kottila said:Haven't we learned by now that Wii games often start out slow?
HoTD Overkill debuted higher than any previous light gun game before it so why is there any negativity???
kottila said:Haven't we learned by now that Wii games often start out slow?
I believe the original argument was about software sales and that the wii saw a boost in the number of charting titles based off of kiddy titles during the holidays, and I have bolded at least one place above where you have made ad hoc changes to your hypothesis.doicare said:Uh huh, what i said was not a throw away comment.
Sooooo, how exactly have i altered my argument? How am i 'carefully' choosing my data?
As for the holiday periods, they cover pal weeks 7,8,14,15,16,17,21,22,23,30,31,32,33,34,35 with pal weeks 28,29 being the last two weeks of the summer term when very little work is done.
Did you make the comment or did i? I did, and you have no idea how i meant it so just because i chose to make the comment funny does not make it a 'throw away comment'.Weisheit said:It most certainly was:
Why would it make any logical sense for wii software to get a boost in only one holiday and not all of them? :lolWeisheit said:It was only after it was pointed out that Wii sw was strong for months previously, that your argument changed from summer holidays to every holiday in Britain.
Why is it anyone else's job to prove you're wrong? Don't you have to prove yourself correct first? The only thing you've proven is correlation, I can do that too:
By your logic Nintendo should be thanking the new moon for their sales...Code:Wk1 sw 14 (new moon) Wk2 sw 7 Wk3 sw 7 Wk4 sw 7 Wk5 sw 15 (new moon) Wk6 sw 8 Wk7 sw 7 Wk8 sw 16 (new moon) Wk9 sw 7 Wk10 sw 7 Wk11 sw 15 (new moon)
It's in the post right above yours.doicare said:I've provided plenty of proof with zero proof being presented against it.
You're getting confused with a general point of view i had and then me trying to prove it with the data that was publically available. The 'theory' i spoke about was that kids are a big factor in wii sales, and when kids have free time to do what they want a.k.a play games, wii sales (software and hardware) go up. Kids have the most free time during school holidays so that translates into higher wii sales then. That point of view has never changed. However during the process of trying to prove that point of view many varying avenues of sales/wii were talked about. Lots of other people brought many other things into the discussion and whilst i was happy to talk about them they were not the point of what i was trying to say.poppabk said:I believe the original argument was about software sales and that the wii saw a boost in the number of charting titles based off of kiddy titles during the holidays, and I have bolded at least one place above where you have made ad hoc changes to your hypothesis.
poppabk said:Anyway the data shows that at p=0.05 no console has any statistically significant trends in mean sales either over the entire year or in each of the three segments between holiday and non-holiday sales.
The closest to being significant is the negative trend of the 360 for the entire year with p=0.055. For the Wii the positive trend has a p=0.08 for the entire year and no better than 0.26 for any particular segment of the year.
Like i said what he has posted doesn't compare the relevent weeks correctly.Psychotext said:It's in the post right above yours.
Your "theory" is irrelevant. Wii is more popular amongst kids, nobody doubts that. Its more popular amongst every other demographic too, just like the PS2 was. The only problem I have with your "theory" is it seems like you're just a bitter Sony fan trying to cope with Nintendo's domination over Sony. Also, only those in denial maintained Sony had any chance this gen after $599...doicare said:Did you make the comment or did i? I did, and you have no idea how i meant it so just because i chose to make the comment funny does not make it a 'throw away comment'.
Why would it make any logical sense for wii software to get a boost in only one holiday and not all of them? :lol
If people want to be dicks and ridcule what i have said then they have to back up what they are saying with some proof, two 'wrongs' don't make a right. I've provided plenty of proof with zero proof being presented against it. By your standards i've proven a correlation, so why don't you ask yourself what is the most likely reason for that finding...?
Basically his theory is school kids are responsible for Wii sales, throw in slow DS software sales when the "data" doesn't correlate with his theory, using number of titles charts then switching to school dates as data points, changing these dates whenever he feels like, accusing of people being Nintendo fanboys.Captain Smoker said:What's this "theory" you're always talking about, I always run over these threads and everytime the same discussion. xD
BishopLamont said:Basically his theory is school kids are responsible for Wii sales, throw in slow DS software sales when the "data" doesn't correlate with his theory, using number of titles charts then switching to school dates as data points, changing these dates whenever he feels like, accusing of people being Nintendo fanboys.
