My understanding is that the big hospital spirometers typically use a direct volume measurement so that a fake high is less likely. The only inaccuracy would be the volume measurement itself or any air leaks in the system (the tester's mouth, lips, and nose being part of that system).
Home spirometers use a flow meter and then derive volume. In addition to the problems of inaccurate measurement and leaks, the flow meter device also has to assume that the flow is moving in a laminar fashion (smooth-even flow) and then do some calculation to derive volume. However if you blow hard enough, or sporadically, or have an obstruction, then you can create a turbulent (chaotic) flow across the meter which causes the meter to calculate a poor reading.
I'm sure the home spirometer makers try to design everything so that it stays laminar (large pipe, smooth surfaces, training/feedback to help the user blow in a smooth fashion) but with the variability of a young users, and a constraint to make the device reasonably priced, you are probably going to get some inaccurate readings due to a turbulent test.
Below are my thoughts on the science behind it. I'm spit balling a little bit here and would love someone like LittleLab4CF to chime in as well.
With laminar flow you can easily predict the speed profile across the sensor (it looks a flat 3D parabola, with flow faster on the middle than at the edges of the pipe.) To better illustrate this, imagine water coming out of a garden hose at a nice smooth stream. You can fairly easily measure the speed of the water (with something as simple as a water wheel) and because you know what the flow profile is uniform and consistent. One you have the speed and cross section you can calculate the flow volume. [Speed (m/s) X Area (m^2) X duration (s) = Volume (m^3)]. Granted even with this method, you’ll need a flow sensor that is reasonable good at measuring flow speed (which doesn't come cheap). Additionally, air density, humidity, pressure, pipe materials, cross section, can all contribute in making the flow profile hard to control, highly variable, or just plain difficult to understand.
To illustrate turbulent flow, imagine water spraying out of a fire hose at full blast. With a spraying stream you can't assume any of the particles of water are moving at the same speed or direction, any single particle speed measurement reading will not be illustrative of the overall flow rate. But you meter will still measure some speed, run the laminar calculations, display a reading, but the reading will be inaccurate.
