It's actually not super uncommon for management to use statistical tools for this, although they may not realize what exactly is going on. For example there ar

It's actually not super uncommon for management to use statistical tools for this, although they may not realize what exactly is going on. For example there are a couple of tools whose names escape me at the moment that extend or wrap MS Project but have a statistical time estimate piece based on quintile time estimates (e.g. your engineers give you 50% and 90% conf estimates, it applies simple modeling). I've also been at one shop where we used a more sophisticated modeling approach, but built in house. Reception was warm in at least most of management. So my experience isn't that these tools don't exist, or that management is unwilling to use them - but rather that management has trouble accepting the implications of using them properly. Specifically, if any of the inputs of the model change, you should update those and re-run the model. With linear (e.g bullshit) GANTT, you just say "oh, we missed this week of work, move the endpoint forward a week". With more careful modelling adding an unexpected dependency or a couple days work in the wrong place can suddenly add 3 months to your worst-case model. Exec's really don't like that happening without a whole song and dance about why, so there is a tendency to "freeze" the models in ways that they become progressively less useful. Worst case they are asymptotic on meaningless.