{"id":"4c6a9fa4-d845-4ce0-a426-6de6fd5598a4","arxiv_id":"2411.13950","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A grounded theory study finds that cognitive and emotional dysfunctions in ADHD and ASD software developers affect performance, and that accommodations and adjusted Agile practices can regulate the impact.","lead":"This paper builds a grounded theory from interviews with 25 neurodivergent software developers to explain how ADHD and autism-related cognitive differences affect their work performance. It proposes that individual journeys and workplace accommodations, including tailored Agile practices, can regulate these effects.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central 'SE Performance' category is not independently operationalized: the effect of cognitive dysfunctions on performance is coded from the same self-reported interviews that form the theory, so the model may establish perceived rather than actual performance impact.","rationale":"The paper is a serious and methodologically transparent socio-technical grounded theory study; I do not dispute the qualitative approach or the value of the reported experiences. For the central claim to hold in the strong sense stated in the abstract, cognitive dysfunctions must actually affect SE performance, not merely be experienced as affecting it. That requires some evidence about performance that is distinct from the accounts of the people whose cognition is under study. The paper does not provide this: the IFERA-II instrument measures self-reported executive-function difficulties, and the interview coding folds performance impact into the same self-reports. The reader's weakest assumption identified self-report bias; I extend this to a structural issue, namely the absence of any independent outcome construct for 'SE performance.' This is a load-bearing concern because the model's causal arrows and the paper's practical recommendations depend on the claim that these difficulties translate into performance effects. Still, the authors honestly acknowledge self-report limitations, provide open data artifacts, and present the theory as grounded rather than as a controlled demonstration. My concern reinforces the conditional verdict rather than moving it: the theory should be framed as a well-grounded account of perceived challenges and accommodations, with the direct performance effect left as a testable hypothesis rather than an established result.","tokens_in":17238,"tokens_out":3998,"duration_ms":44141,"concrete_test":"Have two independent coders, blind to the final theory, re-code all 25 neurodivergent interview transcripts for concrete performance outcomes (missed deadlines, rework, bug escape, poor performance review, promotion delay), classifying each as directly observed/dated versus inferred/feeling-based. If the concrete-outcome subset is small or absent, the model's 'affect SE performance' claims should be downgraded to 'perceived impact on work' and the causal arrows marked as hypotheses. A complementary check: for consenting participants, correlate IFERA-II scores with manager ratings or objective code-review/cycle-time metrics; a null result would further weaken the central claim's current wording.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section IV.A states: 'the effect on SE performance is described together with the Neurodivergent Cognitive Dysfunctions, as our interviewees would report how a specific dysfunction or challenge would affect their performance.' This is the crux: 'performance' is not a separately measured outcome. The IFERA-II instrument captures self-reported difficulties in executive functions, regulation, and delay aversion, not workplace performance. Figure 3 compares group averages of n=7, 5, 9, and 5 without inferential statistics, and Section V.A uses these averages to conclude that the results 'suggest that neurodivergent software engineers face significant cognitive and emotional regulation challenges that are likely to impact their performance.' Every arrow in Figure 2 from dysfunctions to SE performance is therefore backed by participants' retrospective causal attributions, filtered through the researchers' coding. This does not invalidate a grounded theory, but it means the abstract and conclusions overstate the evidence: the theory is currently about perceived impact, not demonstrated effect. Section VI acknowledges self-report bias, but the deeper issue is structural: the outcome variable is defined by the same reports used to define the causes.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents a Socio-Technical Grounded Theory (STGT) study of neurodivergent software engineers with ASD and ADHD. The authors conducted 25 interviews with neurodivergent software developers and 5 neurotypical controls, combined with the IFERA-II pre-interview questionnaire, and developed a conceptual model (Figure 2) in which neurodivergent cognitive dysfunctions (e.g., executive function difficulties, emotional dysregulation, social communication challenges) affect software engineering performance at both individual and team levels. The model also includes stress responses as an intermediate link and identifies the individual journey (diagnosis, therapy, medication, self-awareness) and accommodations (organizational support, Agile practices) as regulatory factors. The paper concludes with a set of recommendations for tailoring Agile practices to support neurodivergent engineers. The authors report transparent coding procedures, include pilot data, make materials available, and discuss reflexive threats from their own neurodivergent status.","tokens_in":17441,"tokens_out":3615,"duration_ms":35293,"significance":"If the central claim is supported, this paper addresses a genuinely understudied area in