How SSA Baby Name Data Actually Works

The source is Social Security card applications, not a birth registry. A name is counted as written on the application, filed under the stated year of birth. In practice this tracks births closely, because most US newborns get a card, but the distinction has consequences: a child without an application never appears, and the name recorded is the name on that form rather than whatever the family uses day to day.

Aiden and Aidan are two entries. Sofia and Sophia are two entries. Nothing in the source merges them, so any ranking that does is applying an editorial rule of its own.

This is the single largest source of disagreement between name sites. A list that combines variants can move a name ten or twenty places, and both lists can be defended. We do not merge, because the merge rule would be ours rather than the data's, and once you start there is no principled place to stop.

In the state file, SSA withholds any name with fewer than 5 occurrences in a given state and year. The reason is privacy: with a small enough count a name plus a state plus a year can identify a specific child.

The consequence is that a blank is not a zero. It means "fewer than 5". Treating those blanks as zeros, which is the easy mistake when loading the file into a spreadsheet, will overstate how regional rare names are and understate their national totals.

The boys' and girls' files are separate. An overall ranking is the two added together, so a name used for both sexes ranks higher overall than in either list alone. That is why a name can sit at #12 overall while placing around #30 among boys.

There is no unisex flag in the data. Whether a name counts as unisex is a threshold someone chose, and different sites choose differently.

SSA publishes the previous year annually, usually in the first half of the following year. Past years do not change once published, so historical ranks are stable and worth citing. The current year does not exist in this data at all, which is worth knowing before you go looking for it. Any site showing a US ranking for a year in progress is not using this file.

The first is treating a withheld state value as zero. Load the state file into a spreadsheet, pivot by name, and the blanks silently become zeros. Every rare name then looks sharply regional and its national total comes out short. This is the most common error in state-level name journalism.

The second is comparing across the rank cap. SSA's web forms give the top 1,000, so a name at rank 1,400 is absent from the form but present in the download. An analysis built on the form will report a name as having disappeared when it merely fell past the cap.

The third is reading rank change as popularity change. Rank gaps are tiny near the bottom of a list and large at the top, so a forty-place jump at rank 80 can represent fewer additional births than a two-place jump at rank 5. Rank is an ordering, not a measurement, and the counts are right there in the file.

It does not record ethnicity, language, religion, or family origin, so any claim about which community favours a name is an inference from geography rather than something the file says.

It does not record middle names, nicknames, or later legal name changes. The count is the name on one form at one moment.

And it says nothing about why. The file can tell you that a name gained forty places; it cannot tell you that a television series caused it. Attributions like that are worth making but they come from outside the data, and it is worth being clear about which half of a claim is measured.