One thing I've learned about computer simulations is that I can make them show anything I want with a simple change in a line of code.

True if you can control the variables. In Circuit design simulators, most of the variables are set. You can change from a weak to a strong model but within process margins, but that is about it.

In this case it is easy to control the variables.

If the polling is off by a small percentage any direction the entire simulation is off.

The data must have been gathered by a poll of some sort via interviews which can be twisted severely. Never heard someone down on their luck blame themselves for the state they are in...more likely it will be this that or the other thing happened. I got sick and didn't have insurance. My boss fired me because of something outside of my control. You get the idea. Those stories were a part of the whole analysis and there's not a way to find out whether there is any truth in the narratives. Who is going to challenge the assertions? Who is going to research the "unlucky" stories? On the other hand the successful folks are more likely to lay out the truth behind how they succeeded. "Unlucky" types are less likely to do the same. It's human nature to deflect blame. If just a handful of bad narratives make it into the study the whole output is skewed into the coin toss theory.

The harder you work, the luckier you get.

Perhaps you and I have lived very different lives or grew up in very, very different neighborhoods. But I can tell you in all honesty that if I have learned anything in this life, it is that that is not even close to being true.