# Some Issues with Cafe Society Polls

At this moment I am collecting data on two topics and have started a third thread that’s headed toward a poll eventually.

The topics are:

Greatest Actors
Best Western Movie
Favorite All-Time TV Show

In each case I’m confronted with how to deal with a large list of candidates, over 50 in each case.

If it comes down to it, I will even do a poll on proposed solutions, but I wanted input from you folks first.

Which of these ideas makes better sense, assuming the polls ought to express the will of the majority of Dopers:

1. Just trim the lists down based on number of replies in the preliminary thread(s)
2. Have some preliminary polls with entries arranged alphabetically in each pre-poll for some relatively small number per poll
3. Have preliminary polls but weight the entries some way – to be determined in this thread
4. Some better idea.

Items 2 and 3 could be seen as a sort of “bracket” like the one used in the NCAA Basketball playoffs. “Seeding” is the big issue as I see it.

The polls I’ve seen so far are for moderate sized numbers of options. Sampiro’s poll on birth-year, birth-month and birth-decade is the biggest group of choices I’ve seen and it seems to be working well. But I have way more potential entries than even that one.

I suspect future polls could profit from some thoughts on how to be as thorough and yet as fair as possible while trying to deal with some “Big Issue” with a lot of potential entries.

What do you think?

Such a hot topic!

I’ve had an idea for how to break down these big lists in such a way as to reduce bias of multiple mentions and yet still give all entries a chance to be voted on.

1. put all entries in alphabetical order
2. divide the total number of entries by 25
3. set up that many separate sub-groups
4. start at the top of the alpha list and assign each item to a subgroup in the order it appears in the alpha list
5. continue down the alpha list until all items are in a sub-group

Example: say there are 26 items, one for each letter, and instead of dividing by 25 we divide by 5. Each subgroup gets 5 plus 1 in Group1.

The assignment would be:

G1 a f k p u z
G2 b g l q v
G3 c h m r w
G4 d i n s x
G5 e j o t y

The five subgroups would each constitute a separate poll. Top two in each subgroup poll would go into the final poll.

How’s that?

The tricky thing about that is, what if in some cases 3 entries under a subcategory would happen to have been the top 3 overall? There would need to be a some fiddling with the groups to prevent something like that

Good point. That suggests a pre-sort by frequency of mention before the alphabetical grouping. This would have the top mentioned items in as many separate groups as there are.

That better?

Not sure Maybe you should post a poll to help decide.

I vote for option 1. Polls with many options, like Sampiro’s, work just fine. The discussion that goes along with a poll is probably just as interesting as the poll itself. If you did several preliminary polls, the discussion would be split up in multiple threads, which doesn’t sound as fun as a single poll.

If it comes to that, I’ll mention you support the idea in the OP.

I’m not concerned about commentary within the preliminary polls. The threads mentioned in the OP are gathering the details that will be polled as soon as I can figure out the fairest way to do that. The idea in the subpolls (preliminary ones) will be to permit dozens of individual items to be polled and yet to wind up eventually with one Top Choice. Please refer to the Cafe Society threads mentioned in the OP.

Zeldar - I did look at the voting threads you mentioned in the OP before I made my comment above, and then I gave my opinion about the four options. I don’t think you can go wrong whatever you decide. I do understand why you want to do the preliminary polls.

Okay, thanks. I’ve reread your comments and this time I saw that your idea of proceeding with the trim-down process was your main recommendation. I was more struck the first time by what seemed to be an issue with discussion and splitting that up over several polls.

Thanks for restating your main point. It does make some sense to trim off the once-only items, since there are so many of them. But since the current threads have only been open a few days, and have gathered a lot of details in that short time, I think it’s closer to fair to include all of the entries in the prelims. Those coming to the preliminary polls as their first exposure to the idea of getting a Straight Dope Favorite TV Show of All Time, may have a stronger pull to the ones that weren’t as popular in the first thread.

The eventual voting in the prelims will need to be explained well so that those voting will realize that a Final Poll will be built from the results of the prelims. I’m still wrestling with the most effective (and clearest) way to present that!

Any ideas on that part? Thanks once more for your input.

Since the TV Shows thread has presented the biggest problem of the three, here’s my thinking on it.

How does this sound – beyond involved?

I’m thinking there are quite a few people who won’t read the thread before they post their choices, but that there will also be quite a few people who read the thread and don’t bother to relist choices that have already been mentioned…

I’m doing the western one so I’ll be interested to hear how you think I approached the subject (since I want to steer clear of long, endless lists)

I agree. There are all kinds of participants in threads of this type. It’s hard to predict how things will go and even harder to address issues that arise during the course of the thread.

Well, if your hope is to include all the ones that were mentiond in the thread, I’m really curious how you handle it. Any idea when your poll will go up?