Is there any justification to treat a sum of ordinal scores as a ratio score?

More than once, I noticed that researchers sum ordinal scores of e.g. items of a psychometric test to gain a total score. This score is e.g. calculated for multiple study participants and metrics like mean or standard deviation are deduced. Isn't this a mistake or am I missing something?


TS Contributor
The sum is often is treated as interval scaled, but nearly never as ratio scaled (the only exception I am personally aware of is the awkward use of the HAM-D Hamilton scale for depression).

As to the question of why the sum of several ordinal scaled items can be treated as interval scaled, discussions of this can be found on the net, e.g. here or here or here .