gender is associated with differences in cognitive load, task outcomes, and tool utilization when developers use the code generation tool Windsurf, compared with internet-only assistance. The planned design is a mixed within-between experiment with 54 participants (27 per gender), two Python tasks, counterbalanced task and condition order, cognitive load questionnaires, screen recordings, Code Watcher telemetry, and a menu of regression models. A pilot study with five participants is reported as feasibility evidence. The paper explicitly states that results are not yet available.

Significance. The research question is timely for the FATE agenda in AI4SE, and the paper has several strengths: it presents a falsifiable protocol, uses a validated cognitive load questionnaire, plans objective telemetry rather than self-report alone, counterbalances order, and reports pilot data that confirm the instrumentation works. If the design were sound and the results supported H1-H3, the study would provide useful evidence on whether CGT interfaces interact with gender-related differences in cognition and tool use. However, the current protocol has two load-bearing weaknesses: the sample-size justification does not match the statistical design, and the central between-subjects gender comparison is vulnerable to an acknowledged but unmeasured prior-experience confound. These issues need to be resolved before the protocol can be executed or accepted as a registered report.

major comments (3)
  1. [IV.A and IV.B] The power analysis is reported for a between-subjects design (approximately 54 participants in a between subjects design), but the primary gender comparisons in H1-H3 are between-subjects with 27 participants per gender, while the CGT-vs-internet comparison is within-subjects. If the effect size of 0.4 is Cohen's d, a two-group t-test with 27 per group has roughly 31% power, not 95%; if 0.4 is a different effect-size metric, the paper should define it. Please provide the exact test, effect-size metric, and design structure used in the power calculation, and compute the sample size for the between-subjects gender contrast that H1-H3 actually test.
  2. [IV.B and IV.G] The central hypotheses concern a non-manipulated between-subjects factor (gender), so confounds must be controlled. The paper acknowledges in Construct Validity that differences in familiarity with VS Code or Windsurf may introduce unintended variability in performance, and the participant survey collects CGT experience level and coding level, yet the analysis plan in Section IV.B lists only gender, task, and condition as predictors, with random intercepts for participant ID, and does not include prior Windsurf/VS Code/Python experience as a covariate, blocking variable, or stratification variable. Because prior exposure plausibly differs by gender (as the paper's own motivation cites unequal technology engagement), any observed gender difference in H1-H3 could be an artifact of experience. Please add an explicit experience measure to the models (or a pre-registered matching/blocking plan) and discuss the resulting interpretation.
  3. [Section III, H4 and Section IV.E] H4 states a statistically significant difference in tool utilization when using a CGT compared to the internet approach, but the utilization metrics defined in Section IV.E - prompt generation count, suggestion acceptance rate, modification rate, and proportion of final code generated by the CGT - are all CGT-specific and have no counterparts in the internet-only condition. The hypothesis therefore cannot be evaluated with the planned measurements. Please either define an equivalent internet-usage metric (e.g., number of searches, pages visited, copy-paste events) and pre-register it, or remove H4 and align the research questions with the hypotheses that are actually testable.
minor comments (6)
  1. [Section I] The phrase 'Largue-Language Models' should be 'Large-Language Models'.
  2. [Section IV.B] The analysis is described as 'semiparametric,' but POM, GLMs, and beta regression are parametric; remove 'semiparametric' or justify the term.
  3. [Table I] The column headers 'CGT Starts With CGT' and 'CGT Starts Without CGT' are confusing; rename them to 'CGT first' and 'Internet first'.
  4. [References] References [4] and [36] are the same paper (Wang et al., FAccT 2024); cite once and cross-reference the later discussion.
  5. [Section V] The sentence 'All the tasks, surveys, and explanations about the experiment can be found here' contains no URL; include the repository link.
  6. [Section IV.A] The inclusion criterion requiring prior CGT experience means the study cannot speak to first-time CGT users; state this as a scope limitation.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: this registered report proposes an experiment and makes no predictive or derivational claims that reduce to their inputs.

