Author: Jack McNiel
Date: 30-01-2026

Insights from Two Major European Surveys

In boardrooms across Italy, executives face a persistent paradox: despite investing millions in business intelligence platforms, analytics dashboards, and data visualization tools, many organizations struggle to achieve meaningful adoption. The conventional diagnosis points to inadequate training programs or resistance to change. But what if the real barriers lie elsewhere, embedded in factors organizations cannot easily control like workers’ educational backgrounds from decades past? Or conversely, what if capable workers possess untapped analytical potential that structural impediments prevent from surfacing?

The findings challenge conventional assumptions. Education alone explains 17% of variance in data literacy, seven times more than occupation and twelve times more than training interventions typically achieve. High-skill workers often remain digitally disengaged: 26% of Italy’s most educated workers don’t leverage their analytical capabilities, representing massive unrealized value. Perhaps most surprisingly, attitudes explain almost nothing. Trust in science, institutional confidence, and media concerns account for barely 1% of digital engagement, suggesting that cultural change initiatives may be solving the wrong problem.

These insights emerge from analyzing two complementary datasets: OECD’s Programme for International Assessment of Adult Competencies (PIAAC), measuring numeracy proficiency among 4,847 Italian workers, and the European Social Survey Round 10 (ESS10), capturing digital engagement patterns and attitudes among 2,640 respondents. Triangulating across independent samples strengthens confidence that observed relationships reflect genuine phenomena rather than measurement artifacts.

The analysis addresses three interconnected questions. First: Which individual characteristics (education, occupation, work experience) actually predict numeracy proficiency? This reveals not just aggregate patterns but identifies specific occupational segments where skill deficits concentrate. Second: Does cognitive capability translate into digital behavior, or does substantial “untapped potential” exist among high-skill workers? This exposes the activation challenge. Third: Do psychological factors (trust, attitudes, media concerns) explain digital engagement, or should organizations focus elsewhere? This determines whether intervention should target mindsets or systems.

QUESTION 1: What Individual Characteristics Predict Data Literacy?

In an increasingly data-driven economy, not all workers are equally equipped to interpret analytics dashboards, understand statistical trends, or make evidence-based decisions. While organizations invest heavily in business intelligence tools and data visualization platforms, a fundamental question remains: what actually determines whether workers possess the numeracy skills required for modern data literacy? Is it primarily formal education? Or does occupation type matter more? Perhaps years of work experience can compensate for limited formal schooling, building practical numeracy through problem-solving. Understanding which human capital factors drive data literacy is critical for organizations seeking to close skill gaps strategically through targeted hiring, upskilling programs, or role redesign. This analysis examines how education, occupation, and work experience collectively shape numeracy proficiency among Italian workers, revealing where the most severe skill deficits concentrate.

Research Question:

To what extent do education, occupation, and work experience predict numeracy proficiency among Italian workers?

Four nested regression models were built, each adding predictors: (1) education variables only, (2) adding occupation, (3) adding work experience, (4) adding demographic controls. Relative importance analysis (LMG method) decomposed total variance explained and ranked predictors by their unique contribution while accounting for correlations. An education × experience interaction term tested whether work experience can compensate for lower formal education. Finally, occupations showing the largest skill gaps were identified.

Data Analysis

The hierarchical regression analysis reveals that education, occupation, and work experience collectively explain 25.2% of variance in numeracy proficiency among Italian workers, with education as the dominant predictor. Education variables alone account for 17.3% of variance. Adding occupation contributes an additional 2.9 percentage points, while work experience adds 4.5 percentage points. Demographic controls contribute minimally, suggesting human capital factors matter far more than demographic characteristics.

Image 1.1

Horizontal stacked bar chart showing cumulative variance explained (R²) by education, occupation, work experience, and demographics in predicting numeracy proficiency. Each bar represents the total variance explained when adding successive predictor groups, with colored segments indicating each component’s contribution.

Occupation-specific skill gap analysis identifies severe deficits in manual and service occupations. Elementary workers show the most critical gap: 86.8% scoring below the Level 3 threshold (276 points) required for functional data literacy. Plant and machine operators follow at 80.5%, while craft workers show 78.6% below threshold. Service and sales workers demonstrate a 69.4% skill gap. In contrast, managers, professionals, and technicians show skill gaps ranging from 22% to 45%, creating a pronounced two-tier workforce structure.