software engineering and offers a theoretical framework that can guide future research on neurodiversity in Agile teams. Its strengths include a clearly described STGT process, a comparison with neurotypical controls, practical recommendations grounded in interview data, explicit treatment of researcher positionality, and open access to data collection instruments and codebook. The work also contributes to a small but growing body of empirical studies on ADHD and ASD in software development, and the proposed model has the potential to inform both organizational accommodations and future confirmatory research. The main reservation is that the central outcome, 'software engineering performance,' is not independently operationalized, so the theory currently captures participants' perceived or experienced performance impact rather than an externally demonstrated effect.","major_comments":[{"comment":"The central claim that neurodivergent cognitive dysfunctions 'affect SE performance' is not supported by an independent operationalization of performance. Section IV.A states: 'the effect on SE performance is described together with the Neurodivergent Cognitive Dysfunctions, as our interviewees would report how a specific dysfunction or challenge would affect their performance.' Every arrow in Figure 2 from dysfunctions to SE performance is thus grounded in participants' retrospective self-reports, filtered through the researchers' coding. The abstract and conclusions present this as an unqualified effect. The authors should either reframe the theory as a model of perceived or self-reported performance impact, or supplement the interviews with objective performance indicators (e.g., task completion, code quality, performance reviews) to support the stronger claim.","section":"Section IV, Figure 2, and Abstract"},{"comment":"The comparison of IFERA-II scores across groups (ASD n=7, ASD+ADHD n=5, ADHD n=9, neurotypical n=5) is based solely on descriptive averages, with no inferential statistics. Despite this, Section V.A concludes that 'the results suggest that neurodivergent software engineers face significant cognitive and emotional regulation challenges that are likely to impact their performance.' The word 'significant' is misleading in this context, and the likely impact on performance is an extrapolation beyond the measured instrument, which captures self-reported difficulties in executive functions, regulation, and delay aversion rather than workplace performance. The group comparisons should be explicitly labeled as exploratory and illustrative, and the discussion should not present these averages as evidence of a demonstrated performance effect.","section":"Section V.A and Figure 3"},{"comment":"The limitations section acknowledges that 'relying on self-reported experiences may introduce bias,' but it does not identify the deeper structural issue: the outcome variable (SE performance) is defined by the same interviews that provide the causal accounts of dysfunctions. This circularity means that even if self-reports are accurate, the model describes how participants perceive their dysfunctions to affect their performance, not whether such an effect is observable externally. The authors should add an explicit limitation to this effect and discuss how future work could validate the model with independent performance measures.","section":"Section VI"}],"minor_comments":[{"comment":"The phrase 'the individuals' individual journey' is redundant and grammatically awkward; it should be 'the individual's journey.'","section":"Abstract"},{"comment":"There is a typo: 'Repetive sounds' should be 'Repetitive sounds.'","section":"Section IV.A.d"},{"comment":"The statement 'reveals distinct patterns' is based on descriptive averages in a small sample; it should be qualified as 'in this sample' or 'descriptively,' and the absence of significance testing should be acknowledged.","section":"Section V.A"},{"comment":"There is a duplicated word: 'still still see stereotypes' should be 'still see stereotypes.'","section":"Section V.E.d"},{"comment":"The phrase 'the the former was frequently cited' contains a duplicated article; it should read 'the former was frequently cited.'","section":"Section V.F.b"},{"comment":"The IFERA-II instrument is described as a 'pre-interview questionnaire' in the method, but the abstract calls it a 'survey.' Please align the terminology to avoid implying a separate survey instrument.","section":"Section III.A"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of software engineering venues and the qualitative methodology is appropriate for the research question. The main concern is that the central claim about 'effect on SE performance' needs to be carefully reframed as perceived impact or supported with additional evidence. The authors are transparent about many threats, but the structural issue of relying on the same self-reports for both causes and outcome should be addressed head-on. If the framing is adjusted, the paper could be a valuable contribution to the growing body of neurodiversity research in SE."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The short version: this is a legitimate grounded theory study and the first one I know of that puts ASD and ADHD developers in the same socio-technical frame and turns the data into concrete Agile practice adjustments. The theory is best read as a model of perceived impact, not demonstrated effect on output. That distinction matters, but it is not a reason to desk-reject.\n\nWhat's new: the combination of both conditions, the explicit link to Agile practices (TDD as task-initiation support, shorter pair programming, written communication, stable teams), and the discussion of co-occurring ASD+ADHD as an especially heavy load. The method section is honest about how STGT was applied. The figshare material with instruments and coding examples is a plus. The authors also build on their pilot and on Liebel et al. rather than pretend the topic is untouched.