Using real numbers is extremely slow, and takes LOT's of code, plus it's also inaccurate - so PIC BASIC's generally don't support them. It's actually very rarely needed, you can scale your calculations and do them as integers - so to get 0.1 resolution simply multily both values by 10, then manually insert the decimal point before you display the value.
This is commonly done on all processors, including PC's, particularly for dealing with money values - you would usually convert all figures to pennies, and use integer maths. Again, because floating point maths is too inaccurate.