pith. sign in

arxiv: 2604.06580 · v1 · submitted 2026-04-08 · 💻 cs.SE

It's Not About Whom You Train: An Analysis of Corporate Education in Software Engineering

Rodrigo Siqueira , Danilo Monteiro Ribeiro This is my paper

Pith reviewed 2026-05-10 18:29 UTC · model grok-4.3

classification 💻 cs.SE
keywords corporate educationsoftware engineering trainingtraining perceptionprofessional developmentsurvey analysistraining effectivenesssociodemographic factorsvoluntary participation
0
0 comments X

The pith

Corporate training perceptions in software engineering depend more on professional experience and voluntariness than on personal demographics.

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

The paper examines whether who receives corporate software engineering training shapes how they rate its quality and effectiveness. It finds that personal details such as gender, age, education level, and company size show almost no influence, while career-stage factors like years of experience, professional level, and area of work produce a few localized differences. Mandatoriness of participation stands out as the strongest driver, lowering ratings across most items. A reader would care because the result points toward designing training around voluntary access and targeted adjustments for experience bands or role types rather than broad demographic tailoring.

Core claim

Analysis of responses from 282 Brazilian software engineering professionals found statistical significance in only 35 of 243 tested combinations of 27 perception items and nine variables. Mandatoriness affected 24 items negatively. Experience displayed a non-linear pattern with reduced engagement between three and six years. Area of work showed gaps in soft-skills coverage for advanced technical positions. Personal-profile variables and company size yielded no relevant differences, and the sample's 23 percent women showed no gender-based perception gaps.

What carries the argument

Non-parametric significance tests applied to 27 survey perception items crossed with nine sociodemographic and professional variables across 243 combinations to isolate which factors influence reported training quality and effectiveness.

If this is right

  • Voluntary participation should be prioritized because it improves perceptions across nearly all measured dimensions.
  • Training programs need adjustment for the experience band between three and six years where engagement dips.
  • Advanced technical roles require stronger soft-skills components than other areas of work.
  • Demographic tailoring by gender, age, or education level is unnecessary based on perception data.
  • Barriers to training equity likely lie in access and representation rather than the learning experience itself.

Where Pith is reading between the lines

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

  • Companies could test whether increasing the proportion of voluntary training sessions raises measured skill retention over time.
  • The mid-career engagement dip suggests a need for role-specific refreshers that the study did not directly measure.
  • Similar perception patterns may appear in non-software technical fields if the same survey items are reused.

Load-bearing premise

The 27 survey items validly measure quality and effectiveness of corporate training and the 282-person Brazilian sample generalizes without major selection bias or unmeasured confounders.

What would settle it

A replication using objective post-training performance measures or a larger sample drawn from multiple countries that finds significant effects from personal demographics such as gender or age would falsify the central claim.

Figures

Figures reproduced from arXiv: 2604.06580 by Danilo Monteiro Ribeiro, Rodrigo Siqueira.

Figure 1
Figure 1. Figure 1: presents the significance matrix resulting from the crossing of 27 perception items with 9 sociodemographic variables, total￾ing 243 combinations tested. Color intensity indicates the level of statistical significance. A methodological caveat is necessary before interpreting the results. With 243 tests conducted at 𝛼 = 0.05, approximately 12 significant results would be expected by chance (243 × 0.05 ≈ 12)… view at source ↗
Figure 3
Figure 3. Figure 3: Q25 (Autonomy at work) by Length of Experience. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Q14 (Organization and structure) by Length of Ex [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 2
Figure 2. Figure 2: Q23 (Performance improvement) by Length of Ex [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 5
Figure 5. Figure 5: Q15 (Useful materials) by Length of Experience. [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
Figure 8
Figure 8. Figure 8: Q7 (Relevance for competitiveness) by State. [PITH_FULL_IMAGE:figures/full_fig_p006_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Q7 (Relevance for competitiveness) by Region. [PITH_FULL_IMAGE:figures/full_fig_p006_9.png] view at source ↗
Figure 7
Figure 7. Figure 7: Q6 (Impact on personal time) by age groups. [PITH_FULL_IMAGE:figures/full_fig_p006_7.png] view at source ↗
Figure 10
Figure 10. Figure 10: Q10 (Training during working hours) by State. [PITH_FULL_IMAGE:figures/full_fig_p007_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Q8 (Motivation to learn) by Area of Work. [PITH_FULL_IMAGE:figures/full_fig_p007_11.png] view at source ↗
Figure 13
Figure 13. Figure 13: Q27 (Career growth opportunities) by Professional [PITH_FULL_IMAGE:figures/full_fig_p008_13.png] view at source ↗
read the original abstract