\n\nSoft spots: the sample is small and self-selected: 25 neurodivergent, 5 neurotypical, recruited via social media. The IFERA-II comparisons in Figure 3 are descriptive only, with groups of 5–9, and one neurotypical outlier moves the averages. More importantly, the stress-test note is right: 'SE performance' is not measured independently. The category is built from the same interviews in which participants attribute their difficulties to their conditions, and the abstract and Section V.A say the challenges 'are likely to impact their performance' as if this were a separate finding. It is a grounded theory of perceived performance impact. That may be exactly what a qualitative study can responsibly claim, but the wording needs to stay at that level.\n\nThe citation pattern looks fine: related work is covered, self-citations are to the pilot and to Liebel et al., and the targeted literature review is post hoc as intended. No circular parameter fitting here; the induction is genuine.\n\nBottom line: for SE researchers working on inclusion, and for practitioners who want evidence-informed accommodations, this is a useful paper. It deserves a serious referee, with the main request being to reframe performance as perceived performance and to either add inferential stats or explicitly say none are claimed. I would take it to our reading group and would cite it.","headline":"A genuine grounded theory of how ADHD/ASD developers experience their work, with concrete Agile adaptations; treat 'performance' as perceived, not measured, and it is a solid contribution.","tokens_in":17965,"tokens_out":2952,"would_cite":true,"duration_ms":27280,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A grounded theory from 30 interviews explains how ADHD and autism-related cognitive dysfunctions affect software engineering performance, and how individual journeys and workplace accommodations can regulate that effect.","keywords":["neurodiversity","software engineering","ADHD","autism spectrum disorder","grounded theory","Agile practices","executive functions","workplace inclusion"],"falsifier":"Run the same interview and IFERA-II protocol alongside objective performance records (completed story points, code-review outcomes, bug rates, or manager ratings) for a comparable sample; if the reported dysfunctions do not correlate with measurable performance differences, the theory's central claim would fail.","tokens_in":17058,"feed_emoji":"🧠","tokens_out":7753,"duration_ms":71811,"temperature":0.7,"pith_summary":"This paper builds a grounded theory from interviews with 25 neurodivergent and 5 neurotypical software engineers, claiming that the cognitive and emotional dysfunctions associated with ADHD and autism affect software engineering performance at both individual and team levels. It argues that this effect is not fixed: an individual's journey—diagnosis, therapy, medication, self-awareness—and accommodations from the organization and from Agile practices can regulate how strongly dysfunctions hurt performance. The study's practical payoff is an empirically grounded list of recommended Agile adaptations meant to improve inclusion. A sympathetic reader would take the central claim as: neurodivergent performance differences are real, identifiable in work settings, and manageable through concrete team-level changes.","feed_headline":"Agile tweaks can help neurodivergent developers perform better","feed_subtitle":"A 30-engineer grounded-theory study maps which Agile adjustments support engineers with ADHD and autism.","key_machinery":"The central object is the conceptual model of Figure 2, organized around the core category 'Software Engineering Performance' at individual and team levels. The model connects categories by explicit causal arrows: Neurodivergent Cognitive Dysfunctions and Social Interaction/Communication Challenges trigger Responses to Stress, those responses affect performance, and both the Individual Journey and Accommodations regulate that effect. The mediating chain is the central mechanism: cognitive, emotional, and social difficulties produce stress (anxiety, burnout, exhaustion, shutdowns), and the strength of that chain depends on the person's journey and on organizational and Agile accommodations. The IFERA-II pre-interview questionnaire supplies standardized self-ratings of executive-function difficulties, state regulation, and delay aversion, giving the qualitative model an auxiliary quantitative comparison across the ADHD, ASD, ASD+ADHD, and neurotypical groups.","core_discovery":"The core discovery is a grounded theory in which Neurodivergent Cognitive Dysfunctions affect Software Engineering Performance—the core category—through a chain of causes and moderators. In the model, cognitive and emotional dysfunctions (organization and planning, task initiation, perfectionism and overengineering, irregular work performance, emotional dysregulation, low self-esteem, context switching, cognitive flexibility, and focus and attention, including hyperfocus) together with social interaction and communication challenges trigger responses to stress such as mental exhaustion and anxiety. Those stress responses in turn affect performance at the individual level (for example, productivity) and at the team level (for example, communication). The Individual Journey and Accommodations—organizational support plus adapted Agile practices—regulate the effect. The authors present as the main novelty their empirically grounded recommendations for tailoring Agile practices, including clear task breakdown, test-driven development for task