Without knowing the assumptions on which the statistics are based (e.g. different definitions of poverty, different assumptions concerning future economic or demographic developments) one cannot evaluate them. Knowing the assumptions, one can understand their differences. For example, at the time of my research at the Institute for Futures Research (IFR), unemployment in South African was officially about 25 percent of the total economically active population, while our estimate was about 45 percent. Both estimates were correct in terms of their definitions and underlying assumptions. To avoid confusion, we referred to our estimate as un- and under-employment and we always quoted both estimates and explained the reason for their difference.
Unfortunately, as a researcher or consumer of research, one cannot always discern the underlying assumptions of quoted figures, because of omission (e.g. researchers are sloppy about them, don’t quote them or are even unaware of them) and commission (e.g. researchers promote a deliberate ideological position and select evidence accordingly). Personally, I have sinned on both accounts on occasions.