Nailing your measurement
Say you test a question about happiness, and most of the responses come back closely clustered, but not really capturing happiness (e.g. maybe it's capturing satisfaction)... what is this measurement said to be...
reliable but not valid