(I quote an email that a scientist (Dr. Van Arsdale) sent me in this post quite a bit, so I'm just crediting him here.)
Now, from my perspective, IQ tests are meaningful to an extent, but I think scientists like J. Philippe Rushton and Charles Murray have been misleading people on this for far too long.
The short answer to your question is: no, IQ tests do not measure "intelligence." Intelligence has an ontological reality, certainly, but in practice, intelligence is inevitably epistemologically defined. Arthur Jensen and Rushton's 'g factor', for example, is merely a statistical construction based on standardized test results. It exists, but its existence is not necessarily connected with biology in any meaningful way.
In regards to race, to get big, consistent differences across races you'd need to have allelic variants with really large effect size. You would essentially need to prove that the complex admixture of genes for intelligence haven't been acquired by non-Eurasian/Jewish populations, when in fact, lots of genes had been admixed INTO those populations for the past 8,000 years. Keep in mind that while we've found some genes for IQ, we're still a long way from mapping the human brain, so no one with full confidence should claim that intelligence genes are distributed discriminately. Also, given the really large number of genes that must inevitably be involved with the varying complex forms of intelligence that exist, unique small effect variants would simply be swamped in the polygenic nature of the phenotype.
We also have a wealth of evidence that intelligence is highly malleable to a multitude of environmental factors; not just genetic.
In short, the claim that there are innate differences in "intelligence" between biological "races" is, quite simply, an absurd, biased interpretation of data. This myth has been debunked many times, and the Heritage Foundation was doing itself a favor by distancing itself from this pseudoscience.
IQ is mutable. Downward mutability has possible causes in heavy-metal pollution (the poor experience more pollution because they are more likely to live near smelters and other dirty industries), trauma from near-drowning to head injuries, and of course unstimulating environments. IQ might be mutable upward because of experiences, which might not be as significant as trauma and poisoning from pollutants. If that has connections to ethnicity, then such may manifest itself in 'racial' differences that become proxies for poverty.
Some people are just simply better at raising their kids to compete intellectually. Do children learn to trust or distrust well-intentioned authority such as school teachers? That may be the difference between kids learning in school and getting little from it. But such depends upon parents being trustworthy.
Is it meaningful? People with below-average IQ are unlikely to get much out of matriculation in college. (Of course, people with bad study habits would get little out of a college education, too). One fast-food company that I heard of in a business management class gives a basic IQ test to people applying for jobs as fast-food location managers. The optimal IQ for an applicant was about 90, and an IQ much above that indicated on the whole poor matches for the job.
OK, so the work isn't comparable to being the director of a scientific research institute. Below 90 one has people who can't cope with the rigid rules and the paperwork. Around 90, rigid rules fit one well, and the paperwork is something of a challenge. In view of the sorts of people who remain workers in such a place for an extended time (dullards), someone with an IQ near 90 can better relate. Someone with an IQ in the 120s might learn the routines quickly but would see the job only as a stepping-stone to something more intellectually-satisfying. People like challenges that validate themselves.
There is an optimal IQ for just about any job. For a lumberjack the level might be low because someone who thinks too much might get excessively cautious for the job description. Assembly-line work is far more a matter of hand speed than a deft mind, and a person who does such a job whose mind is working on something other than the repetitive work of feeding a punch press is easily distracted. A house-painter is near average because although the painting is easy, estimating amounts of paint and calculating costs requires some mathematical sophistication; besides, one needs to manage one's time on the job. Physician, attorney, research scientist, accountant, engineer? You get the general idea.