initiation, stable team composition, written communication, and shorter ceremonies.","pith_inferences":["Beyond the paper's claims, its model predicts that co-occurring ASD+ADHD engineers require the most extensive accommodations, because that group scored highest on all executive-function constructs in the study; a targeted follow-up could test this directly.","An extension the paper does not develop is whether the same regulatory logic applies to other neurodivergent conditions such as dyslexia or dyspraxia, since the theory's categories were built from ASD and ADHD accounts only.","The anecdotal use of ChatGPT to check message tone points toward a testable design direction: AI-assisted communication tools might act as a low-cost accommodation, and their effect on workplace inclusion could be measured experimentally.","One editorial reading is that Agile ceremonies function as externalized executive function—deadlines and daily check-ins become scaffolding for task initiation—so removing or weakening those ceremonies without a replacement could disproportionately harm neurodivergent engineers."],"forward_implications":["Agile teams should make task breakdown explicit and consider test-driven development as a way to help neurodivergent engineers start tasks and see when they are finished.","Keeping team membership stable gives autistic engineers the familiarity that reduces the social load of pair programming and other collaborative ceremonies.","Allowing written communication and shortening or strictly timing meetings, daily stand-ups, and retrospectives can reduce cognitive fatigue and anxiety.","Accommodations should be institutional rather than informal, because informal support varies and leaves some neurodivergent employees with unequal help.","Engineers with both ASD and ADHD are likely to need the most support, since that group reported the highest difficulties across all measured executive-function constructs."],"supporting_citations":[{"why":"Supplies the socio-technical grounded theory method that structures the entire study.","marker":"[13]"},{"why":"Defines the STGT approach by the same methodologist, grounding the iterative data collection and analysis process.","marker":"[30]"},{"why":"Provides the validated IFERA-II inventory used as the pre-interview questionnaire to measure executive-function difficulties, regulation, and delay aversion.","marker":"[31]"},{"why":"Prior case study of software engineers with ADHD whose findings this paper extends to ASD and to concrete Agile practice adaptations.","marker":"[28]"},{"why":"Earlier study of neurodiverse software engineering employees whose open research directions (pairing, process changes, message-nuance tools) this paper directly addresses.","marker":"[27]"},{"why":"Recent multi-vocal literature review documenting the scarcity of empirical studies on ASD inclusion in software engineering, motivating this study.","marker":"[11]"},{"why":"The authors' own pilot study on neurodiverse developers in Agile teams, which this full socio-technical grounded theory builds upon and expands.","marker":"[29]"}],"fun_headline_variants":["Agile tweaks for ADHD and autism: what works for devs","How ADHD and autism affect software dev performance","Grounded theory: agile practices that support neurodivergent devs","Study pins down agile tweaks for devs with ADHD and autism"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The study's link between cognitive dysfunctions and software engineering performance rests entirely on participants' self-reported experiences, with no objective performance data to confirm that the reported struggles actually changed their output, code quality, or workplace evaluations.","fun_headline_variants_meta":{"raw":{"variants":["Agile tweaks for ADHD and autism: what works for devs","How ADHD and autism affect software dev performance","Grounded theory: agile practices that support neurodivergent devs","Study pins down agile tweaks for devs with ADHD and autism"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000528,"raw_usage":{"total_tokens":2544,"prompt_tokens":940,"completion_tokens":1604,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":556,"completion_tokens_details":{"reasoning_tokens":1533}},"tokens_in":556,"tokens_out":1604,"duration_ms":12399,"temperature":1.0,"reasoning_tokens":1533,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T15:41:18.955313+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same interview and IFERA-II protocol alongside objective performance records (completed story points, code-review outcomes, bug rates, or manager ratings) for a comparable sample; if the reported dysfunctions do not correlate with measurable performance differences, the theory's central claim would fail.","supporting_citations":[{"cited_title":"Hoda, Qualitative Research with Socio-Technical Grounded Theory","cited_arxiv_id":null,"evidence_quote":"Supplies the socio-technical grounded theory method that structures the entire study."},{"cited_title":"Socio-technical grounded theory for software engineering,","cited_arxiv_id":null,"evidence_quote":"Defines the STGT approach by the same methodologist, grounding the iterative data collection and analysis process."},{"cited_title":"Development and content validity of ifera-ii for adults,","cited_arxiv_id":null,"evidence_quote":"Provides the validated IFERA-II inventory used as the pre-interview questionnaire to measure executive-function difficulties, regulation, and delay aversion."},{"cited_title":"Understanding and supporting neurodiverse software developers in agile teams,","cited_arxiv_id":null,"evidence_quote":"The authors' own pilot study on neurodiverse developers in Agile teams, which this full socio-technical grounded theory builds upon and expands."}],"review_version":1}