New Entries in the CoasterBuzz Top 100

Sign up a bunch of new profiles, create some track records (including high ratings for whatever new coasters you think are missing) and you should be there.

I know it sounds ironic given my last two posts, but I don't honestly care enough.

My retort is most based on the "statistical significance" defense (which has been a long time sticking point with Jeff on the subject of coaster pools) and is purely an emotional reaction to years of that discussion more than anything.

Simply put, a poll isn't suddenly 'better' because some attempt to create statistical significance was made.

I think that one can argue that there's a difference between "best" and "favorite." I view this poll as a representation of the rides the membership here feels are their favorites, and in that sense, I think that the absence of such rides as Expedition GeForce, Silver Star, Oziris, Space Mountain: Mission 2, and several others is irrelevant. It comes down to what you're looking to get out of the data.

Also, it's SUCH an easy process to just give a ride 1-5 stars. ;)

Sucks.

I think the sample is too narrow...it doesn't account for car comfiness, air time, speed, knee-knocking lift hill, historic significance, track design, inverted, suspended, traditional, i-track/hybrid, how much is a funnel cake? shade, parking, wait times? Passholder benefits, was your head banged at all? did you lose your stomach? light display at night? operators dispatching one per five minutes?

Sarcastic, yes...but a series of questions broadening the scope would answer the final math score more definitively. I just read the list , and i305 is ranked 29th...I haven't rode it yet, but from what I've read it's alot better than MF, and some say better than fury; and if you're addicted to borderline grey-out, I believe them. It may take a lot of work to fine tune it, I'm willing to help....

I've always found it interesting to see where the B:TR clones wind up lists. The scores are fairly consistent which in my mind helps validate (to a certain extent) the scoring system of a given poll.

Lord Gonchar said:

Point is, sticking rigidly to statistical significance creates situations where the list is useless because it's wrong due to incompleteness.

What you're describing is two diametrically opposed concepts. Statistical significance is completeness.

hendrixmarshalljames said:

i305 is ranked 29th...I haven't rode it yet, but from what I've read it's alot better than MF, and some say better than fury; and if you're addicted to borderline grey-out, I believe them.

I have ridden both Millennium Force and I305, and while I prefer I305, I've heard just as much negative feedback on it as I have positive. So 29th is quite possibly a fair and accurate ranking for it.

I greatly prefer MF. I305 is a marvel of intensity and power but it doesn't really speak to me as a coaster rider. It's all power and brain scrambling; no art or drama.

For whatever reason, the Coasterbuzz hive-mind seems to not be completely enamored with Intamin's super mega intense steel rides, relative to other forums and polls I've observed. Note Skyrush too at #36, compared to its #7 and #5 showings on Mitch's poll: http://ushsho.com/steelpoll13yeartable2013.htm Even after pulling the woodies out of the Coasterbuzz 100 that's still a big difference, especially since Mitch's poll has the international contigent that the Cbuzz poll does not.

Jeff said:

Statistical significance is completeness.

It is completeness of process.

Which is exactly my point. Getting hung up in the completeness of the process impedes results at times. (and I know you read that line and it breaks your brain because it makes no sense to you...that just not how it works)

I believe the process doesn't have to be complete to generate reasonable accuracy for a list of rollie coasters. Sticking with the example, Outlaw Run's spot was not going to change dramatically if it was included at 35 votes instead of 38 or 33 or 30 or probably even 25. None of those levels would be statistically significant, yet they would produce results well within what I think would be the acceptable range for pretty much anyone bothering to look at the list.

I would be curious as to what coasters would pop into the top 100 if they just had another rating or five. What rides are teetering on the brink of significance?

matt. said:

Note Skyrush too at #36, compared to its #7 and #5 showings on Mitch's poll: http://ushsho.com/steelpoll13yeartable2013.htm Even after pulling the woodies out of the Coasterbuzz 100 that's still a big difference, especially since Mitch's poll has the international contigent that the Cbuzz poll does not.

Probably because Mitch's poll isn't statistically significant.

hendrixmarshalljames said:

I just read the list , and i305 is ranked 29th...I haven't rode it yet, but from what I've read it's alot better than MF, and some say better than fury; and if you're addicted to borderline grey-out, I believe them.