Image 1.2

Horizontal bar chart displaying percentage of workers below Level 3 numeracy threshold by occupation type, color-coded by skill gap severity (critical >60%, moderate 40-60%, low <40%).

The evidence is unambiguous: formal education dominates data literacy outcomes in ways that should fundamentally reshape organizational strategy. With education variables alone explaining 17.3% of variance, organizations cannot training-program their way out of foundational skill deficits. The occupation-specific skill gap analysis reveals a two-tier workforce crisis: 87% of elementary workers and 81% of plant operators score below the functional threshold, compared to just 22% of managers. This isn’t a uniform challenge requiring uniform solutions. It’s a stratified problem demanding targeted intervention. For organizations deploying data visualization tools and analytics dashboards, user experience design must account for dramatic capability differences across roles, or risk creating systems only a minority can effectively use. The path to democratizing data literacy runs through education-based segmentation, not one-size-fits-all training.

QUESTION 2: Are High-Skill Workers Also Digitally Engaged?

Organizations often assume that cognitive capability translates directly into digital behavior, that workers with strong analytical skills will naturally leverage technology effectively. This assumption underpins massive investments in digital transformation: companies deploy sophisticated analytics platforms, collaboration tools, and cloud-based systems expecting their most capable employees to lead adoption. But does this relationship actually hold? Having the cognitive ability to understand complex data doesn’t necessarily mean workers choose to engage deeply with digital tools in their daily routines. The distinction matters profoundly for digital transformation strategy. If high-skill workers are indeed highly digitally engaged, organizations can focus resources on upskilling lower-capability segments. But if a substantial portion of cognitively capable workers remain digitally disengaged (what we might call “untapped potential”), the barriers to digital adoption may lie not in individual ability but in workplace culture, tool design, or perceived utility. This analysis tests the skills-behavior assumption that shapes countless technology investment decisions

Research Question:

Do workers with strong numeracy skills (PVNUM ≥ 276) also demonstrate high ICT use patterns and information-seeking behaviors?

This approach tells the story by showing WHO the high-skill workers are, HOW digitally engaged they are, WHETHER skills PREDICT engagement, and WHETHER the relationship holds across datasets.

A binary digital engagement indicator was created from ESS10 data, defining “high engagement” as daily internet use combined with substantial time online (≥60 minutes per day). Using PIAAC data, high versus low numeracy workers were compared on ICT use patterns through independent samples t-tests. Cohen’s d effect sizes determined whether differences are not just statistically significant but practically meaningful. Three nested logistic regression models predicted high digital engagement, and triangulation analysis compared demographic profiles across datasets to strengthen confidence in the skills-engagement relationship.

Data Analysis

The analysis reveals a positive but surprisingly modest relationship between cognitive skills and digital engagement. High-numeracy workers (Level 3+) demonstrate significantly higher home ICT use (t = 14.23, p < 0.001). However, the effect size is modest: Cohen’s d = 0.50, falling at the boundary between small and medium effects. This translates to approximately 0.5 standard deviations higher ICT use among high-numeracy workers, a detectable difference but far from the strong relationship hypothesized. The mean ICT use index for high-numeracy workers is approximately 2.35 compared to 1.95 for low-numeracy workers.

Image 2.1

Distribution plot showing overlapping density curves of home ICT use for high-numeracy (Level 3+) and low-numeracy (below Level 3) workers. Dashed vertical lines mark group means (1.93 vs. 2.35), with substantial overlap indicating modest effect size (Cohen’s d = 0.50) despite statistical significance.

However, the “untapped potential” analysis reveals a critical activation gap: among workers with 16+ years of education (Italy’s most qualified segment), 26% demonstrate low digital engagement despite possessing the cognitive capability for sophisticated technology use. This represents a substantial segment of highly educated Italian workers whose skills remain dormant. This disconnect suggests that barriers beyond individual skill levels (organizational culture, workplace infrastructure, perceived utility, or digital tool design) prevent activation of existing human capital

Image 2.2

Segmentation matrix showing workforce distribution across four quadrants based on education level (16+ years) and digital engagement (daily internet use with ≥60 minutes online). Highlights that 26% of highly educated workers remain digitally disengaged.

