164 réponses à ce sujet

[marshalleck]

I'm not sure I like this idea of trying to cherry-pick opinions that are "valid"

Too bad Google doesn't track the number of people responding more than once, that seems like it should be obvious

[TOBY FLENDERSON]

sweet thread, lets try to keep this going.

[DigitalMaster37]

Gerse1 wrote...

Deltaboy37-1 wrote...

fortunesque wrote...

Yeah. We should test the cameo effect.
Though I think we all know how that'll end up

     Funny thing is that the ME1 Lis arent doing bad at all. It's just a bunch of people fudging results and it's clear who is doing it. I have a calculating way to eliminate those who put 10's and 1's on specific people to boost and lower certain people based on past LI wars on these boards. Goog thing being good at math and analytics helps with eliminating those skewed votes.
     The interesting thing is that I didn't do that with the last results and th ME LIs still did very well in the polls. I will do that in the next though. So for all those that will look at my results, you will see slightly different numbers based on different algorithms used in the calculations. This will illiminate fanboy/girlish votes. The funny thing is if you look at the votes closely you can spot these a mile away.

How is that fair though? One could call me a "fangirl" of Garrus but I gave Kaiden a 2 because I honestly think he is a 2.  I thought it in ME1.  So what if that's because he pales in comparison to someone else?  That fact is revealing in itself of a character's strenght or weakness.

right, I don't mean to say that every one that looks like Liara 10, Tali 2 or Ashley 5 Miranda 9  will be eliminated, you have to be more efficient than that I know, but again, there are techniques one can use to eliminate ones like Tali 10, Ashley 1. When you get a large number of those, you are either running into people who haven't played ME1 and/or people who are fanboys of one, not caring to be honest about the others. If you want to be truthful, no one deserves a ten and neither does anyone deserve a one. That's just being realistic. People just want to see their favorite character get the most votes and to me that is 'fudged". IF you are rating a game, you don;t give it a ten, unless your jusdgement is slightly impaired, by (I better not say it) fanboyism. Honestly, any poll done by fanboys isn't a poll at all. Lets face it.

[marshalleck]

Deltaboy37-1 wrote...

right, I don't mean to say that every one that looks like Liara 10, Tali 2 or Ashley 5 Miranda 9  will be eliminated, you have to be more efficient than that I know, but again, there are techniques one can use to eliminate ones like Tali 10, Ashley 1. When you get a large number of those, you are either running into people who haven't played ME1 and/or people who are fanboys of one, not caring to be honest about the others. If you want to be truthful, no one deserves a ten and neither does anyone deserve a one. That's just being realistic. People just want to see their favorite character get the most votes and to me that is 'fudged". IF you are rating a game, you don;t give it a ten, unless your jusdgement is slightly impaired, by (I better not say it) fanboyism. Honestly, any poll done by fanboys isn't a poll at all. Lets face it.

Or that's their genuine opinion. How does your algorithm account for that?

Modifié par marshalleck, 12 février 2010 - 02:25 .

[ThatDancingTurian]

XX55XX wrote...

Yes, it does, unfortunately. This method
is the easiest compared to the old way of inputting dozens of values
into SPSS, but allowing any user to take the survey as many times as
they want really skews the results to a degree. That's probably why the
standard deviations are so much higher this time around.

But how high is that degree? All it could take is one incredibly obsessive fan.

Deltaboy37-1 wrote...

right, I don't mean to say that every one that looks like Liara 10, Tali 2 or Ashley 5 Miranda 9  will be eliminated, you have to be more efficient than that I know, but again, there are techniques one can use to eliminate ones like Tali 10, Ashley 1. When you get a large number of those, you are either running into people who haven't played ME1 and/or people who are fanboys of one, not caring to be honest about the others. If you want to be truthful, no one deserves a ten and neither does anyone deserve a one. That's just being realistic. People just want to see their favorite character get the most votes and to me that is 'fudged". IF you are rating a game, you don;t give it a ten, unless your jusdgement is slightly impaired, by (I better not say it) fanboyism. Honestly, any poll done by fanboys isn't a poll at all. Lets face it.

It shoudn't matter what someone picks. It's -based- on bias, the people decide who gets what, whether that be a ten or a one. What you consider 'fudged' results is kind of shifty and prone to your own personal biases.

While I applaud your efforts to increase the survey size, I can't help but see this system as incredibly flawed.

[sliverofamoon]

KBGeller wrote...

Alright, I did the survey and it was well done. The only thing that may be interesting to know/ask for (if the survey can be modified after creation) is under the LI topic, you do not prompt for multiple LI's. I notice it says primary, so primary could be a personal choice as to who was last or who was most important. But you can "go all the way" with each possible LI in endgame I believe. However, if you break up with an LI in endgame, you cant go back to them. And the savegame does record which/how many LIs you have had.