full rationale

The paper is a registered report (ESEM '25) proposing an experiment to test gender differences in CGT interaction, cognitive load, and task outcomes. It contains no equations, no fitted parameters, and no derivation of outcomes from inputs. The hypotheses H1-H4 are explicitly motivated by prior literature (e.g., Burnett et al. GenderMag, Barke et al. exploratory/acceleration states, Chen et al. cognitive load) but are not derived from the metrics the study plans to collect. The only self-citation involving an author is reference [10], Mendes, Souza, and De Souza, used to support the background claim that CGTs are designed to reduce cognitive load; this is not load-bearing for any hypothesis or design choice. The paper's own stated threat, that 'differences in familiarity with VS Code or Windsurf may introduce unintended variability in performance' (Section IV.G), identifies a potential confound in the proposed design, not a circular step: the independent variable (gender) is not defined in terms of the outcomes, and no outcome is defined in terms of gender. Because the study has not yet produced results, there is no fitted-input-called-prediction pattern and no renamed known result. The derivation chain, such as it is, consists of literature-motivated hypotheses and a planned statistical analysis; it is self-contained and non-circular.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

No data or derivations are present, so the ledger contains one design parameter (assumed effect size) and four background assumptions drawn from cited prior work. Code Watcher is a new measurement tool, but it is not a theoretical entity; its reliability depends on the unpublished extension and manual validation via screen recordings.

free parameters (1)
  • Assumed medium effect size for power analysis = 0.4
    Chosen in Section IV.A to compute the target sample size of 54 participants. If the true effect is smaller, the study will be underpowered.
assumptions (4)
  • domain assumption GenderMag's five cognitive facets (risk attitude, self-efficacy, information processing style, motivation, learning preference) differ by gender and transfer to CGT interaction.
    Cited from [23] and [24] in Section II; hypotheses H1-H3 depend on this transfer from prior technology studies to code generation tools.
  • domain assumption The 15-item cognitive load questionnaire validly separates intrinsic, extraneous, and germane load in this setting.
    Adopted from [32] and [33] in Section IV.C; if the three subscales are not separable, RQ1 cannot be answered as designed.
  • domain assumption Tinkering and exploratory behavior differences, plus the exploratory versus acceleration user states identified in prior CGT studies, apply to Windsurf users.
    Used in Section II to motivate RQ2 and RQ3 via [26] and [27].
  • domain assumption Self-identified gender, treated as a binary women/men variable, is a stable and meaningful independent variable for the planned comparisons.
    Section IV.A limits inclusion to participants who self-identify as women or men; non-binary participants are included only if the sample size is sufficient, which is unspecified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Toward Inclusive AI-Driven Development: Exploring Gender Differences in Code Generation Tool Interactions." pith.science (2026). https://pith.science/paper/BFJZ7LDN

@misc{pith2026250714770,
  author       = {Pith},
  title        = {Pith review of: Toward Inclusive AI-Driven Development: Exploring Gender Differences in Code Generation Tool Interactions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BFJZ7LDN}},
  note         = {Machine review of arXiv:2507.14770}
}
read the original abstract

Context: The increasing reliance on Code Generation Tools (CGTs), such as Windsurf and GitHub Copilot, are revamping programming workflows and raising critical questions about fairness and inclusivity. While CGTs offer potential productivity enhancements, their effectiveness across diverse user groups have not been sufficiently investigated. Objectives: We hypothesize that developers' interactions with CGTs vary based on gender, influencing task outcomes and cognitive load, as prior research suggests that gender differences can affect technology use and cognitive processing. Methods: The study will employ a mixed-subjects design with 54 participants, evenly divided by gender for a counterbalanced design. Participants will complete two programming tasks (medium to hard difficulty) with only CGT assistance and then with only internet access. Task orders and conditions will be counterbalanced to mitigate order effects. Data collection will include cognitive load surveys, screen recordings, and task performance metrics such as completion time, code correctness, and CGT interaction behaviors. Statistical analyses will be conducted to identify statistically significant differences in CGT usage. Expected Contributions: Our work can uncover gender differences in CGT interaction and performance among developers. Our findings can inform future CGT designs and help address usability and potential disparities in interaction patterns across diverse user groups. Conclusion: While results are not yet available, our proposal lays the groundwork for advancing fairness, accountability, transparency, and ethics (FATE) in CGT design. The outcomes are anticipated to contribute to inclusive AI practices and equitable tool development for all users.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

36 extracted references · 28 canonical work pages

  1. [1]

