Yeah, I have an epi degree, so I'm perfectly aware of what the median is. Median and mean both have their advantages and disadvantages, and give you different information.
If you have a distribution that's skewed, the median gives you a more apropriateview of what the typical (i.e. 50th percentile) person is spending. Of course, if there's a peak at both ends and a valley in the middle, the median may not be to helpful.
Obviously, a lovely graph of the actual distribution is what will tell you the most.
I maintain that, in the context of the discussion that's been going on, the median is likely more informative. Having both would be best.
And I continue to say that without data, you are compounding the problem. You know very well that you can't assume anything that you don't know to be true. The only thing we know is the average. nothing else can be assumed.
What would be nice is if we knew the standard deviation. This would reveal how accurae the mean is. A small standard deviation, say about $250, would tell us that 68% of the data points was withn + and - $250 of the mean. And that the mean is truly reflective of the real average amount college students paid for a laptop. At this point the mean and median (and maybe also the midpoint) should be very close to each other.
A large standard deviation, say about $600, tells us that the mean is not so accurate as %68 of the data fell + and - $600 from the mean. That's a large range and most likely caused by a peak in the data at one or both end. It is at this point (like you stated) that the median may be a more revealing value (providing the peak is only at one end) of the true average amount paid for a laptop by college students.
I (we) can only assume that the data taker knew enough about statistic to take enough data so that a normal distribution curve (a Gaussian Bell) forms in the data. Otherwise the mean, median and midpoint may end up being a value where the data points are the least dense and thus are not very accurrate indicators of what college students are paying for laptops.
Anyway, we can't assume any of this as we have no raw data, or even a generalized statement from them about deviation, endpoints, etc.
I tend to believe the number stated is a good one, from what I see, and it's the only one we can go by.
Otherwise, you know that the statistically most likely curve is the gaussian distribution, and that curve sits in the middle of the universe of possible curves determined by a gaussian distribution.
If you're talking about height sure.
We're talking about consumer electronics. Do you dispute that computers that cost say, $600, sell in greater numbers than computers that cost $2000?
That's 700 on either side of a $1300 average. To believe the distribution roughly normal, those computers have to sell in roughly equal quantities.
I'm not disputing the average isn't good. Using just the average to draw inferences, with no other data, one has to realize that one can only have slim confidence in conclusions drawn from that.
It is all we have to go on, and I'm just saying that ain't worth much.
And I notice that apple closed today over $143...what happened to those predictions of going below $100? I guess if students buy as this poll suggests, maybe the company isn't in such bad shape after all?
There was only one guy here that was insisting that it would drop to $80. His name is g3pro, and he's been scarce since his prediction.
We're talking about consumer electronics. Do you dispute that computers that cost say, $600, sell in greater numbers than computers that cost $2000?
That's 700 on either side of a $1300 average. To believe the distribution roughly normal, those computers have to sell in roughly equal quantities.
I'm not disputing the average isn't good. Using just the average to draw inferences, with no other data, one has to realize that one can only have slim confidence in conclusions drawn from that.
It is all we have to go on, and I'm just saying that ain't worth much.
All I'm saying is that, like yu, I see that there is little data, the only data being the average price.
While I have no problem agreeing that it's likely that $700 computers sell more than $2,000 models, there are two questions here to look at.
The first is who is buying these computers, and for what purpose, and the second is who is paying for them.
I don't believe that the average computer buying freshman is equal to the average computer buyer. And therein lies the problem.
Without knowing the differences, we simply can't say anything about this without going to most statistically likely curve that would fit whatever data there may be, and that curve is the normal curve.
Don't forget that students going into engineering, physics, biology, and other "hard" sciences are more likely to buy that MBP, and that will help to bring that average up.
I also find it hard to believe that there are more than a few students buying cheap laptops much under that $700 you mentioned.
Don't forget that students going into engineering, physics, biology, and other "hard" sciences are more likely to buy that MBP, and that will help to bring that average up.
I also find it hard to believe that there are more than a few students buying cheap laptops much under that $700 you mentioned.
Who's making assumptions now?
Seriously though, we really don't disagree on anything.
Seriously though, we really don't disagree on anything.
True, but I wouldn't attempt to put those assumptions into a curve, unless we want to play with the numbers to see a logical way to come up with that $1,290 number.
I wonder how much of this is due to the hype that is going around in the computer world about how 'cool' Apple computers are, I hope is not just a fad but a long term switch into more choices for the users in general.
Seriously though, we really don't disagree on anything.
I was thinking about this after I finally got offline last night, and was lying in bed.
Actually, we do have some corresponding data.
I know that there are a few who don't wan't to believe some of it, but it is there.
