This is something that got me thinking recently. What type of decision process would work well if you had to make choices about a question? Say you have to vote for a leader of your country, or if you had to decide over what can be do with the limited time you have in a day.

Would choosing one answer work best for you? Would choosing multiple answers work better? Would placing priorities over certain tasks work better?



There are 16 questions in the excel sheet provided. They are random questions & are not meant to relate to each other. This survey process will last until either 10 participants have submitted their answers, or 2 months have past, whichever occurs later. After that time, the results for the questions will be released for all to see.

The intent of this experiment is to see what results will occur and to see if any of the methods mentioned above work better than the other.

Behind the spoiler button below is the list of questions being asked. DO NOT CLICK THE BUTTON BELOW IF YOU WISH TO SEE THE QUESTIONS FOR THE FIRST TIME IN THE EXCEL SHEET PROVIDED.

After you complete the survey, please send it to me via PM, e-mail, or a password protected zip or rar file for me to download and compile into a consolidated voting spreadsheet.

What do you think would work for a voting system if you had to implement one?

Wouldn't like method 3 - too much math... As for the other two methods, I can't quite see them working for most of these questions.

Very few of the questions work much for multiple choice (but some do, like favorite weekday). Some are definitive yes/no, others can have a too wide array of answers (like "which pet do you prefer" or "favorite dessert") and would either need an "answer in text" or "pick between these options" variant, and some again might vary from place to place, and from person to person (school subjects, desserts). For a lot of these questions, people would often prefer to answer in text, rather than multiple choice or some other method. To me it looks very much like something that could just as well be a regular off-topic topic. If you want to look at methods of choice for surveys, maybe choose other questions that don't have very wide options?

Questions like "what's your favorite[insert item here]" will only rarely give you a good view on what people actually like, as there are a lot of different favourite flavors, pets, colors, and so forth. YOu can re-phrase it into "What's your favorite [item] out of [multiple choice tick boxes]" and pick some colors or common desserts or whatever. But if someone's favorite color is turqoise, you'll not know the actual answer if the multiple choice only gives you green and blue as the possible options. You will however know if they can answer in text.

Besides, I think people might prefer to just give their answers in a regular post, rather than PM-ing, E-mailing or other ways. I honestly don't see the point of an excel spreadsheet. For different questions - maybe it can work. But for these? Maybe not so much.

I would answer in a post, excel sheet, no.

I just think that people's level of education might be important here. Math students might find the % answering fun, most others would be annoyed I should think.

If you have like a rating from 1 to 10 instead, might be more useful and not so mathmatical.

People who don't have a lot of education might not be interested in a multiple choice questioner at all.

If you ask people about the job they do and what tasks they prefer or think is most important, then a 1-10 rating might be considered useful.

It shouldn't be needed to send it to you? What about using onedrive or something and lett people answer the questions there, or there should be sites that lett you do this. It should even be less work to extract the results.

Home stationary edit: For method 1-10 it's usual to add how much a point count, is 10 the most, meaning that's the task you prefer to do first. If you ask people about what president they would vote for and there is 10 candidates, a only one choice would tell you who's winning if the election was now, while a multiple choice questioner would tell you the difference between these candidates in favor, and if it could easily flip.



I very rarly see text where I come from. I guess the problem if someone can answer in text is the variety of names on a color that others might think is turqoise. It's also some people that don't really know what that color is, some people also think they know what turqoise is, but it turns out to be green. A social anthropology student who deals a lot with color, would say that the only right answer to that would be if you wrote out colored paper of some kind and collected the answer in person.

People would be more likely to agree on a selection of cats and dogs. That's a biological distinction. For an artist though, agreeing on the name for a color might not be that important, sometimes it's far more useful to keep the discussion going.

@simmer22

I don't mind if folks give they answers in a post, although they may find it to be a hassle to keep their responses together with the questions.

Such as the question:
Which flavour do you like to taste in a dessert?

