How to Construct and Analyse a Poll 101

[Almightyjosh]

POLL TYPES
These are several poll types that I have observed during the demogame. A major problem with the polls is that people are unable to interpret the results to give a definate outcome. This is not an exhaustive list, merely a guide.

THE ABSTAIN OPTION
Firstly there are three types of majority that could be required:
1. Majority of population
2. Majority of those who voted per say
3. Majority of those who voted for or against

The first two DO NOT need abstain options, as abstaining is the same as voting NO. These two types of poll are required to pass certain laws and appointments.
The third type MUST contain an abstain option. The abatain option allows people to count their vote toward reaching quorum, without influencing the outcome. Most polls are of this type.

MUTUALLY EXCLUSIVE OPTIONS
These are very simple polls where there are two or more options, only one of which is possible, as it negates the others:
e.g.
Elections (only one candidate can win)
Changes in Law (always yes/no, can't sortof change the law!)

In this type of poll, the option that has the majority required, wins (could be 51% or 33% or whatever)

 

[Almightyjosh]

SPLIT OPTIONS
This is one of the GREATEST problems in our polls. People offer sevel options, some of which are very similar. This is usually because the method of change is split up. I'll explain with an example that is very common, the YES/NO/LATER poll.
e.g. Should we plant spies
1. Now
2. Never
3. Later
The problem is that that 2 & 3 have the same result, no spy! If the citizen votes are thus:
1. Now - 10
2. Never - 6
3. Later - 7
4. Abstain - 0
It seems on the surface that the NOW option has won, but this is not the case. 10 people advocated planting spies NOW, but 13 people (6+7) people advocated NOT planting spies now, even though they disagreed on whether to plant them in the future.
The same problem occurs whenever you try to combine a 'should we' with a 'how should we' poll.
The correct way to deal with this problem is to run seperate polls. The first is a hypothetical:
'Should we plant spies NOW?'
1. Yes, plant them
2. No, don't plant them
3. Abstain
This would give a clear result. If the result was NO! You could run another poll like:
'Should we plant spies in the future, given that we are not planting them now?'
1. Yes, later
2. No, never
3. Abstain
Thus we have a clear result overall.

Two things to consider...
A- In our first poll the NO vote was split 6/7, but this does not mean that the LATER option should win, as those who voted NO may have voted NEVER or LATER if the choice was simply between them.
B- If there are time constraints, the polls could be run concurrently, wiht the second run as a hypothetical.

 

[Almightyjosh]

INCREMENTAL OPTIONS (THIS IS NOT TO DO WITH MULTI-CHOICE)
This is the hardest to understand, as it seems to be counter-intuitive. In this type of poll, there are many options and each option does more (or less) than the one before. The difficulty is defining at what the majority of citizens approved!!
e.g. What do you think is the appropriate level of involvement in the demogame for moderators?
A: No involvement
B: Citizens only
C: Non-game effectual position (e.g. Mayor or census officer)
D: Depudy or Chat Rep
E: Leader of Department or Senator
F: Any position, including President

This is better understood as:
A: No involvement
B: Citizens
C: Citizens or Non-game
D: Citizens or Non-game or Depudy
E: Citizens or Non-game or Depudy or Leader
F: Any position

Each one is greater than the one before. Thus, anyone who votes for B supports mods as citizens, but those who vote for C+D+E+F also support mods as citizens, but with also rights as well. Thus, for mods NOT to be citizens, more people would have to vote for A than for B+C+D+E+F. This logic can be used to move incrementally up the scale until you get to a point that the 'less' outweighs the 'more'. ie if A+B > C+D+E+F but A+B+C < D+E+F then mods would be citizens and non-game officials, but not depudies. This is hard to understand, but I promise it is true!


This is just a start, but I hope I've made things more clear. If anyone is unsure how best to construct a poll, feel free to ask me (or anyone else for that matter). Just remember:
The best polls are YES/NO
NEVER run 'should we' and 'how should we' together
and Multi-choice polls should never be used for incremental polls.

Now, go forth and discuss!

 

[eyrei]

Hehe. Good luck! (seriously) We have been trying to do this since the beginning.

 

[Eklektikos]

Think I should draft him onto the the Polling Standards Commission?

 

[Almightyjosh]

Personally, I'd love to see an active polling standards commission. Polls are the main way in which decisions are made, and I think citizens need to be able to get advice on how to construct their poll so as to achieve the most useful result. After all, our democracy is all about citizen will, and polls are our way of assessing that!!

 

[Eklektikos]

So join up!

 

[Shaitan]

Fantastic! Clear, concise and to the point.

