An in-depth look at Age & Gender Inference: what it is, how it works, and what data it uses.
What is Age & Gender Inference?
Age & Gender Inference enriches donor records in the Dataro Platform with an inferred age and gender for each incomplete contact record in a nonprofit’s database. Inferred Ages & Genders appear as highlighted purple text in Contact & List Views in the Dataro Platform.
Age & Gender Inference in List View:
Age & Gender Inference in Contact View:
What data is used to infer age and gender?
- Government & Population Data: Dataro is able to look at local name and age distribution (primarily through the US Social Security name-age database) to gather data on how likely a person with a certain name is to be a given age and gender.
- Your donor data: From donor records that include provided age and gender information, Dataro can gather age & gender trends among your donors (for instance, if 25% of known donors in your database are 20-30, it could follow that 25% of all of your donors are 20-30).
- Nonprofit sector-wide data: Like all of Dataro’s models, Dataro’s Age & Gender inference is built on a foundation of fundraising data from hundreds of organizations and over 100 million donors. 25% of those individuals have age records, providing sufficient data to train a machine learning model.
How accurate are inferred age and gender?
Inferred Age
Inferred Ages are not exact, Dataro rounds to the nearest 5th year. In our testing, on average the model predicted accurately within +/- 11.2 years of a donor’s true age.
For practical purposes, you can treat Inferred Age as an indicator that a donor is likely within a 5 year range on either side of that age.
Inferred Gender
If Dataro’s model is not reasonably confident, the Gender field will read “Unknown” in the donor record.
Do other Dataro models use inferred age and gender?
Yes! Currently, inferred age and gender are referenced by Dataro's Major Giving and Gift in Will models.
Disabling Age & Gender Inference
We understand that not every organization is comfortable with collecting age and gender information or inferring this information when it is not supplied by donors. If this is the case, just request that your Customer Success Manager disable age and/or gender inference in your environment.
While our client's data is missing ~75% of ages, our Data Pool has age for ~25% of records (>25M individuals). This provides more than sufficient data to train a machine learning model.