Separate names with a comma.
Discussion in 'The OT' started by fluffybear, Oct 9, 2012.
From Fox News:
They must not have used any of the establishments in my area for this survey as Wendy's is by far the slowest. I don't think there has been a time int he last few years that I have not been told to pull up and someone will bring me my food.
Wendy's has been the fastest for at least a decade.
I remember back in 2002/2003, Arby's had a big push to try and beat Wendy's, and spent 2 years developing store contests/incentives and corporate personnel spend a lot of time in the field timing drive thrus and providing feedback. The end results, IIRC, was that they gained about 1 second in their overall time and were still close to last in the study.
Wendy's is definitely the fastest (and best) here, Burger King takes foreeeeeevvvvveeeeerrrrr, I don't even eat there.
I don't trust any of those numbers. I remember the Wendy's by USC had a bucket on a string they would lower out if the pick up window right after you ordered at the menu so the sensor would think a car came and went in like 1 second. Not to mention the McDonalds trick of asking you to pull over and someone will come out with your food later. At least in LA, those numbers are cooked.
I think the whole survey is kind of ridiculous. After all, it all depends on each restaurant location's pace, which can be affected by a variety of factors. I can go to 3 different McDonald's and 2 Wendy's near my abode, and each is very different when it comes to the speed of their service.
And yet we have survey companies call 600 people and tell us what millions are thinking - and we believe it (plus or minus 4%).
Today I had a longer than usual wait at Wendy's ... and I was the only car in line. But a quick look at the figures - 129 seconds to 201 for the whole field? Not worth worrying about. I'm more concerned with hot and fresh.
BTW: The real winner would be Little Ceasers. In, out and on the road in under a minute unless one wants more than the basic pizza and breadsticks. They don't call it "Hot and Now" for nothing.
That has nothing to do with the QSR Magazine survey. Some restaurants will do that to lower their times and impress their managers or district managers (or the franchisee), but that's all internal.
The QSR Survey is done manually by actual people placing actual orders and using stopwatches. They're done at random locations around the country. Does that mean all Wendy's are fast? No. But it does say something that Wendy's wins it every year.
Yeah... our local news station recently talked about a poll (details unimportant because it is political and non-essential to this topic)... and had a segment on the local news + an article on their Web site... where they proudly proclaimed 90% favored some particular thing... only, in the second paragraph they noted the survey polled 641 people!
I don't know how many million people we have in NC... but polling 641 people is hardly a good random sampling to extrapolate and make a big political statement around!
You're making the common error of confusing survey validity with sample size. They are actually only very slightly related.
For example, a valid survey can be had at a 99% confidence level (how "valid" the survey is), with a 5% confidence interval (the "margin of error") on a population of 25,000,000 (or any "large" population as at a certain size, population essentially becomes irrelevant) using a sample size of...you guessed it, only 667!
Additionally, the higher the percentage of the sample size that picks a particular answer, in your cited example 90%, the smaller the sample size required for a valid survey.
The question here is are they all ordering the same thing? Are the vehicles ahead of them ordering the same thing? Are they visiting at the same time of day? There are a lot of factors which need to be looked at in order to make it a fair survey.
All those things are looked at.
They typically order a basic combo. They visit all restaurants at different times (but primarily over lunch) and combine the results. Other cars in line are factored into the equation.
It's as fair as it can get.
+1. They must have missed all the Wendy's I've been to. Because, short of airport Wendy's, the Wendy's I've been to in Houston and North Carolina have all been 5 to 10+ minute affairs from the time I ordered to the time I finally got the food.
I really like the food. However, because they are so slow, I rarely ever eat at Wendy's. Whataburger and Chick-fil-a are the same. Good food, but it takes too long to get it.
Actually, I would argue YOU are the one making the common error... extrapolating data from hundreds of people into millions of people's habits.
Statistics are a dangerous thing anyway... but especially so when applied to predicting human behavior on a large scale. There's just no way to slice the population in a way that is reliably accurate with such a small sample size like you are suggesting.
It is way too easy to find skewed results with small sample sizes. Ignoring all other aspects, the first thing that comes into play with any poll or survey... Some people participate, others do not... The type of people who participate tend to skew one way, and people who don't participate might skew the other... you have no way of knowing, because they don't participate.
It's like assuming the food is good at a restaurant because nobody sends their food back. IF you rate the restaurant by how many (or few) send their food back to the kitchen you might assume you have high quality food... but in truth, most people will not bother to complain unless it is REALLY BAD... they will just not come back and eat again... so as a restaurant you can be losing money hand over fist, because your food is terrible... but have very few complaints registered.
It would be folly to assume from a small sample size, and ignoring people who don't participate at all, anything meaningful for the larger group.
You could, but you would be quite incorrect, not to mention demonstrating your apparent complete lack of either training or understanding of statistical methods.
Perhaps, but I argue that you are demonstrating an apparent complete lack of either training or understanding of human behavior.
Besides the fact that statistics can be twisted to find what you were originally looking for...
Statistics in a small sample size can be relatively useless for predicting human behavior.
There are millions of people in NC... What sample size of 600 or so people are you going to find anywhere that is representative of the entire state's beliefs?
We have farmers and high technology and students and retirees and I can't even imagine the whole grey area spectrum in between. There's simply no way to take a poll of 600 or so people and come to any reasonable conclusion about the views of the entire state's population on average.
Even for the things that statistics does well... in predicting natural events... human behavior is very unpredictable... so taking a survey along the lines of "do you like politician A or politician B" and extrapolating from 600 people to millions is irresponsible.
If you're really hanging your hat on that, I don't see where productive conversation on that particular topic can proceed.
While it may seem like you'd be right, you're actually wrong. Statistics is all about Math and Math is a funny thing (for example, did you know that if you have a random group of 30 people, odds are that two of them will have the same birthday? Doesn't seem possible with 365 different days does it? But is is)
As long as the sample, is RANDOM, small samples can be very accurate (statistically speaking).
HERE'S a pretty good article about it.
And HERE'S a calculator.
Of course there is. Whether or not you choose to believe it doesn't change established and proven scientific and mathematical fact.
You need a valid cross section of the population. Ask 600 people in the red state of Texas who they're voting for. Well, those 600 can't all be black...the results won't represent the state.
You need to first establish what information you are seeking. Then, you seek a random sampling (what you refer to as a "valid cross section") of the group from which you seek that information (if I want to know what whites think, I don't sample blacks. If I want to know what Texans think, I don't sample Kansans. If I want to know what white Texans think, I don't sample...well, you get the idea) using well-established and proven methods. So long as randomness is adhered to from within the targeted group, voilà, we have statistical validity using a relatively small sample size! It's science, but it ain't rocket science.