The only reason I chose my primary on Mass Effect 1,  was because the choice I WANTED wasn't available, so really, statistically, it's a false positive. My choice in ME 1 would have been Tali.

Modifié par sliverofamoon, 12 février 2010 - 02:28 .

[XX55XX]

Aris Ravenstar wrote...

XX55XX wrote...

Yes, it does, unfortunately. This method
is the easiest compared to the old way of inputting dozens of values
into SPSS, but allowing any user to take the survey as many times as
they want really skews the results to a degree. That's probably why the
standard deviations are so much higher this time around.

But how high is that degree? All it could take is one incredibly obsessive fan.

Deltaboy37-1 wrote...

right, I don't mean to say that every one that looks like Liara 10, Tali 2 or Ashley 5 Miranda 9  will be eliminated, you have to be more efficient than that I know, but again, there are techniques one can use to eliminate ones like Tali 10, Ashley 1. When you get a large number of those, you are either running into people who haven't played ME1 and/or people who are fanboys of one, not caring to be honest about the others. If you want to be truthful, no one deserves a ten and neither does anyone deserve a one. That's just being realistic. People just want to see their favorite character get the most votes and to me that is 'fudged". IF you are rating a game, you don;t give it a ten, unless your jusdgement is slightly impaired, by (I better not say it) fanboyism. Honestly, any poll done by fanboys isn't a poll at all. Lets face it.

It shoudn't matter what someone picks. It's -based- on bias, the people decide who gets what, whether that be a ten or a one. What you consider 'fudged' results is kind of shifty and prone to your own personal biases.

While I applaud your efforts to increase the survey size, I can't help but see this system as incredibly flawed.

It hardly affects the character rankings, from what I can see. Mordin is still on top, with Legion, Tali, and Garrus behind. Additionally, while the standard deviations are higher, everything appears to be within scale. (For example, Mordin's standard deviation is still the lowest of the bunch.)

[sliverofamoon]

NuHPgn wrote...

the first question in the survey should be whether you play as male or female Shepard

And if what you play is the same, or opposite gender.

[XX55XX]

sliverofamoon wrote...

NuHPgn wrote...

the first question in the survey should be whether you play as male or female Shepard

And if what you play is the same, or opposite gender.

I have integrated that into the survey. However, people may only make one choice, no matter which game they have played.

[ThatDancingTurian]

Okay, granted, this is from someone who does not understand statistics at all. But are you saying one fan taking it 50 or so times out of 500 wouldn't affect some sort of change?



It takes about two seconds to take, and I've heard of incredibly bored people doing much stupider things to occupy their time.

[Razgriz9327]

is this both games or just Mass 1 or 2

[DigitalMaster37]

marshalleck wrote...

Deltaboy37-1 wrote...

right, I don't mean to say that every one that looks like Liara 10, Tali 2 or Ashley 5 Miranda 9  will be eliminated, you have to be more efficient than that I know, but again, there are techniques one can use to eliminate ones like Tali 10, Ashley 1. When you get a large number of those, you are either running into people who haven't played ME1 and/or people who are fanboys of one, not caring to be honest about the others. If you want to be truthful, no one deserves a ten and neither does anyone deserve a one. That's just being realistic. People just want to see their favorite character get the most votes and to me that is 'fudged". IF you are rating a game, you don;t give it a ten, unless your jusdgement is slightly impaired, by (I better not say it) fanboyism. Honestly, any poll done by fanboys isn't a poll at all. Lets face it.

Or that's their genuine opinion. How does your algorithm account for that?

That just so happens to be the casualty of having to adopt such an extreme method to curtail the fanboys (who effectively mess up a good survey) If you disagree then unfortunately I'd bet many more would agree with me. If your looking for accuracy, never ask a fanboy anything about their opinion. Believe it or not I give a liara a 9 and Tali a 7. That's realistic as much as some would like to slight the Talimob, it is not realistic, same for Liara, Ashley and Kaiden.

[Gerse1]

Well first of all rating a game a 10 and rating a character a 10 are apples and oranges, so just because rating a game a 10 is "incorrect" in the sense that "perfection" is a virtual impossibility, doesn't mean you can't rate a character a 10 and mean it. It is a personal preference. Even if there are 10,000 "fanboys" who think Tali is a 10, well that simply reflects the positive reception of her character. If all those fanboys rate Ashley a 1, they are 100% entitled to have that opinion of theirs reflected. It's not like they rate ALL non-Tali characters a 1. And the reason is because some of the other characters are also likeable to them, if not as *much* as Tali. I just don't see how the degree of one's love for a character is cheaper if it's the highest score, and more "reasonable" if it's lower. Opinions aren't supposed to reasonable. This isn't a critical review of the characters from professional writers and game designers, it's a fan reaction sample.

