Right. I'm sorry. My statistics are a little rusty too. Too much something-or-other, can't remember what. 533 drug-votes. 12% of the votes for types of drug (not 12% of respondents) are for alcohol.
There should be a person counter and then every question should be counted as yes or no. Then you get 93% yes, 7% no for alcohol.
"Never done any except drunk"/"drunk and stoned" are superfluous, and possibly misleading if we have the total. They put alcohol and cannabis in a special category and allow contradictory answers. 68(total) - 12 (only alcohol) = 56 (the gateway drug). It's unlikely, but possible that people will have taken drugs but not drunk, making this way of counting imperfect.
Pot and hash are the same drug. It's like saying beer vs whisky.
Who's going to ask the administrators to build a more scientific polling device? I'm talking radio-buttons, check-boxes, the works!