    Copilot by the numbers: Microsoft’s big ai bet paying off,

    D. Ramel, “Copilot by the numbers: Microsoft’s big ai bet paying off,” May 2024. [Online]. Available: https://visualstudiomagazine.com/ Articles/2024/02/05/copilot-numbers.aspx

    work page 2024

  2. [2]

    The widening gap: The benefits and harms of generative ai for novice programmers,

    J. Prather, B. N. Reeves, J. Leinonen, S. MacNeil, A. S. Randrianasolo, B. A. Becker, B. Kimmel, J. Wright, and B. Briggs, “The widening gap: The benefits and harms of generative ai for novice programmers,” in Proceedings of the 2024 ACM Conference on International Computing Education Research-V olume 1, 2024, pp. 469–486

    work page 2024

  3. [3]

    “it would work for me too

    R. Cheng, R. Wang, T. Zimmermann, and D. Ford, ““it would work for me too”: How online communities shape software developers’ trust in ai-powered code generation tools,” ACM Transactions on Interactive Intelligent Systems , vol. 14, no. 2, pp. 1–39, 2024

    work page 2024

  4. [4]

    Investigating and designing for trust in ai-powered code generation tools,

    R. Wang, R. Cheng, D. Ford, and T. Zimmermann, “Investigating and designing for trust in ai-powered code generation tools,” in Proceed- ings of the 2024 ACM Conference on Fairness, Accountability, and Transparency, ser. FAccT ’24. New York, NY , USA: Association for Computing Machinery, 2024, p. 1475–1493

    work page 2024

  5. [5]

    Exploring the learnability of program synthesizers by novice programmers,

    D. Jayagopal, J. Lubin, and S. E. Chasins, “Exploring the learnability of program synthesizers by novice programmers,” in Proceedings of the 35th Annual ACM Symposium on User Interface Software and Technology, 2022

    work page 2022

  6. [6]

    “it’s weird that it knows what i want

    J. Prather, B. N. Reeves, P. Denny, B. A. Becker, J. Leinonen, A. Luxton- Reilly, G. Powell, J. Finnie-Ansley, and E. A. Santos, ““it’s weird that it knows what i want”: Usability and interactions with copilot for novice programmers,” ACM Transactions on Computer-Human Interaction , vol. 31, no. 1, p. 1–31, Nov 2023

    work page 2023

  7. [7]

    Asleep at the keyboard? assessing the security of github copilot’s code contri- butions,

    H. Pearce, B. Ahmad, B. Tan, B. Dolan-Gavitt, and R. Karri, “Asleep at the keyboard? assessing the security of github copilot’s code contri- butions,” in 2022 IEEE Symposium on Security and Privacy (SP) , May 2022

    work page 2022

  8. [8]

    Accelerating software development using generative ai: Chatgpt case study,

    A. Rajbhoj, A. Somase, P. Kulkarni, and V . Kulkarni, “Accelerating software development using generative ai: Chatgpt case study,” in Pro- ceedings of the 17th Innovations in Software Engineering Conference , ser. ISEC ’24. New York, NY , USA: Association for Computing Machinery, 2024

    work page 2024

Show all 36 references
  1. [9]

    Interactive program synthesis by augmented examples,

    T. Zhang, L. Lowmanstone, X. Wang, and E. L. Glassman, “Interactive program synthesis by augmented examples,” in Proceedings of the 33rd Annual ACM Symposium on User Interface Software and Technology , 2020

    2020

  2. [10]

    “you’re on a bicycle with a little motor

    W. Mendes, S. Souza, and C. R. B. De Souza, ““you’re on a bicycle with a little motor”: Benefits and challenges of using ai code assistants,” in 2024 IEEE/ACM 17th International Conference on Cooperative and Human Aspects of Software Engineering (CHASE) , 2024, pp. 144–152

    2024

  3. [11]

    Interactive prompt debugging with sequence salience,

    I. Tenney, R. Mullins, B. Du, S. Pandya, M. Kahng, and L. Dixon, “Interactive prompt debugging with sequence salience,” arXiv preprint arXiv:2404.07498, 2024

    2024 arXiv

  4. [12]

    Introducing code referencing for github copilot chat in visual studio,

    R. Patel, “Introducing code referencing for github copilot chat in visual studio,” Dec 2024. [Online]. Available: https://devblogs.microsoft.com/visualstudio/ introducing-code-referencing-for-github-copilot-chat-in-visual-studio/