[Gerse1]

"If you are looking for accuracy, never ask a fanboy anything about their opinion." No, if you are looking for a good prediction of how *you* will feel about a character or game, you should be wary of fanboy/girl opinions and just see for yourself. Still has nothing to do with polling who people's favorite characters are.

[.primus]

I took part in this simple survey.



The LI part was pretty simple. Though I have gone through all of them or planning too. The ones I choose are the ones I preferred.



With the characters. For the characters from ME1 I rated based on how I like or disliked their change rather than my opinion of them from the first game or their usefulness throughout my missions. With the new recruits I rated based on a lot of factors from usefulness, how i their interactions with my shepard went (paragon or renegade), and if I would like for them to stay for part 3.

[XX55XX]

Aris Ravenstar wrote...

Okay, granted, this is from someone who does not understand statistics at all. But are you saying one fan taking it 50 or so times out of 500 wouldn't affect some sort of change?

It takes about two seconds to take, and I've heard of incredibly bored people doing much stupider things to occupy their time.

If someone does do that, then the results will be skewed. Your note of this flaw is quite valid. But, as far as I can say, if, for example, the Ashley fanboys and the Liara fanboys are really sparing it out over this silly little survey, then whatever they do will cancel out one another, hopefully.

Unfortunately, as much as I want to, Google doesn't offer such an option of limiting one vote per Google account or something along those lines...

We are really relying on people's good faith to take this survey only once. But fortunately, version 3.0 of the survey is going to be much longer and spans three pages. I really hope no one does this survey fifty times just so that their character will win. I know there are people taking the survey multiple times, but taking an extremely large sample size can alleviate the skewing somewhat.

[DigitalMaster37]

Aris Ravenstar wrote...

XX55XX wrote...

Yes, it does, unfortunately. This method
is the easiest compared to the old way of inputting dozens of values
into SPSS, but allowing any user to take the survey as many times as
they want really skews the results to a degree. That's probably why the
standard deviations are so much higher this time around.

But how high is that degree? All it could take is one incredibly obsessive fan.

Deltaboy37-1 wrote...

right, I don't mean to say that every one that looks like Liara 10, Tali 2 or Ashley 5 Miranda 9  will be eliminated, you have to be more efficient than that I know, but again, there are techniques one can use to eliminate ones like Tali 10, Ashley 1. When you get a large number of those, you are either running into people who haven't played ME1 and/or people who are fanboys of one, not caring to be honest about the others. If you want to be truthful, no one deserves a ten and neither does anyone deserve a one. That's just being realistic. People just want to see their favorite character get the most votes and to me that is 'fudged". IF you are rating a game, you don;t give it a ten, unless your jusdgement is slightly impaired, by (I better not say it) fanboyism. Honestly, any poll done by fanboys isn't a poll at all. Lets face it.

It shoudn't matter what someone picks. It's -based- on bias, the people decide who gets what, whether that be a ten or a one. What you consider 'fudged' results is kind of shifty and prone to your own personal biases.

While I applaud your efforts to increase the survey size, I can't help but see this system as incredibly flawed.

I haven't even explained the system and it's getting shot down. And why are we so defendent of fanboys??? I 'd like to think you'd be subjective, but Zaeed deserves a 2-3 at best, being that a 1 signifies the character was utterly a failure. I can't see any objective person rating any character a 1, I don't care how much you "don't" like them. 

At the end of the day, it's all about how the survey creator describes the meaning of the number system.

It just seems like your defending fanboys, which raises my eyebrow.

Accurate polls and surveys require objectivity not subjectivity.

[jason9880]

Deltaboy37-1 wrote...
...snip...
Honestly, any poll done by fanboys isn't a poll at all. Lets face it.

This.  I find all the work you guys have done to be interesting, but with uber fanboys and double voting it's far from perfect.

[XX55XX]

Deltaboys, surveys like these are always going to be subjective. In fact, the original survey was hardly objective at all - we were asking people how they were feeling about certain characters on a 1-10 scale. There is nothing objective about that.

In my analysis, it seems that the act of people taking the survey multiple times is having limited affect on the means. Overall, the means are slightly down from the first survey, but they are all within the ballpark.

[DigitalMaster37]

Gerse1 wrote...

