Ideas for Another BR vs. HD DVD Poll?

[gorilla]

Stampeder asks

What is your own take on the poll results so far? As of 113 voters, the decided folks are choosing BR 4:3 over HD:

It's currently 140 votes, running about 5:4 but even if this was a proper scientific poll, that's still not enough to give a statistically valid discrimination. Just through chance, it's more than 50% likely that the actual winner is HD, and the poll is just sampling more BR.

[Arthur Dent]

Apart from the error margins, the poll options are not worded well. For example, what is the first option supposed to mean: "I will go/remain with Blu-Ray"? Are you going to go from HD DVD to Blu-ray as a result of the Blockbuster announcement? The way it is worded, in the context of this topic, any Blu-ray owner will automatically choose option 1. As for option 2 - same thing. The chances that a BD owner decides to go with HD DVD as a result of this announcement is zero, so option 2 is pointless too. Options 3 and 4 are just irrelevant.
All this poll shows is how many HD DVD vs BD owners have looked at the topic. Nothing on what the influence of the BB announcement is. The poll question should have been "If you are an exclusive HD DVD owner, will you go BD as result of this announcement? And only 2 options - Yes and No. But, of course, I'm pretty sure that a lot BD owners would feel the urge to give their not so wanted input, which will again skew the results.

[calvin940]

Originally Posted by Arthur Dent

But, of course, I'm pretty sure that a lot BD owners would feel the urge to give their not so wanted input, which will again skew the results.

I would hope that we are all adults here (or at least mature) and that such a poll could work.

Cal

[Arthur Dent]

OK, then. I suggest that we start a new poll (I don't know how), and the question should be: "If you DO NOT currently own a Blu-ray player, did the Blockbuster announcement not to carry HD DVDs in additional stores made you decide to purchase a Blu-ray player?" Yes or No.
I wouldn't be eligible to vote, because I am bi-.

[gorilla]

Apart from the error margins, the poll options are not worded well.

Yes, if I was designing the poll to find the influence of the BB decision, I'd ask it as two questions, what were your intentions before, and what they now after learning of this decision. That would give the same 4 questions on each, but a total of 16 possible pairs of responses. Obviously some of these pairs will have virtually no responses, but you have to ask them.

I would hope that we are all adults here (or at least mature) and that such a poll could work.

Unfortunatly, even if you have 100% mature adults, then people will still misread the intentions of the poll or simply make mistakes. That's one of the reasons that proper polls have rules and procedures to follow, ensuring that only applicable questions get asked. However, even if there was total compliance, it would still not be a scientifically valid poll, as it's not ensuring a statistically valid random sample. Pehaps more people who are already thinking of switching from HD to BR will read the thread, and therefore vote, and obviously the membership of DHC isn't a good sample of the general population.

[calvin940]

Originally Posted by gorilla

obviously the membership of DHC isn't a good sample of the general population.

I think we are just looking for a poll with more meaningfull results rather than being statistically significant. The poll in the current vein tells us little to nothing about the BB announcements and the HD DVD format supporters. A better worded poll would give us better information and more meaningful but may not represent the population - which is fine by me.. I am interested in the members of DHC at this point.

Cal

[gorilla]

Actually, I think we want a poll which is both. If we have a poll which isn't statistically significant, then our results aren't meaningful - there is a significant chance that it's actually a different result. On the other hand, if they're significant, but not meaningful, then we've got something which is pointless.

[calvin940]

We are not going to have population significance in this forum. That's why I am saying we shouldn't worry about it in here as it is unattainable. We are neither random nor a proper cross-section of the consumer group as a whole.

Cal

[stampeder]

As author of the poll, my plain and simple purpose is to survey DHC Members (only) on whether the Blockbuster announcement has had an impact on their plans. It doesn't concern their previous choices, just whether their plans have changed. It is a nuanced difference, and simply asking yes, no, or maybe is boring so the questions were fleshed out to reflect better on what is not necessarily a manichean choice.

[gorilla]

We are not going to have population significance in this forum.

No, but we can still have significance in the poll in itself. If you have 60 voting type "A", 20 people type "B", and 20 undecided, that's overwhelming statistically valid (About 1 chance in 1000 that type "B" actually won). If you have 6/2/2, then that's still statistically valid, but much less so (about 1:100). If you have 4/3/3 then that's not valid at all - 75 chances in 100 that B actually won.

[calvin940]

The statistical significance of which you speak is poll results oriented and not controllable by us other than choosing to create a meaningful and valid set of questions and only if a proper sample size answered the poll in the first place (relative to the entire membership of DHC).

This can be endless, so all I am saying is we create questions that make sense and have enough people answer that will tell us at least what the DHC is thinking. It won't however tell us anything about the rest of the public (nor will it tell us much if not many answer the poll to begin with).

What is 155 people as a percentage of the membership?

Cal

[gorilla]

This can be endless, so all I am saying is we create questions that make sense and have enough people answer that will tell us at least what the DHC is thinking. It won't however tell us anything about the rest of the public (nor will it tell us much if not many answer the poll to begin with).

Agreed.

What is 155 people as a percentage of the membership?

It doesn't really matter. Statistical relevence is largely independant of the size of the larger population, it's a feature of the size of the subsample. If you've got very small samples, they're not relevent. if you have very large samples, they're relevent. It's only in the border zone when the size of the larger population is important.

[calvin940]

Originally Posted by gorilla

It doesn't really matter. Statistical relevence is largely independant of the size of the larger population, it's a feature of the size of the subsample. If you've got very small samples, they're not relevent. if you have very large samples, they're relevent. It's only in the border zone when the size of the larger population is important.

heh.. I was just asking out of curiousity. I often wonder how many people really vote on those things to begin with. Of course, many members probably shy away from the format war section

Cal

[gorilla]

After doing a little looking around, I think the answer for how many people have ever joined is around 36,000.

If you click on someone's id, it links to /member.php?u=x. I'm pretty sure that this is a sequential ID for each member. People who have been members for a long time have low ID's, and decreasing amount of time means higher IDs.

If find someone who joined this month, then x is just under 36,000. Of course, this includes a number of people who have joined and are banned, or otherwise no longer active.

[gorilla]

For the sake of argument, if we were to divide the undecideds evenly, BR wins convincingly.

If I understand you, the vote would be 60 + 49/2 = 85 for Blu-Ray, and 43 + 49/2 = 68 for HD DVD.

This is JUST statistically significant, passing at the 90% confidence, but failing at 95% confidence. The latter is the standard usually used for surveys (the right 19 times out of 20), but the former is sometimes used when the study group is small.