Method 1 gives you only one choice of a flavour...or if a respondent chooses more, their 1 vote gets divided evenly to represent 1 vote (e.g. I place an X on 3 flavours, my vote is split into 1/3 for each flavour)

Method 2 gives you a chance to rate each flavour out of 10, such as 10 being the most desirable flavour and 0 being the least desirable. You can rate each individually, and the vote will be pro-rated to represent a combined total of 10 points used (e.g. I vote 9/10 for chocolate, 4/10 for vanilla, and 3/10 on mint. That's 16/10, therefore the vote will be adjusted downwards by 62.5% to become 5.625/10, 2.5/10, and 1.875/10 respectively)

Method 3 gives you a chance to rate each flavour out of 100% in case you'd like to distribute your vote towards certain items, rather than choosing just one item. (e.g. I prefer my dessert to have 60% chocolate, 35% vanilla, and 5% mint, none of the other options work).

If a respondent chooses to post their answers using Methods 2 or 3, it may seem inconvenient.
---------------------------------
I find also that some of the questions that seem like a clear Yes/No decision may not really be. Such as:
Do you feel you sing well?

You may choose method 2 and say
Yes: 7/10
No: 3/10

Or method 3 and say
Yes: 78%
No: 22%

This may mean that the respondent feels they sing well most of the time, and at other times not. Whereas method 1 confines the respondent to give an absolute yes or no decision (e.g. Yes, they sing well every time, or no, they never sing well at any time).
---------------------------------
Oh actually, maybe even post the excel sheet in a rar file in their post and I can download it.
---------------------------------
@Volvenom

I've never used OneDrive from Microsoft. How does that work?

Having a text answer on the "Do you think you sing well?" question would actually give you a far better answer than percentages or by numbers. Another way is to do it by scale, so something like "On a scale from one to ten, how well do you think you sing?" or have a text scale along the lines of "bad, not so good, medium, good, excellent." If you've ever taken these kinds of questionaires (I've done several), having 3, 5 or 7 options on the scale tend to be enough. You need a "no", a "yes" and a middle "I have no clue either way" question, so 3 is minimum. You can of course do it in a 1-10 way, but not 1-100, because that's a bit much.

As for desserts, it can be easier with a text answer because there are so many options to choose from. One option would have been a multiple choice between regular favorites, say "what's your favorite dessert out of A) chocolate cake, B) ice cream, C) apple pie or D) I don't like dessert." owever, my favorite is always situation based, and depends a LOT on the dessert options. For instance, I love a well made chocolate cake more than most other foods, but I'm very picky, so faced with a mediocre one, I'd rather go for ice cream, unless it's pistachio or chocolate (which is weird, because I love chocolate, so technically that should be my favorite flavor ice cream, right? - but I have found ones I like, and occasion I've even liked some types of pistachio ice cream). There are apple pies I've loved even more than the two others put together. So a rating-based system would not have worked for me, unless I was actually asked to choose between foods I could taste. Whether I like chocolate cake as a 10/10 or as 2/10 depends on the cake. Whether I like one ice cream better than anouther can be not only down to type of ice cream but down to who made it. I can love one strawberry ice cream, and only barely tolerate another. So for some food-related questions, the 1-10 rating kinda works best if you can actually taste them.

As for definite yes/no/maybe questions, there really isn't much use in rating them. It's either yes or no, and if you're less than 100% sure either way, it's a "I don't know/maybe". You don't need to rate those questions, because it does not make sense.

I'm thinking several of the questions you've used could have been formulated to work better with what your intentions are with this survey. I have seen way too many weirdly formulated questions in surveys, and it's often the formulation of the question itself that gives the survey a result that can't be used for anything much. The worst kinds of questionaires are those where questions are formulated in such a way that whatever you answer, the answer would look bad in some way or another. These ones often crop up in official (and tested) surveys, for whatever reason.

As an example, here are some ice cream related questions that have different formulations:
Do you like ice cream? (yes/no)
What kind of ice cream do you generally prefer out of these options? (multiple choice)
What's your favorite ice cream? (text answer, maybe multiple choice)
Rate these 5 ice creams from best to worst (perhaps ratings 1-5)
A taste test between 20 different ice creams (0-10/10 or 1-6 on a dice, or similar)

I'd refrain from "What is your favorite ice cream flavor" and giving a small range of multiple choice options, because if the person's favorite is cookie dough and it's not on the list, the person doing the survey doesn't get a good enough answer anyway. But if you're faced with a slightly different formulation, "Which of these ice creams do you prefer?" and 4-10 very common options, you can in general figure out what kind of ice cream a population would prefer, even if you haven't taken into account all the thousands of flavors out there. It's a small distinction, but can be important for some types of questions.