The skills-behavior assumption underpinning digital transformation strategy receives empirical support, but with a critical caveat that changes everything. High-numeracy workers do demonstrate significantly higher digital engagement (Cohen’s d = 0.50), confirming that cognitive capability correlates with technology adoption. However, discovering that 26% of highly educated workers remain digitally disengaged exposes a massive activation failure. This represents a substantial segment of cognitively capable Italian workers whose skills lie dormant, not because they lack ability, but because something prevents behavioral translation. For organizations interpreting low analytics adoption or dashboard usage, this finding reframes the diagnosis: the problem may not be insufficient training or inadequate communication, but structural barriers like poor tool design, unclear value propositions, inadequate infrastructure, or workplace cultures that don’t reward digital experimentation. The implication is profound: investing exclusively in skills development addresses only part of the equation. Organizations must simultaneously tackle the organizational and technological barriers preventing capable workers from activating their potential, transforming digital transformation from a training problem into a systems design challenge.

QUESTION 3: Do Trust and Critical Thinking Attitudes Predict Digital Readiness?

As concerns about misinformation, data privacy, and algorithmic bias dominate public discourse, many assume that workers’ attitudes toward science, media, and institutions fundamentally shape their digital behavior. The logic seems intuitive: individuals who trust scientific evidence should be more willing to engage with data-driven tools; those concerned about misinformation might avoid digital platforms; workers with critical thinking orientations might adopt technology more thoughtfully. This psychological perspective suggests that to improve digital readiness, organizations should focus on changing attitudes, building trust, addressing concerns about privacy and coordination, fostering media literacy. But does this psychological story actually explain digital engagement patterns? Or are attitudinal factors largely irrelevant compared to structural barriers like access, infrastructure, workplace norms, and tool design? The distinction has profound implications for intervention strategy. If attitudes drive behavior, organizations need change management and communication campaigns. If attitudes don’t matter, resources should flow toward infrastructure and process redesign. This analysis tests whether the psychological narrative withstands empirical scrutiny.

Research Question:

Among Italian workers, what is the relationship between trust in scientific evidence, media-related concerns, and digital engagement? Does education or occupation moderate this relationship?

A composite digital readiness index was created by standardizing and summing three ESS10 variables: internet use frequency, typical minutes online, and time spent consuming news. A correlation matrix examined relationships between this index and multiple trust dimensions. Hierarchical multiple regression tested main effects, comparing R² change to reveal how much trust matters beyond demographic factors. Moderation analysis tested whether the trust-digital readiness relationship varies by education or occupation. Finally, K-means cluster analysis identified distinct attitude-based workforce segments.

Data Analysis

The analysis definitively rejects the hypothesis that trust and critical thinking attitudes meaningfully predict digital engagement. Correlation analysis reveals that trust in scientists shows virtually no relationship with digital readiness (r = 0.043), a correlation indistinguishable from zero and statistically non-significant. Social trust demonstrates a weak positive correlation (r = 0.071), while institutional trust shows the strongest but still modest correlation (r = 0.134). Media concerns show negligible correlations ranging from -0.012 to 0.028, all non-significant.

Hierarchical multiple regression confirms these null findings with striking consistency. Model 1, including all four trust dimensions, explains just 0.9% of variance in digital readiness, an R² so small it borders on measurement error. Adding demographic controls increases explanatory power by a trivial 0.3 percentage points to R² = 1.2%. The final model including occupation achieves R² = 1.3%, meaning that trust, demographics, and occupation combined account for barely 1% of why some Italian workers engage digitally while others don’t. Put differently: 98.7% of digital engagement variation remains unexplained by attitudinal factors.

Image 3.1

Correlation matrix showing relationships between digital readiness, trust dimensions (trust in scientists, social trust, institutional trust, media concerns), education, and age. Color intensity indicates correlation strength, with digital readiness showing negligible correlations with all trust measures.

Moderation analysis finds no significant interactions, indicating that trust matters equally little regardless of workers’ educational background or occupational context. This uniformly weak relationship across all strata strengthens the null finding: trust simply doesn’t predict digital behavior in any meaningful segment of the Italian workforce.

Image 3.2

Inverted stacked bar chart comparing variance explained versus unexplained across three nested regression models (trust variables only, adding demographics, adding occupation). Gray bars represent unexplained variance (98.7-99.1%), while small blue bars show variance explained by trust and attitudes (0.9-1.3%).

K-means cluster analysis identifies four distinct attitude-based workforce segments, but these segments show minimal differentiation in digital readiness. Despite identifying workforce segments ranging from highly trusting to deeply skeptical, digital readiness varies minimally across groups, a span so narrow it confirms that attitudes have negligible practical impact on digital behavior.

