Charles Blow cited a stat from the Southern Poverty Law Center in his poorly sourced and argued defense of Sonia Sotomayor:
Hispanics are the largest and the fastest-growing minority group in the country. And, in recent years, they have increasingly been the victims of racial discrimination. It will be hard to paint the victims, as personified by Sotomayor, as the offenders.
A report entitled “Under Siege: Life for Low-Income Latinos in the South” that was released last month by the Southern Poverty Law Center found “systemic discrimination against Latinos” that constituted “a civil rights crisis.”
…
This finding is borne out by the F.B.I.’s hate crimes data, which show that the number of anti-Hispanic hate crimes have increased by half since 2003, while all other hate crimes have increased by 6 percent.
The Southern Poverty Law Center’s report blames this on the “relentless vilification” of Latinos in the media. I blame it on cherry-picking of the hate crime data.
Look at the chart at the upper right: I lopped off the top of the chart that was in the NYT article (click on the chart to see the whole thing at the NYT’s site) to zoom in on the “Anti-Hispanic” line. You can see the SPLC’s point ==> there was been a terrible increase in the number of hate crimes against Hispanics over that 5-year span.
But 5 years is too short to get a real sense of the trend. What do the hate crimes against Hispanics look like over a longer period? Well, they look like this:

(the SPLC uses the “Offenses” statistic, though I think “Incidents” is probably more useful)
Goodness, it looks like the peak in 2006-2007 is accompanied by similar peaks in 1996 and 2000. That makes it seem like this terrible recent increase in violence is neither unique nor a “civil rights crisis.” And it certainly doesn’t serve as any sort of defense for Sotomayor (not that it would have even if the trend had remained true).
But the real story is that the SPLC and the New York Times picked 2003, the all-time minimum in anti-Hispanic hate crimes, as their point of reference.
That’s cheating, plain and simple.
It’s bad enough that the SPLC and Charles Blow are proclaiming that increasing hate crimes against Hispanics are a current problem, even though they’re using 2-year old data, and even though that data shows that the hate crimes were leveling off in 2006 and 2007. And it’s bad enough that they touted a 5-year span as a trend of significance. These are both instances of sloppy statistics.
But it’s apparent that they also deliberately picked the time span that would make their point.
That’s not careless statistics, that’s outright lying.

Posted by geoff 
Posted by geoff
Posted by geoff 