In general I'd refrain from giving people too much "freedom" in a survey, because it complicates the answer and will mess up the result. Percentages also complicates things. Keep options simple. I don't neccesarily like strawberry ice cream as 75% and mint as 25%. Maybe I like both on an equal rating, but if I'm also taking into account chocolate ice cream and cookie dough, things start getting messy very fast. Also, it's easier to pick from a pool of 2-10 answers, than to have to look over 30 options. The more options you're faced with, the harder the math will be, and the harder the choice. In general, people usually prefer to either tick boxes or write a comment (short or long, simple or complicated, depending on the person), or give a 1-10 rating, or similarly simple answer methods. The simpler the answer method, the more likely you are to get good enough answers to your questions.

I do understand that you're trying to find which rating method works best for which question - but it depends a lot on the formulation of the question, the situation, and the pool of answers. If it's a yes/no/maybe question, it's a yes/no/maybe question. If it's a "pick between these" question, it's multiple choice. If it's "rate these items" it's 1-10, % or a throw of the dice, or 0-10 out of 10. Not all options go with all questions, and if you pick a bad answering method, or formulate questions in weird ways, you will not get good answers. You can of course mix up the type of questions, but you still need to formulate the answering method so it works.

If you use excel onedrive is on net all the time and you can save on onedrive instead. It should be pissible to set access to the file and who can access what. If you know how to use the program access, you can make questioners there as well. It maybe easier though to find an online site setup for the purpose though.

Yes, having text answers would be far better because the respondent can elaborate on their reason for choosing an answer, and you'd get more insight. But that is also far more time consuming to compile and categorize (especially if the respondent is leaning towards one answer, but also likes other answers) than it would be to bin answers in a fixed choice. If the survey wants to have an elaboration on their decision, follow-up questions would be needed.

On the other hand, because the answer is binned in fixed choices, the options provided mustn't be absolute to the extend that the respondent has second thoughts about their answer (such as "I voted Yes because I sing well most of the time, but sometimes I feel that I don't"). So the scale proposed helps provide some measure of their decision. If they want to get more specific, then a percentage rating could be their choice.

Now that's up to the respondent on how they would like to answer the question. How do they want to answer the question, is what we'll see.


Definitely, I'd rather taste the food first before rating them. The dessert question was intentionally vague so the respondent can conjure up their own image of a dessert they like and choose the response(s) to that question. It could be an ice cream, a cake, a pastry. This question doesn't have any follow-ups to clarify what type of dessert the respondents were thinking of when choosing a flavour.


And so if it doesn't make sense to rate them other than Yes/No, the respondents should all take that Yes/No/Maybe route, right? What happens if they don't? Wouldn't that be interesting...

The issue I had was trying to come up with a working for each question to make it possible for a respondent to use either methods provided. Now there is a flaw in the survey provided, as they do not have an "Other" option, since the answers provided are limited.

Like with the singing question:
"Do you feel you sing well?" (Yes/No)
"How do you rate how well you sing?" (1 to 10, 10 being the best)
"What percentage of the time do you feel you sing well compared to other times?" (out of 100%, 100% being all the time)

How do you write the question such that it gives the respondent the choice to use any one of the 3 methods provided?


- What if I like ice cream some of the time and not all of the time? If I say yes, it could imply I like it all the time.
- That's a good one because the options may not have the favour of the respondents, but can bring them to their next favour choices, if any.
- I like Rocky-Road-Chunky-Swirl-Apple-Pie-Deluxe flavoured ice cream. So does that taste like chocolate, I wouldn't know, lol.
- That's a good one too, and then let the surveyor rank the respondent's ice cream choices from first to last choice.
- I'd probably just go back and "re-taste" the ice creams because they're just soooooo hard to pick from, lol.

Yup, and so I tried using "prefer" as the wording for the questions, although I didn't think of searching for other synonyms of that word.