"6000 hardware units is indication of a school holidays impact", saying his theory applies to school holidays then changing to every holiday, separating 360/PS3 initially only to combine them after when it's clear Wii software sells. Repeating the same tired points that have been refuted over and over again by multiple people, only to come back and repeat "You've yet to prove me wrong".
I mean you'd think he'd get the picture by now when people are mocking his theory even when he hasn't posted in the thread.
http://en.wikipedia.org/wiki/Student's_t-testdoicare said:I really have no idea what you have done with those numbers but it is a fact that the numbers show the wii's school holiday sales compared to the relevent non holiday sales received a boost.
Weisheit said:Your "theory" is irrelevant. Wii is more popular amongst kids, nobody doubts that. Its more popular amongst every other demographic too, just like the PS2 was. The only problem I have with your "theory" is it seems like you're just a bitter Sony fan trying to cope with Nintendo's domination over Sony. Also, only those in denial maintained Sony had any chance this gen after $599...
poppabk said:http://en.wikipedia.org/wiki/Student's_t-test
I have performed a t-test on the data, which looks at the two means relative to the variation and attributes the probability that the difference in the two means is due to chance or not. The data say that your fact at the 95% confidence level is not a fact.
Honestly the data will never show what you want it to show anyway.
Even if a few figures were different and the p value dipped below 0.05, it would just mean we could reject the null hypothesis (at 95% confidence) that the two different populations have the same mean. It still would not follow that the difference is related to school holidays. Just as the negative trend that the 360 shows does not demonstrate that once the kids go back to school parents run out and buy themselves a 360 and therefore the 360 is mainly owned by people with kids.
That's what Sony fans like yourself thought at least 5 months after launch. Everyone else (at gaf) not so much. Again $599 US DOLLARS.And again what's with the revisionist goggles looking back at what people thought of sony and how this gen was going to turn out? The xbox360 was just going to be another xbox, the wii was the 'death' of nintendo (do yourself a favour and look back at the thread where the wii was revealed) and the ps3 whilst expensive was going to still be number one with blu ray, cg killzone and all the water tight exclusives such as ffxIII, dmc4,virtua fighter,resevil5 etc...
People aren't debating the Wii's popularity amongst kids, but rather, your stupid theory that has no validity.You accept the wii is more popular amongst kids, i agree it is more popular amongst all demographics
poppabk said:http://en.wikipedia.org/wiki/Student's_t-test
I have performed a t-test on the data, which looks at the two means relative to the variation and attributes the probability that the difference in the two means is due to chance or not. The data say that your fact at the 95% confidence level is not a fact.
Honestly the data will never show what you want it to show anyway.
Even if a few figures were different and the p value dipped below 0.05, it would just mean we could reject the null hypothesis (at 95% confidence) that the two different populations have the same mean. It still would not follow that the difference is related to school holidays. Just as the negative trend that the 360 shows does not demonstrate that once the kids go back to school parents run out and buy themselves a 360 and therefore the 360 is mainly owned by people with kids.
Weisheit said:That's what Sony fans like yourself thought at least 5 months after launch. Everyone else (at gaf) not so much. Again $599 US DOLLARS.
People aren't debating the Wii's popularity amongst kids, but rather, your stupid theory that has no validity.
Again i see people have their revisionist goggles on. Many people predicted the wii bubble would burst shortly, not just people who post on forums but top sales analysts. Do you remember what happened with the N64? It started off selling sh*ttons but after a few months the bubble burst.BishopLamont said:Anyone thinking "Wii software sales are front-loaded due to the cheap price of the Wii" has no business proposing any kind of theories whatsoever.
It's only taken seriously by one poster......doicare said:No case closed here with my theory
I performed the t-test for the entire year (2-42) and then for each individual segment separately (2-8, 9-23, 24-42). The averages are not statistically significant in any individual segment. The t-test takes into account the average and the standard deviation within the populations to determine how confident you can be that any differences are not just due to chance. You may think that the averages of 87800 and 102000 are significantly different but given the variance of these populations they are not because the standard deviations are 18000 and 10000 respectively. Hopefully you can see that your "significant" difference is smaller than the standard deviation. And one standard deviation in a normal distribution only accounts for ~66% of the sample variance.doicare said:Well your t-test failed.
