Women's mobility networks enable more efficient travel

Abstract

Our understanding of gender differences in mobility is marked by a clear tension: surveys portray women's movements as more complex than men's, while digital traces suggest less diverse travel. Here, we resolve the contradiction by modeling trajectories as networks of sequential visits, using smartphone traces linked to self-reported gender for 543,155 individuals across 10 countries. We show that the apparent conflict in the literature arises because women's mobility networks are simultaneously more clustered and more home-anchored -- a nuance obscured by aggregate metrics. This pattern arises because women tend to link multiple destinations within single trips, for trips spanning up to 150 km and multiple days. This organization yields systematically higher travel efficiency, measured as distance saved through destination chaining over monthly sequences.

Figures (22)

• Figure 1: Women explore more locations with fewer visits; men dominate the extremes.a. Distributions of activity and repertoire size for females (purple) and males (orange). Dashed lines mark bootstrapped medians; thin lines show country-level distributions; the bold line shows country-balanced results. Insets display the mean and standard error of the bootstrapped medians (see Methods). b. Gender differences in median repertoire size by activity decile. Error bars indicate the standard error of the difference in medians, estimated via country-balanced bootstrapping (Methods). Background shading highlights where the repertoire is larger for males (light orange) or females (light purple). c. Difference between the share of male and female users across deciles of activity (right) and repertoire size (left), shown separately by country (thin lines) and country-balanced sample (bold line, Methods). Insets report Kolmogorov–Smirnov (KS) statistics for the hypotheses M > F (light orange) and M < F (light purple), compared against random baselines with permuted gender labels (gray violin plots). d. Robust coefficient of variation (RCV, see Methods) for females (purple) and males (orange) by country. Error bars show the standard error of the bootstrapped RCV.
• Figure 2: Women's networks are more interconnected, with home as a central anchora. (bottom) Average relative difference across male-female pairs matched by activity and repertoire size (see Methods)for different network metrics (from left to right): clustering, number of cycles, and degree of the first, second, and third highest-degree nodes. For each metric, markers represent different activity groups (see legend). Gray violins represent the reference distribution obtained from 1,000 random shuffles of gender labels. (top) The fraction of countries in which each metric is significantly larger for females (purple bars), males (orange bars), or not significant (gray bars). Significance is assigned when the observed difference exceeds the 95th percentile of the shuffled distribution in the corresponding direction. b. Example networks for moderately active males (orange) and females (purple). Nodes represent locations, links indicate sequential visits, and node size reflects visit frequency (scaled exponentially to enhance visibility). Distances between nodes are proportional to their geographical separation. The example networks were selected to illustrate the contrasts shown in subplot (a), with the following criteria: female networks with high clustering and top-ranked degree centrality, and male networks with low clustering and high second-ranked degree centrality.
• Figure 3: Women make more two-stop tours, men favor back-and-fortha. Identification of tours in a monthly sequence of visits. The rectangles illustrate a visit sequence, with each distinct location represented by a different letter. Anchor locations are shown in dark gray; other visited locations in light gray (see Methods). b. Distribution of tours by length (right) and their corresponding differences between the share of male and female tours (left). Marker colors denote activity groups: inactive (light blue downward-pointing triangles), moderately active (green diamonds), active (dark green upward-pointing triangles), and all individuals (black circles). Results show means and bootstrapped standard errors averaged across countries; error bars are too small to be visible (see Methods). c. Gender differences (as defined in panel b) in the prevalence of back-and-forth tours (top) and two-stop tours (bottom), grouped by round-trip distance to the furthest location. Lines show country-averaged results for activity groups, with marker colors as in panel b.
• Figure 4: Women’s visit patterns support more efficient mobility.a. Schematic of two routing models from a source to a target node: random diffusion (top network, pink dashed line) and shortest path (bottom network, yellow line). The blue arrow depicts the physical distance between source and target. b. Average relative differences across male-female pairs matched by activity and repertoire size (see Methods) in network-wide and home-based mean first passage time (MFPT), global efficiency (unweighted and weighted by distance). Marker colors denote activity groups. Gray violins show reference distributions from 1,000 random shuffles of gender labels. c. Minimal illustration of tour efficiency, defined as the proportion of distance saved when sequencing visits instead of making independent round-trips from home (marked as H in black). Chaining nearby stops reduces the total travel cost while achieving the same overall reward. d. Relative gender difference between the average male and female tour efficiency against tour reward (see panel c). Results are shown for different activity quantiles (colorbar). The tour efficiency is computed as $1-\frac{\text{cost}}{\text{reward}}$ for each monthly sequence. Error bars show the standard error of the bootstrapped difference.
• Figure S1: Country-level bootstrapped medians of activity and repertoire size Mean and standard error of the bootstrapped medians (see Methods) for activity (left) and repertoire size (right, shown separately for men (orange) and women (purple).