Well it's easy enough to test. Select any input, '0' would be best and measure the input voltage and the output voltage, they should be the same. The transmission gate has about 3 - 400 ohms resistance @ 5Vdc Vcc. Pin 13 is the 0 input and pin 3 is the I/O pin (the device is unique in that it is totally bi-lateral... to at least a Mhz) so you might load the pin slightly with an appropriate value resistor and see that the change is equal on both pins, verify that 6 is grounded or at least pulled down 100k is ok. and also one of my mistakes was not to verify the state of the switch address pins as well, I forgot pull-down resistors. When you have an issue like this a DVM can be your best friend. It 'should' work, the chip is reasonably bullet proof and it can source and sink a little current. If you put pull downs on the data address lines they can be pulled from the processor and tested that way easily. IMO
Doc