One is the $1,290 average price paid by college freshmen that we're discussing.
Two is the sales of Macs, particularly portables, rising three times as fast as that of PCs the past few years. This is not in dispute. As part of that, we see in Feb, and hopefully, repeated again in March, that Mac sales are 60% higher year over year. It is about 6 times the PC sales growth for the month. That is preliminary data, until Apple releases the actual numbers later in April, but is likely close enough. This is also important, because as we well know, business is a small portion of Mac sales, so consumer and educational purchases predominate. And of course, the Sept quarter, which is the heavy school purchasing quarter that Apple always does so well in, and is expected to do very well in again this year, when these students are expected to be buying these computers this year for the following school year, has yet to come up.
Third is the information that of the 1,200 students polled (subject to about a plus or minus 3% error, assuming the poll was done properly), 40% claimed to be purchasing a Mac as their next computer. This number is in dispute between myself and cameronj, who refuses to beleve it's close to being true, but nevertheless is supported by numerous college data sets given by the colleges themselves, along with more colleges every day requiring a student to purchase a Mac. I provided a few links to that information.
With all three of those bits of information, and knowing the educational pricing of Macs, we can see the number of $1,290 as being much more likely to reflect an actual, useful average, residing somewhere smack dab in the middle of the curve, rather than being a couple of deviations over to the side.
I would still like to have the actual data, but this does tend to support it, from numerous independent sources.
We're talking about consumer electronics. Do you dispute that computers that cost say, $600, sell in greater numbers than computers that cost $2000?
That's 700 on either side of a $1300 average. To believe the distribution roughly normal, those computers have to sell in roughly equal quantities.
I'm not disputing the average isn't good. Using just the average to draw inferences, with no other data, one has to realize that one can only have slim confidence in conclusions drawn from that.
It is all we have to go on, and I'm just saying that ain't worth much.
The question is not whether there are as many consumers buying $700 computers as there are $2000 (and over) computers. The question is whether there are as many consumers that are buying $700 AND $2000 (and over) computers than there are consumers buying computers near the average. And the anwser is most likely no.
You see, we don't have to assume whether the data is "good" or not. We can almost be sure of it. The data wasn't collected by some Joe Shmo doing a blog. The data was collected and analyzed by a Morgan Stanley analyst. If anyone knows how to collect and analyze data, you would figure that an analyst for an investment banker would. Tens of millions of dollars can change hand based on them analyzing data and drawing a conclusion based on that data. I doubt very much that someone that makes a living by collecting and analyzing data, so that they can inform their clients on how best to invest their money, would not know what "bad" data looks like. I doubt very much that they would come to a conclusion of a value for the "average" if it wasn't around where most of the data points fell. I doubt very much that the average they came up with was some phantom value between two peaks in the data. The average is around where most of the data points fell. The average is near the top of a bell curve, around where the data points are most dense. Not in a valley, where the data is least dense. If I know this much about analyzing data, then someone working for an investment banker (like Morgan Stanley) must surely know much more.
Plus an analyst don't need to lie (That's not to say that they don't.). They make money for their clients whether a stock goes up or down. They get pay for coming up with investment advice. They collect data and use statistic as a way to arrive at a meaningful conclusion. Not the other way around. It's the people that don't want to believe their numbers that somehow want to find some other way to analyze the data or disprove the existing data. These people have already arrived at a conclusion without the help of any data. And will try to find some statistical means to prove what they already think they know. Which gives creedence to the phrase "There are lies, damn lies and statistics."
The data was collected and analyzed by a Morgan Stanley analyst. If anyone knows how to collect and analyze data, you would figure that an analyst for an investment banker would.
<whump>
<roll_on_floor>
But seriously,
Just about the entire set of large scale economic problems the US is facing are because investment banker analysts couldn't predict risk correctly or even bother to do any cross checks on their raw data from sub-prime mortgages. Enabled the whole damn housing bubble in the first place.
And you think with analyst's firms PUBLICLY backing away from their statements they collectively have any more clue than my Beagle about WTF IS REALLY GOING ON?
Apple is posting huge gains. We are witnessing the beginning of the end for Windows and the grief Microsoft has inflicted upon users.
Don't be mistaken, MS will be around for a very, very long time. Even when the fall of Rome was imminent it still took centuries for the empire to crumble, even well after the last emperor abrogated the throne. But we have yet to see that happen with MS.
In 2007, MS grossed $51B and netted $14B, while Apple only grossed $24B and netted $3.5B. MS may be hemorrhaging billions in some areas but they are still managing to make double the profit margin Apple is... and that won't be changing overnight.
From what I've seen, hardly any students buy a desktop, be it an Apple or anything else, and then a large majoirty buying a laptop are opting for a Mac.
