How the members were counted

What are the sources for this report?

Two undated lists of presumed academy members were obtained from industry sources; those lists comprised 5,500 names, some with branch affiliations. Since 2004, the academy has annually published the names of those invited to become members, providing 1,026 potential inductees, some 700 not on the other lists. The Times combined these three sources into a master list of more than 6,200 names. According to the academy, the organization had 5,765 active voting members as of Feb. 16. This study looked only at active voting members.

How did The Times confirm those names?

A team of more than 20 reporters and researchers then used multiple sources to confirm the identity of active academy members. Those methods included in-person, telephone and e-mail interviews; confirmation from members' publicity and talent agencies and managers; media reports; personal biographies, social networking profiles and resumes; lists of academy members who participated in academy fundraising; and academy publications.

How did The Times determine the members' demographics?

The age, sex, race and Oscar nomination and win history of people on the master list were determined through public records, commercial databases, guild publications, the oscars.org website and individual interviews.

Did The Times identify all the academy members?

In fact, The Times' list has more names than the academy's reported membership. According to information provided by the academy, the organization had 5,783 active voting members as of Dec. 27, 2011, as well as 152 non-voting active associate members. (There were also 534 retired voting members and 139 retired associate members.) Some members have died since Dec. 27, 2011. The Times reporters confirmed more than 5,100 members, about 88% of the active voting membership reported by the Academy.

In some branches The Times confirmed more names than the academy said were actually members. Because only academy officials know for sure who is an active voting member of the organization, these discrepancies could not be resolved.

Considering the discrepancies, how were the statistics derived?

To calculate percentages, The Times assumed that the small fraction of those who could not be confirmed as members or whose demographics were not obtainable were, in fact, very much like the academy as a whole. Thus, in each branch The Times divided the number of people whose demographics were known by the total on its list of confirmed and presumed members, a denominator that was usually higher than the academy's reported membership.

So how accurate are the statistics?

The Times approach was conservative. Most likely, the actual percentages are more white, more male and older than The Times' estimates. That's because those on the list who were hardest to authenticate were most often members of long standing. They were more likely to be older, more likely to be white and more likely to be male than the current academy.

In the interest of complete transparency, The Times made a secondary calculation assuming that every potential member who could not be accounted for was a non-white woman in her 30s. In this calculation, those whose demographics were known were divided by the academy's reported membership. In addition, if The Times' list was greater than the academy's, the difference was subtracted from the white and male members and the oldest prior to the calculation. Those are published below.

This project was reported by John Horn, Nicole Sperling, Rebecca Keegan, Steven Zeitchik, Amy Kaufman, Gina McIntyre, Elena Howe, Julie Makinen, Deborah Vankin, Laura Randall, Ben Fritz, Susan King, Reed Johnson, Emily Rome, Mark Olsen, Oliver Gettell, Jasmine Elist, and Dima Alzayat. Research and statistical support was provided by Doug Smith, Sandra Poindexter, Cary Schneider, Robin Mayper, Scott Wilson and Kent Coloma.