Fast rising US baby names had lower retention at five years but no stable deficit at twenty years in matched 1950 to 1995 cohorts

We tested whether fast rising US baby names kept their use at fixed future dates. We used Social Security applicant counts in babynames 1.0.1. At ten starting years from 1950 to 1995, we selected names with positive five-year share growth and at least 100 current uses. We matched the upper and lower growth quartiles within sex and year, with current shares within about 5%. The primary test had 1,847 pairs. At twenty years, 66.11% of fast names and 65.13% of slow names kept at least half their starting share, a gap of +0.97 percentage points. This gap changed sign under tighter matching and differed by sex and starting year. A planned five-year measure gave rates of 92.58% and 96.81%, a gap of -4.22 points. The five-year gap stayed negative in all planned matching settings. Counts hidden below five cannot change any endpoint decision. Separate calculations checked all endpoints and primary matches. A clean run reproduced the analysis files byte for byte. These are descriptive results for selected historical records. Matching does not control all name history or identify a cause. The package includes input data, code, provenance, figures, all matched pairs and repeat instructions.

baby names · cultural change · retention · matching · administrative data · reproducibility