Here is the CSM:
The practice is common among all religious groups - Hindus, Sikhs, Jains, Muslims, and Christians - but appears to be most common among educated women, a fact that befuddles public health officials and women's rights activists alike.
"More educated women have more access to technology, they are more privileged, and most educated families have the least number of children," says Sabu George, a researcher with the Center for Women's Development Studies in New Delhi, who did not participate in the study. "This is not just India. Everywhere in the world, smaller families come at the expense of girls."
I don't see hard and fast evidence in this statement or elsewhere CSM article for the claim, but several sources in the article suggest a correlation between wealth and education, and higher incidences of female foeticide. It makes some sense: wealthier families can afford the multiple ultrasounds and the abortion procedure. And the strongest evidence for the claim comes from the fact that birth ratios tend to be the most skewed in relatively prosperous agricultural states like Haryana and Punjab.
It doesn't account for everything, of course -- boy/girl birth ratios in relatively prosperous states in South India are at or near normal levels (according to this map based on recent census information). So there definitely are some cultural factors at work, but it's not as simple as "Punjabis are more patriarchal."
Incidentally, the Indian Medical Association, though it has condemned female foeticide, is questioning Lancet's claim that 10,000,000 female foetuses have been aborted via sex-selection in the past 20 years, of which 5 million abortions are said to have taken place after the procedure was banned in 1994.
According to the BBC article, the IMA claims that since 2001 a crackdown on ultrasound equipment has led to a dramatic drop in female foeticide. But it will be another five years (the next census) before we have any reliable data on that, so there's almost no point in even discussing it at present. (Or not: are there other sources for sex-ratio statistics that might be used to sort this out?)