What if half the men in science, engineering and technology roles dropped out at midcareer? That would surely be perceived as a national crisis. Yet more than half the women in those fields leave -- most of them during their mid- to late 30s.

In this month's Harvard Business Review, Sylvia Ann Hewlett, Carolyn Buck Luce and Lisa J. Servon describe the Athena Factor, their research project examining the career trajectories of such women. Hewlett, founding president of the Center for Work-Life Policy in New York, told Kathleen Melymuka about what they learned.

Your research shows that there are more women on the lower rungs of science and technology fields than most people suspect.

Women are actually excelling in science, engineering and technology, despite the fact that the schools are not very good at encouraging them. Many don't just survive the educational process but get some distance in terms of careers. The story is very encouraging in the early run. Between ages 25 and 30, 41 per cent of the young talent with credentials in those subject matters are female. It's a more robust figure than many suspect. That's the good news.

What happens later?

The bad news is that a short way down the road, 52 per cent of this talent drops out. We are finding that attrition rates among women spike between 35 and 40 -- what we call the fight-or-flight moment. Women vote with their feet; they get out of these sectors. Not only are they leaving technology and science companies, many are leaving the field altogether.

How many women are we talking about?

We reckon that maybe a million well-qualified women are dropping out in that age range. We reckon that if you could bring the attrition rate down by 25 per cent, you would hang on to about a quarter of a million women with real experience and credentials in these fields -- fields that are suffering a labour shortage.

Based on the demographics, it seems likely that they leave to start families. Is that what happens?

No. I'm not trying to pretend that work-life balance is not important, but we found four other more important factors about the culture and the nature of the career path. We call them "antigens," because they repel women.

Tell me about those.

The most important antigen is the machismo that continues to permeate these work environments. We found that 63 per cent of women in science, engineering and technology have experienced sexual harassment. That's a really high figure.

They talk about demeaning and condescending attitudes, lots of off-colour jokes, sexual innuendo, arrogance; colleagues, particularly in the tech culture, who genuinely think women don't have what it takes -- who see them as genetically inferior. It's hard to take as a steady stream. It's predatory and demeaning. It's distressing to find this kind of data in 2008.

Is this data global or national?

We studied private-sector employers in the US, and then we looked at three large, global companies with women working across the world. We also did a bunch of focus groups in Australia, Shanghai and Moscow. The data were pretty consistent. Actually, India is a little better than the US. But there's not much variation across geography.