Weeks 2-6 which are the first school term have average weekly sales of 87,800. Weeks 7-8 which are the weeks of the first school holiday have average weekly sales of 102,000. That's an increase of 14,200 per week. That cannot be ignored, where's that in your t-test?
Same happens for the next period, weeks 9-23 there were 7 weeks of school holidays and 8 weeks of non school holidays. The average weekly sales of non holiday weeks was 105,000 whilst the school holiday average was 130,100. That's an increase of 25,100! Where's that show up in your test?
Then finally from weeks 24-42 (i would go further but that's where the charts data ends) there were 6-8 weeks of school holidays and 11 weeks on non holidays. Average non holiday sales were 114,800 whilst average holiday sales were 120,100. An increase of almost 6000.
If your test just takes every week that's a holiday and averages it out and compares to an average of every non holiday week then that is flawed because hardware sales have seasonal highs and lows and whilst you can compare some weeks that are close together you can't compare an entire years worth of data like that. Like wise for most holidays you can't compare every non holiday week before and after a holiday for the same above reason.
Only those with their heads firmly implanted in Sony's rectum thought this, including the analyst.doicare said:Again i see people have their revisionist goggles on. Many people predicted the wii bubble would burst shortly, not just people who post on forums but top sales analysts. Do you remember what happened with the N64? It started off selling sh*ttons but after a few months the bubble burst.
Again:I performed the t-test for the entire year (2-42) and then for each individual segment separately (2-8, 9-23, 24-42). The averages are not statistically significant in any individual segment. The t-test takes into account the average and the standard deviation within the populations to determine how confident you can be that any differences are not just due to chance. You may think that the averages of 87800 and 102000 are significantly different but given the variance of these populations they are not because the standard deviations are 18000 and 10000 respectively. Hopefully you can see that your "significant" difference is smaller than the standard deviation. And one standard deviation in a normal distribution only accounts for ~66% of the sample variance.
poppabk said:I performed the t-test for the entire year (2-42) and then for each individual segment separately (2-8, 9-23, 24-42). The averages are not statistically significant in any individual segment. The t-test takes into account the average and the standard deviation within the populations to determine how confident you can be that any differences are not just due to chance. You may think that the averages of 87800 and 102000 are significantly different but given the variance of these populations they are not because the standard deviations are 18000 and 10000 respectively. Hopefully you can see that your "significant" difference is smaller than the standard deviation. And one standard deviation in a normal distribution only accounts for ~66% of the sample variance.
doicare said:Well to be perfectly honest the t-test doesn't take into account many other things, ...
One last time. The null hypothesis for comparing two means is - there is no difference between the two means. To reject the null hypothesis you need data that are statistically different, ie you have to prove significance not non-significance. The data in this case are not different statistically at the 95% confidence level. The t-test is a very well accepted statistical test which is mathematically rigorous and has stood up to 100 years of scrutiny, it is certainly not laughable, and it is definitely not dismissible.doicare said:Well to be perfectly honest the t-test doesn't take into account many other things, and the t-test will never change the fact that the wii during the weeks i have compared against gains a weekly increase of anywhere in between 6000-25,000. Trying to chalk that up to 'statisical' noise is laughable when the vast majority of the time the preceeding week to the holidays has lower sales then the first week of the holiday and the first week after the holiday has finished has lower sales then the last week of the holiday.
poppabk said:One last time. The null hypothesis for comparing two means is - there is no difference between the two means. To reject the null hypothesis you need data that are statistically different, ie you have to prove significance not non-significance. The data in this case are not different statistically at the 95% confidence level. The t-test is a very well accepted statistical test which is mathematically rigorous and has stood up to 100 years of scrutiny, it is certainly not laughable, and it is definitely not dismissible.
The data do not support your original hypothesis that the means are significantly different. They may support your new hypothesis that the last week of non-holiday and the first week of a holiday are statistically different, but I am not going to bother checking because a) shifting from hypothesis to hypothesis until you get the answer you want is bad science and b) it wouldn't demonstrate anything remotely interesting anyway.