    2024

  5. [13]

    Is your code generated by chatgpt really correct? rigorous evaluation of large language models for code generation,

    J. Liu, C. S. Xia, Y . Wang, and L. Zhang, “Is your code generated by chatgpt really correct? rigorous evaluation of large language models for code generation,” Advances in Neural Information Processing Systems , vol. 36, 2024

    2024

  6. [14]

    Balancing chatgpt and data protection in germany: challenges and opportunities for policy makers,

    F. Leboukh, E. B. Aduku, and O. Ali, “Balancing chatgpt and data protection in germany: challenges and opportunities for policy makers,” Journal of Politics and Ethics in New Technologies and AI , vol. 2, no. 1, pp. e35 166–e35 166, 2023

    2023

  7. [15]

    Rai guidelines: Method for generating responsible ai guidelines grounded in regulations and usable by (non-) technical roles,

    M. Constantinides, E. Bogucka, D. Quercia, S. Kallio, and M. Tahaei, “Rai guidelines: Method for generating responsible ai guidelines grounded in regulations and usable by (non-) technical roles,” Proceed- ings of the ACM on Human-Computer Interaction , vol. 8, no. CSCW2, pp. ...

    2024

  8. [16]

    A comparative study of code generation using chatgpt 3.5 across 10 programming languages,

    A. Buscemi, “A comparative study of code generation using chatgpt 3.5 across 10 programming languages,” arXiv preprint arXiv:2308.04477 , 2023

    2023 arXiv

  9. [17]

    Mconala: a benchmark for code generation from multiple natural languages,

    Z. Wang, G. Cuenca, S. Zhou, F. F. Xu, and G. Neubig, “Mconala: a benchmark for code generation from multiple natural languages,” arXiv preprint arXiv:2203.08388, 2022

    2022 arXiv

  10. [18]

    Exploring the effect of multiple natural languages on code suggestion using github copilot,

    K. Koyanagi, D. Wang, K. Noguchi, M. Kondo, A. Serebrenik, Y . Kamei, and N. Ubayashi, “Exploring the effect of multiple natural languages on code suggestion using github copilot,” in Proceedings of the 21st International Conference on Mining Software Repositories, ser. MSR ’2...

    2024

  11. [19]

    Whose language counts as high quality? measuring language ideologies in text data selection,

    S. Gururangan, D. Card, S. K. Dreier, E. K. Gade, L. Z. Wang, Z. Wang, L. Zettlemoyer, and N. A. Smith, “Whose language counts as high quality? measuring language ideologies in text data selection,” arXiv preprint arXiv:2201.10474, 2022

    2022 arXiv

  12. [20]

    The effects of diversity in global, distributed collectives: A study of open source project success,

    S. Daniel, R. Agarwal, and K. J. Stewart, “The effects of diversity in global, distributed collectives: A study of open source project success,” Information Systems Research , vol. 24, no. 2, pp. 312–333, 2013

    2013

  13. [21]

    Gender differences and bias in open source: Pull request acceptance of women versus men,

    J. Terrell, A. Kofink, J. Middleton, C. Rainear, E. Murphy-Hill, C. Parnin, and J. Stallings, “Gender differences and bias in open source: Pull request acceptance of women versus men,”PeerJ Computer Science, vol. 3, p. e111, 2017

    2017

  14. [22]

    She Elicits Requirements and He Tests: Software Engineering Gender Bias in Large Language Models ,

    C. Treude and H. Hata, “ She Elicits Requirements and He Tests: Software Engineering Gender Bias in Large Language Models ,” in 2023 IEEE/ACM 20th International Conference on Mining Software Repositories (MSR) . Los Alamitos, CA, USA: IEEE Computer Society, May 2023, pp. 624–6...

    2023

  15. [23]

    Gendermag: A method for evaluating software’s gender inclusiveness,

    M. Burnett, S. Stumpf, J. Macbeth, S. Makri, L. Beckwith, I. Kwan, A. Peters, and W. Jernigan, “Gendermag: A method for evaluating software’s gender inclusiveness,” Interacting with computers , vol. 28, no. 6, pp. 760–787, 2016

    2016

  16. [24]

    A common language of gender?

