Los Angeles has been called the murder capital of the United States. But a new study of homicide rates that takes poverty into account challenges this notion: By this new measure, Los Angeles actually falls nearer the middle of the pack for homicide rates -- 24th on a list of 67 large American cities in 2002, and 42nd on the same list in 2003.

That put Los Angeles just below Chicago and Dallas for 2002, and just above Denver and Philadelphia -- although there were 653 killings in Los Angeles that year, the most of any city in the nation. After Los Angeles homicides fell by 23% in 2003, the city's ranking dropped even further, placing it closer to Tucson and Milwaukee. The finding suggests that Los Angeles' large number of homicides can mostly be explained by its high poverty rate and other demographic characteristics.

Social scientists are able to calculate the expected number of homicides for a city based on certain social and economic factors. Hoping to refine conventional crime statistics, this study examines homicide rates that are in excess of the expected rate.

Using that method, researchers found that Los Angeles ranks lower in homicides than many other cities -- including smaller, wealthier ones, particularly San Francisco.

Researchers were surprised to find that, if differences in wealth, demographics and racial composition were taken into account, San Francisco ranked first in the nation in homicides in both 2002 and 2003.

San Francisco's homicide rate of about nine per every 100,000 people is moderate by traditional standards. But the rate is strikingly high given San Francisco's wealth and low-risk demographics, the researchers said.

The broader finding, according to researchers, is that big cities may not be the crucibles of violent crime they are often assumed to be.

Instead, the study suggests, homicide rates are high in some of America's large cities largely because that's where poor people live: Poverty and homicide tend to go hand in hand.

With a few exceptions, "the blighted urban core -- the classical crime areas," didn't pop out as the hotspots, said George State University professor Robert Friedmann, one of the study's researchers.

Conventional crime-data reports, besides lagging several years behind, are much less refined than those routinely used to measure educational or economic trends, Friedmann said. Comparisons of cities based on raw per capita rates unfairly pit wealthy San Diego against economically distressed Detroit, he said, yielding little practical insight.

Besides San Francisco, a number of other medium-sized cities emerged with higher-than-expected rankings. These included Riverside; Anaheim; San Jose; Santa Ana; Omaha; Raleigh, N.C.; Albuquerque; Anchorage; and Seattle.

Factors other than socioeconomics or demographics -- cultural or policing issues -- might be pushing homicides up in those cities, said Richard Rosenfeld, criminology professor at the University of Missouri-St. Louis. Rosenfeld conducted the study with Alfred Blumstein of Carnegie Mellon University and Friedmann.

The study was funded by the National Institute of Justice for the Improving Crime Data project. The three researchers calculated homicide rates in cities with populations of more than 250,000, taking into account such factors as male unemployment rate, poverty, single-parent homes, median income, length of residency, divorce rates, and the percentage of blacks in the population.

(Blacks are killed at disproportionately higher rates than other groups even when the data are adjusted for poverty, so the researchers adjusted their findings to compare cities with small black populations to those with large black populations.)

Researchers for the study engaged in a kind of handicapping. They first predicted what cities' homicide rates would be, based solely on their socioeconomic and demographic characteristics, then compared the predictions to the study's findings.

For some cities, such as Washington and New Orleans, adjusting for socioeconomic factors did little to change their national ranking. Both kept high slots in the years studied, suggesting their homicide problems go deeper than the high-risk demographics.

But other cities notorious for homicides surprised researchers.

Detroit, Cleveland and Newark, N.J., for example, have high homicide rates -- an astonishing 42 per 100,000 people in 2002 in Detroit's case. But when extreme poverty and other factors were taken into account, homicide rates in these cities were actually lower than researchers expected. New York City also had lower-than-predicted rates.

The finding suggests something is working to suppress homicides in those cities, the researchers said. This was news to some police officials. "It's kind of shocking," said Lt. Russell Solano, a homicide detective supervisor in Detroit. "But what I want to know is, do they have suggestions for bringing it lower? We need ideas."

Officials in cities that came out high on the adjusted scale also voiced surprise. Some were quick to argue that taking into account variables such as poverty doesn't account for the idiosyncrasies of such quirky places as San Francisco -- with its vast spread between rich and poor -- or Anchorage, where the usual rules linking wealth and employment are turned upside down by a seasonal labor market.

Such a study "doesn't take into account the complexity of the modern city," said Peter Ragone, spokesman for San Francisco Mayor Gavin Newsom.

Interviews with police officials yielded no obvious explanation of why cities differ.

Lt. John Hennessey, officer in charge of the homicide detail in San Francisco, described conditions facing homicide investigators in his city which are virtually identical to those in Los Angeles -- or those of Detroit, for that matter. In San Francisco, Hennessey said, detectives juggle high caseloads as they are deluged with street murders involving black victims, committed in neighborhoods where witness cooperation is difficult to secure.

Los Angeles remains a relatively dangerous city, with a crude homicide rate of about 17 deaths per 100,000 people in 2002, and 13 per 100,000 in 2003. But poverty and demographics account for most of this rate, the researchers concluded. And in 2003, Los Angeles actually had fewer homicides than would be expected based on its demographic profile.

One implication of this may be that a focus on Los Angeles' gang culture may be misplaced. The study cast doubt on the uniqueness of such factors, suggesting that Los Angeles may not be so special after all. "It is important not to make too much of how distinctive L.A. is," Rosenfeld said.

Study authors acknowledge their analysis is limited, relying on only two years of data. But they argue that it is a starting point for further research.

Eventually, Rosenfeld said, the researchers hope to help isolate factors that authorities control, such as resources for homicide investigations, the number of patrol officers, or the degree of police corruption, as part of the effort to lower homicide rates.

There is a further aim, Friedmann added: To underscore that there is more to homicide than police.

Police departments are often taken to task for homicide rates, but it would make about as much sense to hold health or education departments responsible, Friedmann argued.

Many police seem to agree.

Homicide "isn't just a police issue, it is a society issue," said Anchorage Police Capt. Thomas Nelson, echoing the comments of police in other jurisdictions. "The police department doesn't control poverty. It doesn't control jobs."