How about a write up on multi-choice polls? When to use them, how to rate them, etc.

 

[disorganizer]

Maybe one addition:
We should not count the abstain options as no in some polls.
On apo, i just read a good reason for abstain in any poll:
People not wanting to influence the game maybe want to see the result!
So they would vote abstain, as for example i do as compromise on apo...
We should think of that and maybe change our interpretations accordingly.

 

[donsig]

Originally posted by Almightyjosh
POLL TYPES
THE ABSTAIN OPTION
Firstly there are three types of majority that could be required:
1. Majority of population
2. Majority of those who voted per say
3. Majority of those who voted for or against

Well now, we can't very well know if a Majority of population has been reached when we can't even agree on what the population is. Or am I missing something?

We had 51 citizens vote in the last presidential election. Seems to me a majority would then be 26. But wait, we want to base the census on active citizens so we're going to change the census to the average number who voted in the elections. Any way you cut it this dis-enfranchises some citizens. And to think we spent so much effort building universal suffrage in Phoenatica.

 

[Shaitan]

Disenfranchisement is went you prevent someone from voting. That has nothing to do with determining a number of active citizens.

Is it just the term "census" that is bothering you so much? We can change it to "census of active citizens" without too much trouble, you know.

 

[donsig]

Originally posted by Shaitan
Disenfranchisement is went you prevent someone from voting. That has nothing to do with determining a number of active citizens.

Is it just the term "census" that is bothering you so much? We can change it to "census of active citizens" without too much trouble, you know.

No Shaitan, what is bothering me so much is that you are trying to get a lower population number. This lower base population number will result in fewer votes being needed to make new rules. This means we will continue to spend more time making rules than playing the darn game! Enough is enough already.

I have only been trying to defend the constitution here. We do not need to lower the bar for constitutional amendments. If you all decide you want to lower the bar for passing laws then go right ahead. You can do that by passing a law without changing the constitution at all!

Section D, Point 3 from the CoL:

(4) The quorum for changes in the Code of Laws is 1/2 of the census.

(5) A 2/3 majority of support is required to adopt or alter a law.

If you all want laws that are easier to change then this is what must be changed, not the constiution. There is nothing in the world (or the constitution) that says the passage of laws must be based on the census. Go ahead and base the passage of laws on the average number of voters in the elections but please leave the constitutional census alone!

 

[Cyc]

Quoting Donsig:
Well now, we can't very well know if a Majority of population has been reached when we can't even agree on what the population is. Or am I missing something?

Donsig, ol' buddy, for as many times as you've asked that question, I am going to have to answer yes this time. But I'll try to get Bill to give you some extra gold stars for persistence.

 

[Shaitan]

Let's not hijack this thread as well. Rehash on this topic can go in the appropriate discussion thread or poll thread.

 

[disorganizer]

ok. so come back to topic:
should abstain be counted or not?
as i stated above, abstain could also be a read-only non-citizen

 

[Shaitan]

If abstain is an option then it's votes must be counted, just as any other option. It is also possible that someone who is not a citizen will vote for a deciding option. Should we disount those also because of the possibility? No, if we throw out votes we are disenfranchising our citizens.

 

[disorganizer]

ok, then how about adding an option for non-citizens (sounds strange somehow though)?
i can just tell how i as embassador handle interesting polls in apolytonia.
or maybe we make standard that in any poll, we add the sentence "if you voted abstain as non-citizen, please state so below".
or something like that.

 

[Almightyjosh]

I hate doing this, but not much of whats been said has to do with the topic at hand! There is already a discussion on changing quorum levels elsewhere, as Shaitan said.
As to the absain rule, this was desogned by the polling commision so that citizens could vote and help reach quorum without influencing the outcome. What I wan trying to illustrate was WHEN using the abstain option was appropriate. For instance, in a normal poll, YES must out poll NO to win, regardless of whether there are abstainers. If the poll is one that requires, for example, a 2/3 majority of those who vote to pass, then abstaining is never an option, as it is as good as a NO vote.

 

[disorganizer]

yes, but it does not harm the outcome.

 

[donsig]

Let's see the title of the thread is How to Construct and Analyse a Poll 101.

How in heck do we analyse a poll based on our population when we do not agree on what our population is? How is this not relevant to this thread?

What am I missing Cyc. Please tell me! If 51 people vote in an election does that not mean we have at least 51 citizens?

And BTW, the discussion on changing quorum levels has resulted in a proposed constitutional amendment to change our constitutional census rather than directly change any quorum levels!

There is little point in rehashing ground that the Polling Standards Commision has already covered if we cannot first resolve the underlying problems!