    K. Beckwith, “A common language of gender?” Politics & Gender , vol. 1, no. 1, pp. 128–137, 2005

    2005

  17. [25]

    Computer self efficacy, computer anxiety, attitudes towards the internet and reported experience with the internet, by gender, in an east european sample,

    A. Durndell and Z. Haag, “Computer self efficacy, computer anxiety, attitudes towards the internet and reported experience with the internet, by gender, in an east european sample,” Computers in human behavior , vol. 18, no. 5, pp. 521–535, 2002

    2002

  18. [26]

    Gender differences and program- ming environments: across programming populations,

    M. Burnett, S. D. Fleming, S. Iqbal, G. Venolia, V . Rajaram, U. Farooq, V . Grigoreanu, and M. Czerwinski, “Gender differences and program- ming environments: across programming populations,” in Proceedings of the 2010 ACM-IEEE international symposium on empirical software en...

    2010

  19. [27]

    Grounded copilot: How programmers interact with code-generating models,

    S. Barke, M. B. James, and N. Polikarpova, “Grounded copilot: How programmers interact with code-generating models,” Proceedings of the ACM on Programming Languages , vol. 7, no. OOPSLA1, pp. 85–111, 2023

    2023

  20. [28]

    Gender differences in cog- nitive load when applying game-based learning with intelligent robots,

    B. Chen, G.-H. Hwang, and S.-H. Wang, “Gender differences in cog- nitive load when applying game-based learning with intelligent robots,” Educational Technology & Society , vol. 24, no. 3, pp. 102–115, 2021

    2021

  21. [29]

    Gender differences in episodic memory and visual working memory including the effects of age,

    F. Pauls, F. Petermann, and A. C. Lepach, “Gender differences in episodic memory and visual working memory including the effects of age,” Memory, vol. 21, no. 7, pp. 857–874, 2013

    2013

  22. [30]

    Experimental research in hci,

    D. Gergle and D. S. Tan, “Experimental research in hci,” in Ways of Knowing in HCI . Springer, 2014, pp. 191–227

    2014

  23. [31]

    How many participants do we have to include in properly powered experiments? a tutorial of power analysis with reference tables,

    M. Brysbaert, “How many participants do we have to include in properly powered experiments? a tutorial of power analysis with reference tables,” Journal of cognition , vol. 2, no. 1, p. 16, 2019

    2019

  24. [32]

    A systematic meta-analysis of the reliability and validity of subjective cognitive load questionnaires in experimental multimedia learning research,

    F. Krieglstein, M. Beege, G. D. Rey, P. Ginns, M. Krell, and S. Schnei- der, “A systematic meta-analysis of the reliability and validity of subjective cognitive load questionnaires in experimental multimedia learning research,” Educational Psychology Review , vol. 34, no. 4, p...

    2022

  25. [33]

    Development and validation of a theory-based ques- tionnaire to measure different types of cognitive load,

    F. Krieglstein, M. Beege, G. D. Rey, C. Sanchez-Stockhammer, and S. Schneider, “Development and validation of a theory-based ques- tionnaire to measure different types of cognitive load,” Educational Psychology Review, vol. 35, no. 1, p. 9, 2023

    2023

  26. [34]

    A deeper look at the nasa tlx and where it falls short,

    R. D. McKendrick and E. Cherry, “A deeper look at the nasa tlx and where it falls short,” in Proceedings of the Human Factors and Ergonomics Society Annual Meeting , vol. 62, no. 1. SAGE Publications Sage CA: Los Angeles, CA, 2018, pp. 44–48

    2018

  27. [35]

    Nasa-task load index (nasa-tlx); 20 years later,

    S. G. Hart, “Nasa-task load index (nasa-tlx); 20 years later,” in Pro- ceedings of the human factors and ergonomics society annual meeting , vol. 50, no. 9. Sage publications Sage CA: Los Angeles, CA, 2006, pp. 904–908

    2006

  28. [36]

    Investigating and designing for trust in ai-powered code generation tools,

    R. Wang, R. Cheng, D. Ford, and T. Zimmermann, “Investigating and designing for trust in ai-powered code generation tools,” in Proceedings of the 2024 ACM Conference on Fairness, Accountability, and Transparency, ser. FAccT ’24. New York, NY , USA: Association for Computing Ma...

    2024

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.