In the name of science

Yes, quite frankly. A human life is quite precious. That's the sole and strong reason for my argument.

And of course your results in your experiment would have been different. However, telling the users the purpose and telling the users the consequences are two entirely different things. They are not the same thing.

Rarely do you ever see experiments these days on humans that can potentially be life-threatening. Things like drug-testing go through many trials in bacteria cultures, then small mammals, and if it's deemed safe, then you've got human subjects. The consequences of such an experiment, if any, should be minimal to the subject to avoid subjects conceding from the experiment. If you only have the 'resolute' subjects left, then you will have a substantial amount of bias in your data.