In the course of an otherwise useful article about FDR's approach to the great depression, the New York Times offers a misleading assessment of the unemployment rate under Roosevelt:
During the 1930s, the unemployment rate fell somewhat under Roosevelt, but remained stubbornly high, averaging more than 17 percent for the decade.
There are a few problems with this.
First, the Times doesn't provide a starting point - what was the unemployment rate before Roosevelt took office? 17 percent sounds awfully high, but it could actually be an impressively low figure if the starting point was much higher. Which it was - 25 percent in 1933, Roosevelt's first year in office; 24 percent the year before. (The Times does note the 25 percent starting point later in the article, too late for it to provide effective context for the 17 percent figure.)
Second, in assessing the efficacy of a program begun in 1933, it is basically meaningless to use average unemployment rate for the entire decade. We can hardly credit or blame the New Deal for the unemployment situation in place before the New Deal began. What matters isn't the average unemployment for the entire decade, which includes several years before Roosevelt even took office - what matters is the trend line. Did unemployment go up or down? How much? That's what matters; raw numbers - particularly raw numbers averaged over the entire decade - are badly misleading.
For an example of how misleading it is to use average unemployment numbers to assess the effectiveness of a president in combating unemployment, we need only look to the conservative talking point that average unemployment under George W. Bush was lower than average unemployment under Bill Clinton.
Maybe true, but meaningless - that measure credits Bush for Clinton's success, and penalizes Clinton for the failures of Bush's father.
See, the average unemployment rate under Clinton is artificially high due to the high unemployment rate when he took office; Bush's is artificially low due to the low rate when he took office. If you look instead at trend lines, you see that unemployment went down under Clinton and up under Bush. That's far more useful in assessing the two presidents effectiveness in fighting unemployment than average numbers that paint a misleading picture.