Context: Corporate education is a strategic investment in the software industry, but little is known about how different professional profiles perceive these initiatives. Objective: To investigate whether sociodemographic and professional variables influence the perception of quality and effectiveness of corporate training in Software Engineering (SE). Method: Non-parametric significance tests were applied to data from a survey with 282 Brazilian professionals, crossing 27 perception items with 9 sociodemographic variables (gender, age, education level, state, experience, professional level, company size, area of work, and nature of participation), totaling 243 combinations. Results: Of the 243 combinations tested, only 35 showed statistical significance. Training mandatoriness was the dominant factor, affecting 24 of 27 items. Length of experience revealed a non-linear descriptive pattern with a low-engagement zone between 3 and 6 years. Differences by area of work indicated an expressive gap in soft skills training for advanced technical roles. Personal profile variables and company size produced no relevant significant differences. Conclusion: Personal profile variables do not determine the perception of quality and effectiveness, while professional trajectory variables (experience, level, area of work) produce localized differences. The voluntariness of participation remains a determining factor, in line with the literature. The absence of gender differences in a sample with 23\% women suggests that barriers operate before training, in access and representation, not during the learning experience.

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 / 2 minor

Summary. This manuscript reports results from a survey of 282 Brazilian software engineering professionals investigating whether sociodemographic and professional variables influence perceptions of corporate training quality and effectiveness. The authors apply non-parametric significance tests to 243 combinations of 27 perception items with 9 variables (gender, age, education, state, experience, level, company size, area of work, nature of participation). They find only 35 combinations statistically significant, with mandatoriness dominant (24 items), localized effects from experience/level/area, and no relevant differences from personal profile variables or company size. The conclusion states that personal profiles do not determine perceptions, professional trajectories produce localized differences, and voluntariness is key, consistent with literature; gender findings suggest barriers precede training.

Significance. If the statistical claims hold after appropriate corrections, the work supplies empirical data on corporate SE training perceptions from a sizable sample, underscoring mandatoriness and professional experience over demographics. This could guide industry practices in training design and diversity initiatives by indicating that access rather than in-training experience drives gender disparities. The non-linear experience pattern and soft-skills gap for technical roles offer practical hypotheses for further study.

major comments (2)
  1. [Results] Results section (analysis of 243 combinations): The headline claim that 'personal profile variables do not determine the perception of quality and effectiveness' (repeated in Abstract and Conclusion) rests on non-significant results for gender, age, education, state, and company size. With 243 tests at nominal α=0.05 and no multiple-comparison correction (Bonferroni, FDR, or similar) reported, ~12 false positives are expected by chance; absence of significance cannot be interpreted as evidence of no effect. This is load-bearing for the central distinction between personal and professional-trajectory variables and is compounded by lack of power analysis for subgroups (e.g., 23% women ≈65 respondents).
  2. [Method] Method section: The manuscript provides no details on survey validation (e.g., pilot testing or reliability metrics for the 27 items), exact response rate, data exclusion rules, or the specific non-parametric procedures applied to each combination. These omissions prevent evaluation of whether the 35 significant results (mostly mandatoriness) are robust and directly undermine confidence in the 'no determination' conclusions for personal variables.
minor comments (2)
  1. [Abstract] The abstract and Results could more explicitly state the exact non-parametric tests (e.g., Mann-Whitney U for binary variables) and any descriptive statistics supporting the 'low-engagement zone' between 3-6 years of experience.
  2. [Results] Table or figure presenting the 35 significant combinations would improve readability; currently the breakdown (24 for mandatoriness, etc.) is described only in text.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback, which has helped us strengthen the statistical rigor and methodological transparency of the manuscript. We address each major comment below and have made revisions to incorporate the suggestions.

read point-by-point responses

  1. Referee: [Results] Results section (analysis of 243 combinations): The headline claim that 'personal profile variables do not determine the perception of quality and effectiveness' (repeated in Abstract and Conclusion) rests on non-significant results for gender, age, education, state, and company size. With 243 tests at nominal α=0.05 and no multiple-comparison correction (Bonferroni, FDR, or similar) reported, ~12 false positives are expected by chance; absence of significance cannot be interpreted as evidence of no effect. This is load-bearing for the central distinction between personal and professional-trajectory variables and is compounded by lack of power analysis for subgroups (e.g., 23% women ≈65 respondents).

