Governments need to plan for roads and schools and other services, so they estimate the number of people who might be around in coming years to use them.

A new study may help make the educated guesswork a little more scientific. The study, funded by the National Institutes of Health, proposes a new method of predicting fertility rates by using a new statistical technique.

The new method mathematically compensates for uncertainty and should allow governments that more specific information on large-scale population changes that they need to plan for infrastructure and services.

The conventional method of predicting the fertility rate relies on the average number of times a woman gives birth during a lifetime and the estimated change to the number as a woman ages. Analysts create a range by adding and subtracting .5 children to the average rate predicted. But they could not calculate how likely it was that variations would actually occur.

The new method uses a statistical formula to take into account historical fertility estimates and the likelihood of future trends. It uses the historical rates for the country as well as 195 other countries, since fertility patterns are the same in all countries.

Authors of the study say the new method has a 1 in 10 chance that the fertility rate will be greater or less than that actually observed. The findings appear in the journal Demography.

“More accurate forecasts of fertility trends will allow officials to better plan for a country’s municipal, economic and social needs,” said the program official for the study, Michael Spittel, in a statement. Spittel was formerly of the Demographic and Behavioral Sciences branch at the National Institute of Child Health and Human Development, the NIH institute that funded the study, and now at the NIH’s Office of Behavioral and Social Science Research.

Researchers came from the National University of Singapore, the University of Washington in Seattle, the United Nations, South Africa’s University of Witwatersrand and the INDEPTH research network.

They looked at fertility rates over time to help develop the model. Women used to give birth to an average of six or seven children, then five children and then a stable two, or replacement level. In the last five years, 20 countries have entered this phase. The United States fell below the replacement rate and then recovered to two today.

Using this information, the researchers were able to forecast trends through 2100 using the new method.