Let's take "Input Voltage Regulation": The value listed is 32 Typical and 200 Max. The units are mV or millivolts. The conditions are Vi belongs to the range [7..20] Volts and the temperature is 25 degrees C. Datasheet numbers assume that all of the voltage regulators produced will have parameters that follow a standard normal distribution unless there is a reason to suspect some other distribution. The typical number of 32 mV is the mean of the distribution and it means that as the input voltage is varied from 7V to 20V the nominal 5V output will change by about 32 mV. Some voltage regulators will change by less than 32 mV. The maximum value of 200 mV suggests that some voltage regulators will change their output by up to 200 mV as the input voltage is varied between 7 V and 20 V. At this point I would begin to suspect that this parameter is not normally distributed because the chance of one with 200 mV seems much greater than one with zero.
Ripple rejection. This refers to an AC signal on the input at a frequency of 120 Hz. This would be typical for a full wave bridge rectifier followed by a capacitor filter. The value is 49 dB. This means that any AC signal on the input at 120 Hz. will show up on the output attenuated by 49 dB. How much is 49 dB?
Code:
-49 dB = 20 * log (Vo/Vi)
-49/20 = log(Vo/Vi)
3.55e-3 = Vo/Vi
Vo = 3.55e-3*Vi
So 1V p-p of ripple on the input will be knocked down to 3.55 mV p-p on the output.
Output voltage regulation. Here we are changing the load current from 1 mA to 100 mA at 25 degrees C, you should see that the output voltage will change by 15 mV Typical and 60 mV Maximum. Again we suspect an asymmetrical, non normal distribution with a mean of 15 and a spread from zero to 60 mV.
The dropout voltage of 1.7V is the minimum difference between the input voltage and the output voltage for the device to regulate. With a 5V output this means the input needs to be at 6.7 Volts. What happens with the input below 6.7 volts? Well the output is a linear function of the input, but there is NO regulation going on.
Bias current is how much current is required from the input in order for the regulator's circuits to operate. You can think of this as the current required to operate the voltage regulator even when the ouput current is zero.
How crucial are these numbers? I guess dropout voltage is critical in a battery situation and ripple rejection is crucial in sensors and instrumentation. BTW I'm sure you can find better parts than these. They are almost half a century old designs. That might as well be forever in this business.