Comments
Yeah, I have an epi degree, so I'm perfectly aware of what the median is. Median and mean both have their advantages and disadvantages, and give you different information.
If you have a distribution that's skewed, the median gives you a more apropriateview of what the typical (i.e. 50th percentile) person is spending. Of course, if there's a peak at both ends and a valley in the middle, the median may not be to helpful.
Obviously, a lovely graph of the actual distribution is what will tell you the most.
I maintain that, in the context of the discussion that's been going on, the median is likely more informative. Having both would be best.
And I continue to say that without data, you are compounding the problem. You know very well that you can't assume anything that you don't know to be true. The only thing we know is the average. nothing else can be assumed.
What would be nice is if we knew the standard deviation. This would reveal how accurae the mean is. A small standard deviation, say about $250, would tell us that 68% of the data points was withn + and - $250 of the mean. And that the mean is truly reflective of the real average amount college students paid for a laptop. At this point the mean and median (and maybe also the midpoint) should be very close to each other.
A large standard deviation, say about $600, tells us that the mean is not so accurate as %68 of the data fell + and - $600 from the mean. That's a large range and most likely caused by a peak in the data at one or both end. It is at this point (like you stated) that the median may be a more revealing value (providing the peak is only at one end) of the true average amount paid for a laptop by college students.
I (we) can only assume that the data taker knew enough about statistic to take enough data so that a normal distribution curve (a Gaussian Bell) forms in the data. Otherwise the mean, median and midpoint may end up being a value where the data points are the least dense and thus are not very accurrate indicators of what college students are paying for laptops.
Anyway, we can't assume any of this as we have no raw data, or even a generalized statement from them about deviation, endpoints, etc.
I tend to believe the number stated is a good one, from what I see, and it's the only one we can go by.
Otherwise, you know that the statistically most likely curve is the gaussian distribution, and that curve sits in the middle of the universe of possible curves determined by a gaussian distribution.
If you're talking about height sure.
We're talking about consumer electronics. Do you dispute that computers that cost say, $600, sell in greater numbers than computers that cost $2000?
That's 700 on either side of a $1300 average. To believe the distribution roughly normal, those computers have to sell in roughly equal quantities.
I'm not disputing the average isn't good. Using just the average to draw inferences, with no other data, one has to realize that one can only have slim confidence in conclusions drawn from that.
It is all we have to go on, and I'm just saying that ain't worth much.
I
And I notice that apple closed today over $143...what happened to those predictions of going below $100? I guess if students buy as this poll suggests, maybe the company isn't in such bad shape after all?
There was only one guy here that was insisting that it would drop to $80. His name is g3pro, and he's been scarce since his prediction.
If you're talking about height sure.
We're talking about consumer electronics. Do you dispute that computers that cost say, $600, sell in greater numbers than computers that cost $2000?
That's 700 on either side of a $1300 average. To believe the distribution roughly normal, those computers have to sell in roughly equal quantities.
I'm not disputing the average isn't good. Using just the average to draw inferences, with no other data, one has to realize that one can only have slim confidence in conclusions drawn from that.
It is all we have to go on, and I'm just saying that ain't worth much.
All I'm saying is that, like yu, I see that there is little data, the only data being the average price.
While I have no problem agreeing that it's likely that $700 computers sell more than $2,000 models, there are two questions here to look at.
The first is who is buying these computers, and for what purpose, and the second is who is paying for them.
I don't believe that the average computer buying freshman is equal to the average computer buyer. And therein lies the problem.
Without knowing the differences, we simply can't say anything about this without going to most statistically likely curve that would fit whatever data there may be, and that curve is the normal curve.
Don't forget that students going into engineering, physics, biology, and other "hard" sciences are more likely to buy that MBP, and that will help to bring that average up.
I also find it hard to believe that there are more than a few students buying cheap laptops much under that $700 you mentioned.
Don't forget that students going into engineering, physics, biology, and other "hard" sciences are more likely to buy that MBP, and that will help to bring that average up.
I also find it hard to believe that there are more than a few students buying cheap laptops much under that $700 you mentioned.
Who's making assumptions now?
Seriously though, we really don't disagree on anything.
Who's making assumptions now?
Seriously though, we really don't disagree on anything.
True, but I wouldn't attempt to put those assumptions into a curve, unless we want to play with the numbers to see a logical way to come up with that $1,290 number.
And, yes, I think we do agree.
Who's making assumptions now?
Seriously though, we really don't disagree on anything.
I was thinking about this after I finally got offline last night, and was lying in bed.
Actually, we do have some corresponding data.
I know that there are a few who don't wan't to believe some of it, but it is there.