This is another part of the experiment to see what method the respondent chooses. The percentages would be more difficult with folks who may just want to give a straight response without having to elaborate or split up their preferences. So if the respondent doesn't like as much "freedom" to express their choices, they'll likely go for the one-vote method. I'd like to see how that goes.

The grander scheme of the experiment includes how a respondent chooses to answer a survey question. As well as do any method produce results that are more representative of a population given a handful of people who respond to the questions (e.g. how representative would any of the results be for the total membership of ModTheSims' forum).

E.g. Say there's 20 respondents for the "Do you feel you sing well?" question:

Method 1
12 say yes, 8 say no. That's 60% to 40%.

Method 2
Average scaled for yes is 7.2/10, and no is 2.8/10. That's 72% to 28%.

Method 3
Average yes as a percentage given is 69.7%, and no is 30.3%.

Do any of these methods seem more representative given the option of answering methods? To me it looks like random numbers (and it is!). Or maybe this randomness will slide to about the same percentages once this survey is over, who knows!

I just had a survey from surveymonkey. I filled in and it was easy peasy. Just a suggestion.

Forgive me if this has already been brought up, I confess I only scanned the contents of the replies rather than reading them thoroughly (my brain is way too fried to take in that kind of debate right now), but I noticed that the methods of calculating votes that was suggested doesn't include a system that I really like, so I wanted to share it.

It's called the Single Transferable Vote, and it's the system we use for everything (except a referendum) in Ireland. You rank your options in order of preference, with 1 being the most preferred. You can rank as many or as few as you like. The idea is that if your 1st choice can't win due to too few votes, they then go back and check who your 2nd choice was, and transfer your vote to that candidate. If that candidate is unsuccessful, your vote transfers to your 3rd choice, and so on until either your vote is used or you run out of next choices. It's a very complicated system to administer on the counting side, involving mathematical formulas and a lot of counting and recounting (it typically takes us a few days to finish counting a general election, despite having teams of people spending all day counting in centres all over the country), but it's simple enough on the voter's end of things, and allows people to vote for the candidate they most agree with, without having to risk "throwing away" their vote if that candidate is unpopular and unlikely to get elected.

I believe the cities of San Francisco and Oakland use that system, or something very much like it. That's Northern California, so, they surely have a mix of traditional party politics and numerous small activist communities. The ranked-choice voting system, from what I've heard, does a good job of letting the electorate express their real actual preferences (without game tactics), yet it usually elects candidates that would be able to win a traditional first-past-the-post contest against any one of the others. For that reason, they call it "instant runoff" voting. The effort on the counting side may be more elaborate, but, the results haven't been so weird as to cause people to question its transparency or integrity.

What I like about ranked-choice voting is that everybody, including the eventual officeholder, can see what the voters were really supporting -- because it's right there on the ballots & requires no guesswork or spin-doctoring.

Yeah, there's a bunch of different but very similar systems based on the idea. Our presidential elections are essentially instant runoff voting, because there's only one winner. Our general elections are way more complicated though, because of our multiple seat constituencies (which is another thing I'm very in favour of, because it allows two people in the same district with very different political views to both feel represented). In our last election I had nine candidates to rank, three of which were elected. Some areas have as many as five representatives. It's based on population, I think. For multi-winner contests there's way more maths involved. Formulas and quotas and things. If anyone wants examples of how it works, The Irish Times has some really comprehensive coverage from this year's election. For every district they have a chart showing the votes each candidate had at each round of counting, and lower down on the page they have an animation which shows how the candidates overtake each other as each round of counting and new vote transfers come in.

The particular district I linked to as an example tends to be one of the longer counts. They have five seats, and a lot of candidates. This year it took them 15 rounds of counting to fill all five seats.

It always amazes me how American elections can have a definite winner before people have even finished voting. We don't even start counting until the next day, and it takes us at least 2 days to finish. Sometimes longer, if a constituency ends up having a bunch of recounts (that happened a lot this year. I think it was wednesday or thursday before we had officially filled every seat. Voting happened the previous friday. And then it took literally two months for them all to come to an agreement on who was going to run the government. It was a mess )