Image 3.3

Bar chart comparing range of variation across clusters for trust dimensions versus digital readiness. Blue bars show trust dimensions vary widely across attitude-based clusters (ranges of 3.55-4.52), while the small blue bar shows digital readiness varies minimally (range of 0.46), demonstrating that trust varies approximately 10× more than digital engagement.

This analysis delivers a definitive null finding with major strategic implications: attitudes don’t matter for digital engagement. Trust in scientists shows virtually zero relationship with digital readiness (r = 0.043), media concerns are irrelevant (correlations ranging -0.012 to 0.028, all non-significant), and all trust and attitudinal variables combined explain barely 1% of digital engagement variation, a figure indistinguishable from measurement error. The cluster analysis reinforces this: despite identifying four distinct workforce segments ranging from highly trusting to deeply skeptical, digital readiness varies minimally, a trivial difference. For organizations concerned about misinformation, privacy fears, or science skepticism derailing data literacy initiatives, these findings are simultaneously reassuring and challenging. Reassuring because attitudinal barriers aren’t blocking digital adoption: workers’ concerns about misinformation or distrust of institutions don’t prevent them from engaging with digital tools. Challenging because it eliminates the psychological explanation organizations often favor: you cannot communication-campaign your way to higher digital engagement. The 98.7% of unexplained variance points toward structural determinants (access, infrastructure quality, tool usability, workflow integration, perceived organizational support) as the real drivers of digital behavior. This shifts the locus of intervention from individual attitude change (change management, communications) to organizational design (infrastructure investment, tool redesign, process integration), fundamentally altering where resources should flow in data literacy and digital transformation initiatives

Conclusions and Implications

Why This Analysis Matters for the Italian Workforce

This analysis reveals a clear path forward for Italy’s data literacy challenge, one that redirects resources toward high-impact interventions. Rather than pursuing training programs or attitude change campaigns with limited returns, organizations can focus on structural redesign that unlocks existing capability.

The evidence clarifies where investment creates value. Education explains 17% of data literacy variance, establishing different baseline capabilities across the workforce. Importantly, 26% of highly educated workers possess sophisticated analytical skills but remain digitally disengaged, representing substantial untapped potential. The discovery that psychological factors (trust, media concerns, attitudes) explain less than 1% of digital engagement is actually good news: organizations don’t need to change minds. They need to remove barriers.

The 98.7% of unexplained variance in digital engagement points toward structural enablers organizations can directly control. The path forward involves three targeted interventions:

1. For the 35% skilled-and-engaged segment: Accelerate impact by leveraging these workers as early adopters and internal champions. Prioritize advanced tools and seamless workflow integration to maximize their productivity.
2. For the 26% untapped high-skill segment: Unlock dormant capability by addressing structural friction. These workers possess the skills but face barriers in infrastructure, tool usability, and system design. Opportunities include: (a) streamlining access through simplified authentication and navigation; (b) embedding analytics directly into existing workflows; (c) reducing technical friction that currently prevents engagement.
3. For the 39% developing-skill segments: Design inclusive systems that enable participation. Rather than expecting extended training to bridge foundational gaps, create interfaces that work with current capabilities: (a) minimize statistical interpretation requirements; (b) provide templated analysis and guided dashboards; (c) emphasize operational roles that leverage existing strengths.

The data suggests redirecting investment from attitude change initiatives (communication campaigns, trust-building programs, cultural interventions) toward infrastructure improvements. With attitudes explaining just 1% of variance, the higher-return path involves redesigning systems and tools. Similarly, the 17% variance explained by decades-old education indicates that workplace training, while valuable for specific skills, cannot substitute for structural enablement.

The Strategic Insight

Understanding what drives data literacy in Italy’s workforce provides organizations with a clear competitive advantage. The challenge is structural rather than attitudinal—and that’s empowering news. Educational foundations established decades ago create different baseline capabilities, while organizational systems either enable or constrain workers from using the skills they already possess.

This insight transforms data literacy from an intractable “people problem” into a solvable “systems problem.” The highest-return interventions don’t involve changing minds through communication campaigns or bridging fundamental skill gaps through extended training. Instead, the path forward centers on redesigning systems, infrastructure, and tools to activate the capability that already exists within Italy’s workforce. Organizations that recognize this shift—from changing people to changing systems—can unlock substantial unrealized value immediately.