G-SWA Wave 4 · 2024–25 × Hofstede 6-D • n = 37 countries

Which strands of culture predict who works from home?

An explorer across all six Hofstede cultural dimensions and cross-country remote-work rates. Pick a dimension to see its scatter, map and formal statistics; the bar chart ranks how strongly each one tracks WFH. Spoiler from the data: Individualism is the one statistically robust correlate — the others are either weak or just re-tracings of it. But every one of these links, Individualism included, is confounded with national wealth and structure: correlation, not cause.

WFH = avg. full paid days/week from home, college-educated full-time workers · G-SWA Wave 4 (Nov 2024–Feb 2025), from the PNAS author replication data of Zarate, Barrero, Bloom, Davis, Dolls & Aksoy, "Why working from home varies across countries and people." Cultural dimensions = Hofstede 6-D (Individualism already in the author data; the other five added from the Hofstede matrix). Full provenance in the notes below.

The culture–remote-work plane

Each point is a country. Line is the OLS fit. Points coloured by region.

Formal statistics

IDV → WFH

WFH among college-educated full-time workers · G-SWA Wave 4.

+0.54 Pearson correlation, Individualism vs remote-work days

The standout. Autonomy- and self-reliance-valuing cultures adopt WFH far more — p < 0.001, ~29% of the spread. It is also the one cultural dimension the source paper itself models, and the only one of the six that survives a multiple-comparison correction.

Measure	Value	p-value

How each Hofstede dimension correlates with WFH

Sorted by absolute strength (|r| or |ρ|), sign preserved. Teal = more of this trait ↔ more WFH; clay = less. Faded bars are not significant; ✱ marks p < 0.05 — but with six dimensions tested, only Individualism survives a multiple-comparison correction.

Remote work — days/week from home

College-educated full-time workers. Higher = more common. Anglophone economies lead; East Asia trails.

Individualism (IDV)

Higher = more self-reliant, autonomy-valuing.

Reading the six dimensions

Individualism is the clear winner: r = +0.54, p < 0.001, ~29% of the cross-country spread — and the one cultural dimension the source paper itself models (the other five were added here). Everything else is weaker. Long-Term Orientation (r = −0.39) and Power Distance (r = −0.33) are the next strongest and cross p < 0.05, but both largely re-trace the same individualism/development gradient and neither would survive a correction for testing six dimensions.

Indulgence (+0.27), Uncertainty Avoidance (−0.22) and Masculinity (+0.05) are not significant — Masculinity is essentially flat. Flip the summary chart between Pearson and Spearman and the middle of the ranking reshuffles (LTO softens, PDI holds), a sign those mid-table links are fragile.

Multiple-comparison caveat: with six dimensions tested, the 0.05 threshold should be tightened (Bonferroni ≈ 0.008). Only Individualism clears that bar — treat the rest as suggestive, not established.

Why most fits are loose

The dimensions are intercorrelated: Individualism vs Power Distance is r = −0.71, so PDI mostly re-tells the Individualism story with more noise. Long-Term Orientation's negative link is largely the East-Asian cluster (China, Korea, Japan, Taiwan — all high-LTO, low-WFH) rather than an independent force.

And all of it is confounded, not causal: GDP per capita, lockdown stringency, population density, commute times and occupational mix co-vary with culture. In the source paper, English-speaking countries stay positive outliers even after controlling for those factors — so "culture" is a label for residual structure, not a proven mechanism.

Data provenance & measurement notes — read before citing

Remote-work & control data — PNAS author replication. WFH and the country covariates come from the replication data of Zarate, Barrero, Bloom, Davis, Dolls & Aksoy, "Why Working from Home Varies Across Countries and People" (PNAS 122(51), e2529036122, 15 Dec 2025). The plotted WFH value is the G-SWA Wave 4 (Nov 2024–Feb 2025) country collapse — the mean full paid days/week from home among the survey's college-educated full-time workers (37 countries). The replication file also carries the Individualism (IDV) score and controls (log GDP per capita 2019, industry-mix WFH-suitability, population density, lockdown stringency, round-trip commute minutes), which are not plotted here.
The other five Hofstede dimensions — added matrix values. Power Distance, Masculinity, Uncertainty Avoidance, Long-Term Orientation and Indulgence are not in the author data; they were merged in from the Hofstede 6-D dimension matrix (numeric values as transcribed in the earlier Culture & Remote Work report; the source workbook cites the Hofstede comparison tool, hofstede-insights.com/country-comparison-tool). A built-in audit in the workbook confirms the added IDV matches the author's IDV for every country (delta = 0), so the five new dimensions sit on a consistent footing with the one already present. Note different Hofstede presentations can differ by a few points for some countries.
Validity caveats. WFH covers college-educated full-time workers — read as "remote work among people whose jobs could be done from home," not the whole labour force (levels run higher than an all-workers measure). Hofstede scores are IBM/WVS-derived and debated for age and ecological-inference issues; Malaysia PDI = 104 and Greece UAI = 112 exceed the nominal 0–100 scale; South Africa combines white-sample IBM scores (PDI/IDV/MAS/UAI) with national LTO/IVR. ISO-3 codes and the regional colour groupings are added here for the maps and are not part of the source data.
Stats. Pearson assumes linearity/normality; Spearman is the rank-based robustness check — use the summary-chart toggle to compare. p-values are two-tailed and uncorrected; testing six dimensions inflates false positives, so apply a Bonferroni-style threshold (≈ 0.008) — only Individualism survives it. Detectable ≠ large ≠ causal.

▸ Full data table (37 countries)

Country	WFH days/wk	PDI	IDV	MAS	UAI	LTO	IVR