    Authors: We agree that the absence of multiple-comparison correction limits strong claims of no effect from non-significant results. In the revised manuscript we have applied the Benjamini-Hochberg FDR procedure to the full set of 243 p-values. After correction the mandatoriness variable retains the large majority of significant associations (22 of the original 24), while the personal-profile variables remain non-significant. We have revised the abstract, results, and conclusion to state that personal profile variables showed no statistically significant associations after correction, rather than asserting they 'do not determine' perceptions. We have also added a dedicated paragraph on multiple testing and a post-hoc power analysis (using the observed sample sizes and non-parametric effect-size conventions) that acknowledges limited power to detect small effects in the gender subgroup. These changes preserve the observed pattern that professional-trajectory variables produce localized differences while personal variables do not. revision: yes

  2. Referee: [Method] Method section: The manuscript provides no details on survey validation (e.g., pilot testing or reliability metrics for the 27 items), exact response rate, data exclusion rules, or the specific non-parametric procedures applied to each combination. These omissions prevent evaluation of whether the 35 significant results (mostly mandatoriness) are robust and directly undermine confidence in the 'no determination' conclusions for personal variables.

    Authors: We have expanded the Method section to include the requested details. The survey instrument was reviewed by three SE academics and pilot-tested with 15 practitioners; item wording was refined on the basis of their feedback. Internal consistency is now reported (Cronbach’s α = 0.86 across the 27 items). The survey was distributed through LinkedIn groups and corporate mailing lists; 282 complete responses were obtained after excluding 41 incomplete submissions (those missing >20 % of items). For each combination we specify the test: Mann–Whitney U for binary factors (gender, mandatoriness) and Kruskal–Wallis H with Dunn post-hoc tests for multi-level factors (age bands, experience, company size, etc.). These additions allow readers to assess the robustness of the 35 significant results. revision: yes

Circularity Check

0 steps flagged

No circularity: observational survey with direct statistical tests on external data

full rationale

The paper reports results from applying standard non-parametric significance tests (Kruskal-Wallis, Mann-Whitney, etc.) to 282 external respondent answers. No equations, fitted models, predictions, or derivations exist that could reduce to self-definition, fitted inputs, or self-citation chains. The 243 test combinations and the 35 significant results are computed directly from the survey responses; the conclusion that personal-profile variables show no relevant differences follows from those external data outcomes rather than any internal construction. Self-citation is absent from the provided text and not load-bearing for any claim.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Empirical survey study with no mathematical derivations, free parameters, or postulated entities; the claim rests on the unstated validity of the survey instrument and appropriateness of the chosen statistical tests.

pith-pipeline@v0.9.0 · 5551 in / 1061 out tokens · 37732 ms · 2026-05-10T18:29:58.211807+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

27 extracted references · 27 canonical work pages · 2 internal anchors

  1. [1]

    Nana Assyne, Hadi Ghanbari, and Mirja Pulkkinen. 2022. The state of research on software engineering competencies: A systematic mapping study.Journal of Systems and Software185 (2022), 111183

    work page 2022

  2. [2]

    Timothy T Baldwin, Richard J Magjuka, and Brian T Loher. 1991. The perils of participation: Effects of choice of training on trainee motivation and learning. Personnel psychology44, 1 (1991), 51–65

    work page 1991

  3. [3]

    J Martin Bland and Douglas G Altman. 1995. Multiple significance tests: the Bonferroni method.Bmj310, 6973 (1995), 170

    work page 1995

  4. [4]

    Giovana Giardini Borges and Rogéria Cristiane Gratão de Souza. 2024. Skills development for software engineers: Systematic literature review.Information and Software Technology168 (2024), 107395

    work page 2024

  5. [5]

    Paul D Coverstone. 2003. IT training assessment and evaluation: A case study. In Proceedings of the 4th conference on Information technology curriculum. 206–215

    work page 2003

  6. [6]

    Breno Alves de Andrade, Rodrigo Siqueira, Lidiane C. S. Gomes, Antonio Oliveira, and Danilo Monteiro Ribeiro. 2025. A Mapping Study About Training in Industry Context in Software Engineering. InProceedings of the XXXIX Brazilian Sympo- sium on Software Engineering (SBES ’25) (CBSoft 2025). SBC, Recife, PE, Brazil, To appear

    work page 2025

  7. [7]

    Wellynton Diniz, Marcela Valença, César França, Alessandro Santos, and Mariana Pincovsky. 2024. The Skill Gap in Software Industry: A Mapping Study. In Simpósio Brasileiro de Engenharia de Software (SBES). SBC, SBES, BRA, 192–200

    work page 2024

  8. [8]

    Olive Jean Dunn. 1964. Multiple comparisons using rank sums.Technometrics6, 3 (1964), 241–252

    work page 1964

  9. [9]

