Data is often treated as a plural noun in writing related to science, mathematics, finance, and computing. Elsewhere, most English speakers treat it as a singular mass noun. This convention is well established and widely followed in both edited and unedited writing. Keep in mind, though, that some people consider the singular data incorrect. This view is based on a misunderstanding of how English develops, but those who hold it tend to feel strongly about it, so we might approach data with caution in writing for school or work.
The reason some people believe the singular data to be incorrect is that data is a plural word in Latin, its singular being datum, meaning a thing given. The problem with this view is that data is an English word when English speakers use it, and we’re not required to continue following Latin rules with words that have been in English for centuries. Some Latin forms are preserved by convention, but the plural data is not one of them, and those who wish to make it conventional are fighting for a lost cause.
How lost is the cause? Using Google’s various search tools, we find that there are about four instances of ”data is” for every “data are” overall on the web. The ratio is about 6:1 in newswriting from this century and about 3:1 in published books from this century. ”Data are” still has the edge in scholarly writing (where the ratio is practically 1:1), which makes sense because those searches covers large amounts of scientific, medical, and financial writing, where the plural data remains customary.
People love their pet language peeves, though, so the view that the singular data is wrong in all contexts is likely to live on indefinitely among a handful of English speakers.
The view that the singular data is incorrect still holds sway over some copyeditors, which is why, as these examples show, the word continues to appear both ways in mainstream 21st-century newswriting:
Japan Economic Data Worsen [Wall Street Journal]
[B]y the time the data is published, copycat investors would have made an annualised loss of almost 10 per cent. [Financial Times]
Data are still being analyzed but will be ready to present at the conference. [Denver Post]
GDP Data Shows Japan’s Economic Growth Slowing [LA Times]
Obama’s campaign staff members said that all that data is not gathered to shape the message. [Washington Post]
Money data are not everything. [Telegraph]
The plural data is typical in scientific and financial writing—for example:
From a statistical point of view the data are related to a nonlinear mixed effects model involving repeated measures. [British Journal of Clinical Pharmacology]
We show that Howrey’s method for producing economic forecasts when data are subject to revision is easily generalized to handle the case where data are produced by a sophisticated statistical agency. [Journal of Business and Economic Statistics]