Homicide rates doubled in the United States and Canada within 20 years of the spread of television, says a psychiatrist's seven-year statistical study which he contends shows a link between video and violence.

Other experts cautioned Friday, however, that while Dr. Brandon Centerwall's study may have merit, blaming television for the rise in homicide rates is too simplistic.Centerwall's study, published in the April issue of the American Journal of Epidemiology, is billed as the first to look at the statistical relationships between exposure to television and acts of violence for the entire country.

Centerwall's study also suggests that as many as half of other violent crimes - including rapes and assaults - are related to TV exposure.

"Television is a factor in approximately 10,000 homicides each year in the United States," said Centerwall, a member of the psychiatry faculty at the University of Washington School of Medicine.

"While television clearly is not the sole cause of violence in our society, and there are many other contributing factors, hypothetically if television did not exist there would be 10,000 fewer homicides a year."

To arrive at this conclusion, Centerwall studied the white population of South Africa, where television was not introduced until 1975. Using statistics from 1945 to 1974, he compared homicide rates among South African whites to the rates among U.S. whites and the entire Canadian population.

He found that homicides remained roughly flat in South Africa before television was introduced. In the United States and Canada, however, homicide rates doubled within 20 years after the widespread introduction of television.

Centerwall also said the homicide rate among South African whites in 1983 - the last year for which statistics were available - was 56 percent higher than in 1974, indicating a trend similar to what occurred in the United States after the introduction of television.

Ronald Kessler, a professor of sociology at the University of Michigan, said Friday that the major problem with Centerwall's study is that when television arrives in a society, many other things come with it.

"It's hard to imagine, when you look at what's been happening in South Africa, that the main change between the 1950s and 1970s has been television," he said.

Dr. Michael Rothenberg, professor emeritus of psychiatry, behavioral sciences and pediatrics at the University of Washington, said he believed Centerwall's study "should be taken very, very seriously."

Rothenberg, co-author of the current edition of "Dr. Spock's Baby and Child Care," added that people who commit homicide do so for many different reasons, and Centerwall's figures struck him as high.

"He compared entire populations, so the statistical study is more accurate," Rothenberg said. "But precisely because they are entire populations, there are multiple variables."

Centerwall said regions of the United States that had widespread television before the rest of the country also saw earlier increases in homicide rates.