What the data says
A guided report of the dataset, its strongest findings, the machine-learning results and the limits on what any of it can mean.
Two connected datasets
The historical layer reconstructs 4,250 published profile rows from 1988 through 2026, links 2,924 recovered presidency intervals to 720 mission identities, and preserves 350 recorded changes to the mission system. Names are resolved across renames and splits rather than treated as permanent identifiers.
The statistical layer is the audited 2026 roster: all 504 missions, with 501 named presidents and near-complete biographies for age, children, origin and prior leadership. The comparisons and models below use that cleaner cross-section; the historical archive is best used for search, timelines and change over time.
Public-record coverage is systematic from 1988, but archive completeness varies by year and field. Unknown values remain unknown. A missing calling in an older biography is not evidence that the person never held it, and apparent year-to-year changes can partly reflect reporting style.
The 2026 profile, in plain numbers
A mission president is typically called in a narrow season of life, but there is no single career path. The mean and median age are both near 55, while prior service is nearly universal at the ward level and much less uniform at the stake level.
The sharpest divide is not age. It is leadership pathway.
International presidents are about five years younger and have roughly one fewer child, yet they are far more likely to have already presided over a stake. That gap survives controls for age, children and call cohort: the adjusted odds of prior stake-president service are about 4.45 times higher for an international president. Age itself contributes almost nothing after origin is known (OR 1.003, p = 0.84).
| Measure | International | United States | Utah | US, excluding Utah |
|---|---|---|---|---|
| Mean age | 51.7 | 56.4 | 56.8 | 56.2 |
| Mean children | 3.60 | 4.68 | 4.63 | 4.71 |
| Previously stake president | 69.9% | 38.0% | 29.3% | 43.8% |
| Previously bishop | 88.0% | 92.5% | 91.0% | 93.5% |
| Wife previously RS president | 51.8% | 33.8% | 33.1% | 34.3% |
Utah presidents look like other Americans in age and family size, but only 29.3% had led a stake, versus 43.8% elsewhere in the US. Utah’s dense church structure appears to create more routes to mission leadership that stop short of the stake-president seat.
98.3% had served in a bishopric, but 91% had been the bishop. At the stake level, 79% had sat on the council while only 48.6% had led it. Counting only top titles erases a large share of institutional experience.
What unsupervised machine learning found
The model was given five standardized attributes—age, number of children, prior service as bishop, prior service as stake president, and the wife’s prior service as Relief Society president—with no labels for origin or call year. PCA described the main directions of variation; K-means, Ward hierarchical clustering and DBSCAN independently searched for recurring profiles.
K-means peaked at a silhouette score of 0.380 for k=7. That is meaningful structure, but only moderate separation: these are overlapping pathways, not seven natural castes.
One 42-person cluster was unusually young, 78.6% international and averaged 2.83 children; 85.7% of their wives had served as Relief Society president.
All had served as bishop, none as stake president, and none had a wife reported as a former RS president. Utah supplied 41.1% of this group.
The model finds pathways, not destiny.
A random forest could identify four origin groups from these five attributes only 37.4% of the time, barely above the 33.2% majority baseline. It could not identify call year at all: 30.9% accuracy, below the 39.1% baseline. The profiles differ, but age, family size and three prior callings do not encode a clean national type—and they do not show a detectable 2024-to-2026 cohort shift.
Three storylines worth remembering
The United States supplies 334 presidents where its share of worldwide membership implies about 192. Brazil and Mexico each supply fewer than half their proportional expectation; Peru supplies four where about eighteen would be expected.
Despite dramatic-looking state rankings, the 51-state distribution is consistent with random variation around membership share (χ²=54.6, p=0.305). No state survives multiple-testing correction.
Only 19.0% serve in their own state or country. International presidents are much more likely to remain in-country than Americans are to remain in-state, revealing two very different assignment systems.
Method and limits
Sources and tests
- Official assignment lists and public Church News biographies.
- Newsroom membership, stake and congregation totals by place.
- Welch and Fisher tests with Benjamini–Hochberg correction.
- Poisson proportionality tests and HC3-robust logistic regression.
- Median imputation and z-score standardization before PCA and clustering.
What this cannot establish
There is no comparison group of people who were not called, so the project cannot predict who becomes a mission president or infer why an individual was selected. Biography text measures what was publicly reported, not every calling ever held. Associations are descriptive; they are not causal.