:loldoicare said:Just because some statisitcal test doesn't reconise 6000, 14,000 and 25,000 increases in weekly sales doesn't change the fact that they exist. They are significant and it also doesn't take into account the last week of non-holiday and the first week of a holiday sales match what i say, and f.y.i. i've been stating that fact going all the way back to when i was comparing weekly sales on the weighted sales chart over 4 months ago, so no i'm not changing my hypothesis until i get the answer i want because i already have the answer.
glolRaist said:
Here you go again repeating the same crap when he just refuted you. Jesus Christ dude you make Sony's work with percentages amateur in comparison.doicare said:Just because some statisitcal test doesn't reconise 6000, 14,000 and 25,000 increases in weekly sales doesn't change the fact that they exist. They are significant and it also doesn't take into account the last week of non-holiday and the first week of a holiday sales match what i say, and f.y.i. i've been stating that fact going all the way back to when i was comparing weekly sales on the weighted sales chart over 4 months ago, so no i'm not changing my hypothesis until i get the answer i want because i already have the answer.
poppabk said:Ok seriously for the last time.
Hypothetical situation.
Imagine 3 weeks of school time, followed by 3 weeks of non-school time
Case A
wk1 100000
wk2 101000
wk3 101000
wk4 102750
wk5 102000
wk6 102000
mean school - 100667 SD - 577
mean non-school - 102250 SD - 433
% increase = 1.5%
p=0.02 = statistically significant
Case B
wk1 100000
wk2 150000
wk3 50000
wk4 100000
wk5 200000
wk6 150000
mean school - 100000 SD - 50000
mean non-school - 150000 SD - 50000
% increase = 50%
p=0.28 = not statistically significant
The percent increase means nothing without taking into account variance. In case A the difference is very small but each population is distinct. In case B the % increase is large but is not significant because the difference can be accounted for just by the variance within the two groups.
BishopLamont said:Here you go again repeating the same crap when he just refuted you. Jesus Christ dude you make Sony's work with percentages amateur in comparison.
Reading doicare's replys, I don't think he has any idea what a t-test involves.
Thank you poppabk, no longer must we listen to his insane ramblings...poppabk said:doicare's theory being obliterated.
Oh goodness, you're questioning ironclad statistical test?doicare said:Well there we go, the t-test says a 25% increase in sales is insignificant, well i'm glad we cleared that one up, nothing to see here folks, statistical signifcants is more important than actual significants.
Sorry for screwing up your thread I know a lot of work goes into it, I will leave week 07 well enough alone .Raist said:t-tests now?
WTF
Why not ANOVA while we're at it.
Ok look, the numbers are NOT statistically significant (p around 0.1)
@doicare: he just used an example to show that a 1.5% increase might be significant, while a 50% increase might not be. IT all depends on the standard deviation.
Nevertheless even if it's not significant, it doesn't mean there's not a trend.
Is there such a trend?
YES
Is such a trend logical, or totally unexpected?
Considering that a very significant proportion of the Wii audience is family etc, of course sales would tend to go up during holidays. And of course, Nintendo is smart enough to release some of its carefully targeted games in such periods.
Does everyone agree? Ok? Yes?
Fine.
Now stop with this shit and move along.
Oh and LBP kicks ass and has nice legs and Sackboy is the cutest thing ever. There.
Raist said:t-tests now?
WTF
Why not ANOVA while we're at it.
Ok look, the numbers are NOT statistically significant (p around 0.1)
@doicare: he just used an example to show that a 1.5% increase might be significant, while a 50% increase might not be. IT all depends on the standard deviation.
Nevertheless even if it's not significant, it doesn't mean there's not a trend.
Is there such a trend?
YES
Is such a trend logical, or totally unexpected?
Considering that a very significant proportion of the Wii audience is family etc, of course sales would tend to go up during holidays. And of course, Nintendo is smart enough to release some of its carefully targeted games in such periods.
Does everyone agree? Ok? Yes?
Fine.
Now stop with this shit and move along.
Oh and LBP kicks ass and has nice legs and Sackboy is the cutest thing ever. There.
Weisheit said:Thank you poppabk, no longer must we listen to his insane ramblings...
Oh goodness, you're questioning ironclad statistical test?
Gosset or doicare? That's a tough one.Not.
You should take your own advice smart guy.doicare said:do everybody a favour and drop it,
Parl said:Recently, my main issue was with a claim that it's proven. My view is that most of these significant variations are software driven, which is pretty much what Raist said, and thankfully, I think doicare agreed.
We won't know how much "kids being off school" matters because Nintendo has been timing releases in a way which bumps up holidays, but that there's no significant unexplainable trend relating to hardware sales and holidays, any boost by kids being out of school is likely insignificant.
Lol, fair enough.doicare said:In a word no, and i think we'll leave it at that.