OK, from other side of fence...
For smaller organisations I would be wary about these forms because they can end up going hand in hand with application and you don't want your private life bandied about. But with larger organisations I would be very comfortable giving this out because I have seen how it is used.
I have been working for a large organisation looking at diversity. I was allowed access to this data and was able to look at what stage people (all of them anonymous) joined the firm, when they left, and how quickly they were (or weren't promoted). I was able to find statistical significance at certain points of promotion and could map out when the diversity agenda was going awry by department. I could see some quarters were being very diverse in promoting people and encouraging part time work and I could see some departments that became progressively whiter and more hetero the further up the chain you were. It is very easy to say that you recruit equally, but when (giving a random and not accurate example) 15% of your graduates are Jewish but after 5 years only 1% is Jewish, then you know something is odd and needs fixing. Likewise it is easy for a manager to say that he supports part time working but then you find that all the people with dependants in his team work full-time whereas in other departments the majority with dependants work part-time, then you know you have something to look at in more detail. Do they work full time because they are so engaged they can't bear going part time, or because their manager is a tosser and won't let them? You don't know the answers but you do find the questions to ask. It is easy for directors to assume that women don't get to the board because they choose to leave when they have children, but the questioning for them gets much more awkward when you are able to say: well, even when adjusted for this, I found that it still takes 4 years longer for a woman without kids to get promoted than for a man, so why is that?
Most of the really large organisations monitor this information in miniscule detail and it is never at identifyable individual level. The data in the firm I was involved with was being used to implement diversity policies, to track them, and to check what was happening when departments consistently fell outside the bell-curve. In the org I looked at they have decided to investigate every single senior promotion which is statistically significant. So if 15% of staff are Jewish at one level then they expect the people to be promoted to include 15% Jewish representation, and if it doesn't then they will ask why and make sure the answer makes sense, and it would only be at those points (where you are actively trying to get minorities up) that you would start investigating the team, the management, and the promotion process and potential issues.
And to be honest, one of the most frustrating things in doing all of this was that you never worked with a full pack of cards because often up to 35% of people hadn't ticked the boxes and therefore couldn't be tracked. It was like trying to moinitor the success of eliminating a disease when loads of people hadn't been vaccinated - it just messed up what could have been very clear black and white information.