If I might add my two cents, 29 is still too high.

And it dropped to 30 this week. The coaster that's much too low is the new 29 - Phantom's Revenge.

#29 - the most disrespected number on the list. When will The Beast land there.

Re: Mitch's poll, sure. There's the case that methodology differences will sure lead to different results.

Still, I think it's easy to unscientifically observe that CBuzz's taste as a group will be different from the sub-populations represented by Toilet Paper Rollers, old school ACErs, Mitch's voters, etc. Does Mitch's poll *really* overrate i305 and Skyrush that much purely purely for that reason?

I was being facetious (and taking a dickhead jab at the statistical significance thing).

No, I don't think Mitch's poll is overrating it - and certainly not that much for that reason. If anything, I think you nailed it in the first post - there's a certain kind of ride the CoasterBuzz crowd doesn't like as much as a larger, more general enthusiast group might or does.

Interestingly enough, I wonder if we skew older as a group?

Lord Gonchar said:

Jeff said:

Statistical significance is completeness.

It is completeness of process.

No... a thousand times no. I've brought this up a hundred times and I don't know how else to put it to make you understand. The larger the sample, the more significant the results are and the more they're likely to be "right." That isn't my opinion, it's the foundation of statistics. It's why drug trials aren't particularly meaningful with a dozen patients but pretty important when there are a thousand patients. It's why a political poll of a dozen people from Billings, Montana is unimportant compared to one sampling tens of thousands of people from all over the country.

Lord Gonchar said:

Interestingly enough, I wonder if we skew older as a group?

I was about to speculate on some of the reasons for why the taste of this group might be different, and that was one of them. No hard data to back that up though. While other forums have a significant number of posters who very, very, very much seem to be teeny boppers, there aren't so many here (mostly because they get weeded out so quick).

Another thing I was thinking about was geography. CBuzz still seems to have that lingering Ohio / Midwest / flyover state centric thing going for it, when Toilet Paper Rollers for example still seems to have a much larger California contingent.

Jeff said:

No... a thousand times no. I've brought this up a hundred times and I don't know how else to put it to make you understand. The larger the sample, the more significant the results are and the more they're likely to be "right." That isn't my opinion, it's the foundation of statistics. It's why drug trials aren't particularly meaningful with a dozen patients but pretty important when there are a thousand patients. It's why a political poll of a dozen people from Billings, Montana is unimportant compared to one sampling tens of thousands of people from all over the country.

Great.

Your list would be just as useful (probably more so) not adhering so strictly to statistical significance.

I don't know how to make you understand. You're stuck on how you get the list, not what the list actually says.

Coasterbuzzers have deemed Outlaw Run a top ten coaster. For the last three years, your list didn't even include it. You certainly have a list that's statistically significant, but at what point does it cease to be useful?

You're worried about process and I feel results are more important. You argue that results that don't adhere to the process are incorrect or broken. I argue that a more useful list of roller coasters can be achieved without adhering so strictly to the process.

Notice I'm making a distinction between usefulness and accuracy. Obviously more data is more accurate. But waiting for more data is less useful in the short term.

Or to sum it up (as I have before) - My take is that a list that's 'close' (even remotely in the ballpark) even if the data isn't there, will always be better than a list that's incomplete but has the stats behind it. And eventually the stats will catch up and correct it on the fly - the difference is you have something in place.

And I'm not suggesting the list suddenly gets filled with rides that have one five-star vote either. You have to draw the line somewhere. I just don't think you have to draw it at actual scientific statistical significance to make a worthy list.

But whatever, we've done this dance a million times.

matt. said:

I was about to speculate on some of the reasons for why the taste of this group might be different, and that was one of them. No hard data to back that up though. While other forums have a significant number of posters who very, very, very much seem to be teeny boppers, there aren't so many here (mostly because they get weeded out so quick).

Another thing I was thinking about was geography. CBuzz still seems to have that lingering Ohio / Midwest / flyover state centric thing going for it, when Toilet Paper Rollers for example still seems to have a much larger California contingent.

I think you're right on both counts. That's why it's fun to look at the different polls.