    William Fitzgerald. 1992. Training versus development.Training & development 46, 5 (1992), 81–84

    work page 1992

  10. [10]

    M Galster, A Mitrovic, and M Gordon. 2018. Toward enhancing the training of software engineering students and professionals using active video watching. In 2018 IEEE/ACM 40th International Conference on Software Engineering: Software Engineering Education and Training (ICSE-SEET)(pp. 5–8). Gothenburg: IEEE

    work page 2018

  11. [11]

    Andreas Gegenfurtner, Karen D Könings, Nikola Kosmajac, and Markus Gebhardt

    work page

  12. [12]

    Voluntary or mandatory training participation as a moderator in the relationship between goal orientations and transfer of training.International Journal of Training and Development20, 4 (2016), 290–301

    work page 2016

  13. [13]

    William H Kruskal and W Allen Wallis. 1952. Use of ranks in one-criterion variance analysis.Journal of the American statistical Association47, 260 (1952), 583–621

    work page 1952

  14. [14]

    Virginia W Kupritz. 2002. The relative impact of workplace design on training transfer.Human resource development quarterly13, 4 (2002), 427–447

    work page 2002

  15. [15]

    Zion Lee. 2025. Overcoming Barriers to Corporate Training: A Strategic Frame- work for Success.New Advances in Business, Management and Economics Vol. 6 (2025), 156–173

    work page 2025

  16. [16]

    Henry B Mann and Donald R Whitney. 1947. On a test of whether one of two random variables is stochastically larger than the other.The annals of mathematical statistics(1947), 50–60

    work page 1947

  17. [17]

    MC Ohlmann, C Patenaude, and P Bourgault. 2019. The Perception and Impact of Workplace Training in Manufacturing SMEs: A Multiple Case Study.European Journal of Training and Development43 (2019), 498–515

    work page 2019

  18. [18]

    Tatalina Oliveira, Ann Barcomb, Ronnie De Souza Santos, Helda Barros, Maria Teresa Baldassarre, and César França. 2024. Navigating the path of women in software engineering: From academia to industry. InProceedings of the 46th International Conference on Software Engineering: Software Engineering in Society. 154–165

    work page 2024

  19. [19]

    Eduardo Salas and Janis A Cannon-Bowers. 2001. The science of training: A decade of progress.Annual review of psychology52, 1 (2001), 471–499

    work page 2001

  20. [20]

    Eduardo Salas, Scott I Tannenbaum, Kurt Kraiger, and Kimberly A Smith-Jentsch

    work page

  21. [21]

    It’s Not About Whom You Train: An Analysis of Corporate Education in Software Engineering CONF ’26, Month, 2026, City, Country

    The science of training and development in organizations: What matters in practice.Psychological science in the public interest13, 2 (2012), 74–101. It’s Not About Whom You Train: An Analysis of Corporate Education in Software Engineering CONF ’26, Month, 2026, City, Country

    work page 2012

  22. [22]

    Amalia Santos and Mark Stuart. 2003. Employee perceptions and their influence on training effectiveness.Human resource management journal13, 1 (2003), 27–45

    work page 2003

  23. [23]

    2006.Estatística não-paramétrica para ciências do comportamento

    Sidney Siegel and N John Castellan Jr. 2006.Estatística não-paramétrica para ciências do comportamento. Artmed Editora

    work page 2006

  24. [24]

    Rodrigo Siqueira, Antonio Oliveira, Breno Alves Andrade, Lidiane C. S. Gomes, and Danilo Monteiro Ribeiro. 2026. Corporate Training in Brazil- ian Software Engineering: A Quantitative Study of Professional Perceptions. arXiv:2604.05263 [cs.SE] https://arxiv.org/abs/2604.05263

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  25. [25]

    Rodrigo Siqueira, Antonio Oliveira, Breno Alves de Andrade, Lidiane C S Gomes, and Danilo Monteiro Ribeiro. 2026. Corporate Training in Brazilian Software Engineering: A Qualitative Study of Useful Learning Experiences. arXiv:2604.05209 [cs.SE] https://arxiv.org/abs/2604.05209

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  26. [26]

    Rodrigo Siqueira and Danilo Monteiro Ribeiro. 2026. The professional’s opinion: Suggestions for improving the corporate education training process in Software Engineering. arXiv:2603.13413 [cs.SE] https://arxiv.org/abs/2603.13413

    work page arXiv 2026

  27. [27]

    Charles Spearman. 1961. The proof and measurement of association between two things. (1961)

    work page 1961