In “Point Blank” Gary Kleck writes:

“One way one might crudely and partially control for United
States-Japan cultural differences is to compare homicide rates among
Japanese-Americans, who live where guns are plentiful, with the homicide
rates of their presumably culturally similar brethren in Japan, where
private gun ownership is nearly nonexistent. Certainly this pair of
populations is more comparable than the population of Japan compared
with the entire U.S. population.

Not necessarily. The US does not accept immigrants with criminal
records, so this group of Japanese-Americans will be less likely to
commit homicide than the general Japanese population.

Up through 1979, the FBI reported
homicide arrests sorted by racial breakdowns which included “Japanese.”
For the period 1976-1978, 21 of 48,695 arrests for murder and
non-negligent manslaughter were of Japanese-Americans, or 0.04% (U.S. FBI

A 95% confidence interval is 0.03% to 0.07%.

Applying this fraction to the total of 57,460 homicides
yields an estimate of 24.78 killings by Japanese-Americans for 1976-
1978, or about 8.26 per year.

This assumes that there is no racial bias in the arrest patterns of any
US police force.

With 791,000 persons of Japanese ancestry
in the United States in 1980 (U.S. Bureau of the Census 1984), this
translates into an annual rate of 1.04 homicides per 100,000 population.

There are two different definitions of “Japanese” being used here –
Japanese ancestry (note that a person can have multiple ancestries)
and police ticking a box marked “Japanese” on an FBI form (what do
they do if they don’t know/care?).

Considering all the factors above, we can guess that the homicide rate
for Japanese-Americans is somewhere between 0.5 and 3.0 per 100,000 pop.

For the same 1976-1978 period, the annual homicide rate in Japan
averaged 2.45 (United Nations 1982, pp. 192, 718).

No it didn’t. Kleck can’t read. That’s the rate for BE50 “All other
external causes”. The reference does not tell us the homicide rate
for 76-78, but if Kleck had turned to page 777 he would have discovered
that for 79 they split category BE50 into AM54 (Homicide) and AM55 (Other
violence), with rates of 1.0 and 1.5 respectively. That is, homicides
were 40% of the “All other external causes” deaths. If this was true
for 76-78, the homicide rate in those years was 0.98 per 100,000 pop.

Thus, crudely
controlling for Japanese culture in this way indicates that in Japan,
where civilian gun ownership is virtually nonexistent and gun control
laws are extremely strict, the homicide rate is 2.3 times as high as it
is among Japanese-Americans living where guns are easily available and
gun laws are far less restrictive.”

The Japanese homicide rate is probably lower than that of Japanese
Americans, but there is insufficient data to have any certainty.