I think I've said it before and I'm stupid for still not doing it, but we need a "Trivago" style site that collects all the various poll data and displays it in one place...perhaps even creating it's own poll based on the aggregate data.

You know for all two lists that exist. :)

And did we ever establish why Mitch's poll went away?

I wonder if there's a way to compromise. Could the list remain as is with the minimum ridership requirement in tact, but a separate "On the Horizon" or "Coming Attractions" or "Rising the Lift Hill" (whatever, you get the idea...) list be created of like no more than 5 - 10 or so coasters that have the promise of a high ranking, but not enough ridership just yet?

I have to admit that I kind of like that there's a minimum ridership to the main list. Lord knows that I have different coaster tastes than a lot of enthusiasts. And I would propose that just as it's been stated that CBuzz overall has a distinct taste over other park goers, the same may be true of the subset of enthusiasts who make it out to a coaster opening at the onset. But I can see where it would be interesting to know what those early opinions are shaping up to be, so maybe the second list could complement the main one.

Just a thought.

An interesting discussion. I tend to prefer a list generated from Cbuzzers with a certain number of coasters under their belt; that suggests to me that the opinion of those with broader experience is given more weight.

A demographic breakdown of Coasterbuzz would be fascinating!

My take is that a list that's 'close' (even remotely in the ballpark) even if the data isn't there, will always be better than a list that's incomplete but has the stats behind it.

I don't know what to tell you, Gonch. You're hell bent on convincing me that hundreds of years of statistics as a practice don't matter. An incomplete list does not have the stats behind it. That's why I'm discontent with my list, because there are a lot of rides that don't have statistical significance to include them, and those at the current cut off frankly have too small of a sample.

Carrie J. said:

I wonder if there's a way to compromise. Could the list remain as is with the minimum ridership requirement in tact, but a separate "On the Horizon" or "Coming Attractions" or "Rising the Lift Hill" (whatever, you get the idea...) list be created of like no more than 5 - 10 or so coasters that have the promise of a high ranking, but not enough ridership just yet?

While reading this topic, I was thinking something along the same lines. I think it's a fantastic idea worth taking a serious look at. It would give some people an idea of which rides they should attempt to get to, to ride and rate if they want to help get it on the Top 100 list. It would give an idea of what rides are considered well-liked, but just haven't had enough riders yet to make it to the list.

Jeff said:

I don't know what to tell you, Gonch. You're hell bent on convincing me that hundreds of years of statistics as a practice don't matter. An incomplete list does not have the stats behind it. That's why I'm discontent with my list, because there are a lot of rides that don't have statistical significance to include them, and those at the current cut off frankly have too small of a sample.

No. I'm hell bent on the idea of convincing you that a 'best coaster' list doesn't need hundreds of years of statistics as a practice behind it to be useful.

I get what you're saying though. Based on how you would do it ideally, you're already making the compromise that I'm suggesting.

Maybe something where all the tracks record count, but the more experienced you are, the more your opinion counts for. Something as simple as a multiplier for different levels of experience. The most inexperienced 20%'s track record counts as a single rating. The next 20%'s count as two. The next 20%'s as three and so on. Each rating by someone among the top 20% would carry five times the weight of a newbie's opinion.

Plus, that multiplier serves as a 'filler' of sorts to create an artificial significance, if you will. Right now we're looking at about 40 votes to get on the list. That could be 40 newbs. 20 barely above newbs. 13 or 14 riders from the middle third. 10 pretty experienced riders. Or just 8 of the most experienced riders (who's opinion I would suggest it's ok to trust at that level).

Or whatever combination of all of the above the case may be. If some inexperienced rider gets on a wholly average ride but thinks it's just the bees knees and gives it five stars, a single experienced rider giving a more representative three stars will balance that rating down to a 3.33333.

It's not perfect but it's creating data to fill in the missing pieces based on something tangible. The only real downside might be a very experienced rider that has quirky tastes, but still that should be balanced by the sheer number of riders that don't share those quirks.

You could even break it into smaller groups (15%, 10%, whatever) or change the multiplier to fine tune it further.

Is it messing with a lot of numbers? Sure, but it combats all the issues with number of votes and taking the experience of the rider into account.

And it's fun. :)