Well first of all rating a game a 10 and rating a character a 10 are apples and oranges, so just because rating a game a 10 is "incorrect" in the sense that "perfection" is a virtual impossibility, doesn't mean you can't rate a character a 10 and mean it. It is a personal preference. Even if there are 10,000 "fanboys" who think Tali is a 10, well that simply reflects the positive reception of her character. If all those fanboys rate Ashley a 1, they are 100% entitled to have that opinion of theirs reflected. It's not like they rate ALL non-Tali characters a 1. And the reason is because some of the other characters are also likeable to them, if not as *much* as Tali. I just don't see how the degree of one's love for a character is cheaper if it's the highest score, and more "reasonable" if it's lower. Opinions aren't supposed to reasonable. This isn't a critical review of the characters from professional writers and game designers, it's a fan reaction sample.

I don't know if you have been on these forums lately, but it is clear there is a LI war going on regardless of who is openly waging it or not. In particular it is being played in the threads and in particularly, these polls. So if you want to go by blind fanboyism then look on the boards and you'll see those results. If you want to get an objective look at things, then you need to exort some limited amount of control (across the board) over the results.

You guys act like I would just go in and randomly pick out votes, no that is not how it works. There are other methods to look at. It's a lot to explain so I won't bore you, but I do related type things at work. So aside from it being fun, it is also more helpful to see a more objective view in the results.

[XX55XX]

I have shut the poll down. 1,026 responses. Amazing.

Thank you everyone!

[DigitalMaster37]

Gerse1 wrote...

"If you are looking for accuracy, never ask a fanboy anything about their opinion." No, if you are looking for a good prediction of how *you* will feel about a character or game, you should be wary of fanboy/girl opinions and just see for yourself. Still has nothing to do with polling who people's favorite characters are.

I peacefully disagree, you don't ask a fanboy nothing. At best you will get lucky and agree with them, most commonly you will look at yourself and see a donkey.

It is just a reality and even in the corporate world they know they must do certain things to adjust for the margin of error, this is all I am adding, nothing more, nothing less.

[marshalleck]

Deltaboy37-1 wrote...

I haven't even explained the system and it's getting shot down. And why are we so defendent of fanboys??? I 'd like to think you'd be subjective, but Zaeed deserves a 2-3 at best, being that a 1 signifies the character was utterly a failure. I can't see any objective person rating any character a 1, I don't care how much you "don't" like them. 

At the end of the day, it's all about how the survey creator describes the meaning of the number system.

It just seems like your defending fanboys, which raises my eyebrow.

Accurate polls and surveys require objectivity not subjectivity.

The problem is that you're deciding whose opinions are valid and whose aren't, and your criteria are highly subject to your own personal bias. You've said several times that no character deserves a 10 or a 1. That's your opinion. If you're running a survey of opinion, you can't just discard opinions you don't like.

I'm interested in the subject of how characters are received by the community and how their development affects their popularity, but people being able to vote multiple times and the suggestion of arbitrarily defining some responses as "fanboy" and discarding them just leaves the results very suspect.

If someone wants to vote a character a 10 or a 1 or anything in between, that's up to them. It's a survey of opinion.

Modifié par marshalleck, 12 février 2010 - 02:51 .

[ThatDancingTurian]

XX55XX wrote...

If someone does do that, then the results will be skewed. Your note of this flaw is quite valid. But, as far as I can say, if, for example, the Ashley fanboys and the Liara fanboys are really sparing it out over this silly little survey, then whatever they do will cancel out one another, hopefully.

Unfortunately, as much as I want to, Google doesn't offer such an option of limiting one vote per Google account or something along those lines...

We are really relying on people's good faith to take this survey only once. But fortunately, version 3.0 of the survey is going to be much longer and spans three pages. I really hope no one does this survey fifty times just so that their character will win. I know there are people taking the survey multiple times, but taking an extremely large sample size can alleviate the skewing somewhat.

Considering Ashley/Kaidan/Liara fans are generally agree, I doubt they'd cancel each other out, just like Garrus and Tali fans wouldn't cancel each other out since they tend to agree... But I get your point. Unfortunately, relying on good faith doesn't work with groups like these where apparently liking one option means you have to hate others. Hopefully a large number of legitimate, non-duplicated votes will be enough to solve this issue.

@Deltaboy37-1: I'm not defending 'fanboys', I'm defending everyone's right to vote as they see fit and have that vote counted. It doesn't matter if they're 'fanboys'. They're voting on a poll about personal opinion. Like it or not, their opinion is no less valid than yours.

[ThatDancingTurian]

XX55XX wrote...

I have shut the poll down. 1,026 responses. Amazing.

Thank you everyone!

Nice work, that's an impressive sample size.