Skin-Deep: What Polls of Minorities Miss

The media's knowledge of African Americans, Asians and Latinos is woefully lacking. Opinion polls break out minority-group results from general populations, but the meaningfulness of the findings is moot at best. Lacking reliable data on the variable and textured hopes, needs and fears of minority communities, the media instead turn to personal anecdotes and self-appointed spokespeople to gauge community sentiments. That can be terribly misleading -- and risky -- when reporting on crime, police misconduct and elections. It's no surprise that racial and ethnic tensions and misunderstanding endure in cities such as Los Angeles.

When it comes to polls, news organizations have apparently decided that their typical reader or viewer is most interested in studies of a general population, be it a city, state or the nation. Polling is expensive, of course, and polling subpopulations even more so.

But there's a price to be paid for not regularly polling minority communities. The quality of the information in minority-group breakouts is inferior because the small sample sizes have large margins of error. As a result, not much can be reliably said about the differences between black men and women, or Latino young adults and seniors. Such ignorance has serious policy implications.

The standard margin of error in general population surveys is plus or minus 3 percentage points, at a confidence level of 95%. Media pollsters achieve this low margin of error by randomly interviewing a sufficient number of respondents.

Consider The Times' March 7 poll on the mayor's race. To obtain a 3-percentage-point margin of error, 1,113 Angelenos were polled. The margin of error for the poll's subgroup of 257 African Americans was 6 percentage points. To achieve that, 134 blacks in addition to the 123 in the general sample were interviewed. If The Times had not obtained that additional sample, the margin of error for blacks in its survey would have been about plus or minus 9 percentage points.

The higher standard was costly. Polling minority communities is more expensive per interview than for the general population, for a number of reasons. Because they make up a smaller proportion of the population, minority group members are harder to reach. Potential interviewees must determine who's suitable for the sample, which adds to the number of calls (a significant cost factor).