One is the $1,290 average price paid by college freshmen that we're discussing.
Two is the sales of Macs, particularly portables, rising three times as fast as that of PCs the past few years. This is not in dispute. As part of that, we see in Feb, and hopefully, repeated again in March, that Mac sales are 60% higher year over year. It is about 6 times the PC sales growth for the month. That is preliminary data, until Apple releases the actual numbers later in April, but is likely close enough. This is also important, because as we well know, business is a small portion of Mac sales, so consumer and educational purchases predominate. And of course, the Sept quarter, which is the heavy school purchasing quarter that Apple always does so well in, and is expected to do very well in again this year, when these students are expected to be buying these computers this year for the following school year, has yet to come up.
Third is the information that of the 1,200 students polled (subject to about a plus or minus 3% error, assuming the poll was done properly), 40% claimed to be purchasing a Mac as their next computer. This number is in dispute between myself and cameronj, who refuses to beleve it's close to being true, but nevertheless is supported by numerous college data sets given by the colleges themselves, along with more colleges every day requiring a student to purchase a Mac. I provided a few links to that information.
With all three of those bits of information, and knowing the educational pricing of Macs, we can see the number of $1,290 as being much more likely to reflect an actual, useful average, residing somewhere smack dab in the middle of the curve, rather than being a couple of deviations over to the side.
I would still like to have the actual data, but this does tend to support it, from numerous independent sources.
If you're talking about height sure.
We're talking about consumer electronics. Do you dispute that computers that cost say, $600, sell in greater numbers than computers that cost $2000?
That's 700 on either side of a $1300 average. To believe the distribution roughly normal, those computers have to sell in roughly equal quantities.
I'm not disputing the average isn't good. Using just the average to draw inferences, with no other data, one has to realize that one can only have slim confidence in conclusions drawn from that.
It is all we have to go on, and I'm just saying that ain't worth much.
The question is not whether there are as many consumers buying $700 computers as there are $2000 (and over) computers. The question is whether there are as many consumers that are buying $700 AND $2000 (and over) computers than there are consumers buying computers near the average. And the anwser is most likely no.
You see, we don't have to assume whether the data is "good" or not. We can almost be sure of it. The data wasn't collected by some Joe Shmo doing a blog. The data was collected and analyzed by a Morgan Stanley analyst. If anyone knows how to collect and analyze data, you would figure that an analyst for an investment banker would. Tens of millions of dollars can change hand based on them analyzing data and drawing a conclusion based on that data. I doubt very much that someone that makes a living by collecting and analyzing data, so that they can inform their clients on how best to invest their money, would not know what "bad" data looks like. I doubt very much that they would come to a conclusion of a value for the "average" if it wasn't around where most of the data points fell. I doubt very much that the average they came up with was some phantom value between two peaks in the data. The average is around where most of the data points fell. The average is near the top of a bell curve, around where the data points are most dense. Not in a valley, where the data is least dense. If I know this much about analyzing data, then someone working for an investment banker (like Morgan Stanley) must surely know much more.
Plus an analyst don't need to lie (That's not to say that they don't.). They make money for their clients whether a stock goes up or down. They get pay for coming up with investment advice. They collect data and use statistic as a way to arrive at a meaningful conclusion. Not the other way around. It's the people that don't want to believe their numbers that somehow want to find some other way to analyze the data or disprove the existing data. These people have already arrived at a conclusion without the help of any data. And will try to find some statistical means to prove what they already think they know. Which gives creedence to the phrase "There are lies, damn lies and statistics."
The data was collected and analyzed by a Morgan Stanley analyst. If anyone knows how to collect and analyze data, you would figure that an analyst for an investment banker would.
<whump>
<roll_on_floor>
But seriously,
Just about the entire set of large scale economic problems the US is facing are because investment banker analysts couldn't predict risk correctly or even bother to do any cross checks on their raw data from sub-prime mortgages. Enabled the whole damn housing bubble in the first place.
And you think with analyst's firms PUBLICLY backing away from their statements they collectively have any more clue than my Beagle about WTF IS REALLY GOING ON?
<whump>
Apple is posting huge gains. We are witnessing the beginning of the end for Windows and the grief Microsoft has inflicted upon users.
Don't be mistaken, MS will be around for a very, very long time. Even when the fall of Rome was imminent it still took centuries for the empire to crumble, even well after the last emperor abrogated the throne. But we have yet to see that happen with MS.
In 2007, MS grossed $51B and netted $14B, while Apple only grossed $24B and netted $3.5B. MS may be hemorrhaging billions in some areas but they are still managing to make double the profit margin Apple is... and that won't be changing overnight.
In our Comp Sci lectures Macs are the majority.