VB or C?

[=?ISO-8859-1?Q?Arne_Vajh=F8j?=]

And what is your basis for substantiating this analogy?

You do know that *most* people like Ginger Ale, right? I have no facts or
definitive statistics, but I asked 3 people and they all said it. All the
other people are irrelevant.

Bogus analogy.

The right analogy is that you ask 9379 people in a town in US
and 6867 of them liked Ginger Ale and 34986 people in UK
and 26956 liked Ginger Ale and 602 people in a town in German any 511
liked Ginger Ale - and then conclude that people like Ginger Ale.

I would conclude that people like Ginger Ale.

Arne

 

[=?ISO-8859-1?Q?Arne_Vajh=F8j?=]

When did I say I didn't understand your meaning? The ???? is because I
don't see your logic.

A few lines below.

"How so?"

In my understanding of English that means that you do
not understand what I am writing.

Now, it's clear you don't understand me. Nor, apparently your own point.

So you do not understand what I am saying, but you are sure that I do
not understand it myself ??

No, but I disagree with your assesment of the many thousands of job sites.

Ah. Then you do understand it.

It should be very obvious from the numbers that there are a huge
overlap.

Which sites?

Go into itjobwatch.co.uk and read where they have their data from.

How big? And what is the ratio for the listed jobs to all jobs actually out
there?

Much bigger than what is needed to make a good sample.

Yes, all 3 of them.

With a lot of jobs in them.

The probability of getting those results with more VB.NET jobs
than C# job in total is so close to zero that it is almost
non existing.

Wow, gee thanks. I guess anyone who disagrees with you then is stupid? You
must be a joy to be around yourself. But this mere statement is my point,
you seem pretty big on making blanket statements with no actual proof of
what you are talking about.

Well - I have provided lots of facts.

So far you have just produced some babble "I do not believe you".

Arne

 

[Scott M.]

But Tom, my friend, please see inline and remember that I'm not in a
condition that I need to relax from. I'm simply making a point (to which
you seem more receptive of than earlier.

Scott my friend... if I might offer a bit of advice, relax

If you're willing to buy a fast-food franchise, drop all colas and replace
them with ginger ale I think we might buy into your argument. The fact is
you "know" that Coke and Pepsi account for a substantial percentage of
cola sales. You also know that failing to sell these two brands is
"likely" to reduce your sales. It may not but don't ask somebody earning
money selling Coke and Pepsi to prove it. Rather invest your money and
win or lose based upon your gamble.

Yes, I do "know" that Coke/Pepsi are the largest non-alcoholic carbinated
beverages. But I know this because of the consitent results of the reliable
market research in "Nation's Restaurant News" (in another life I was a
full-service restaurant manager and then a restaurant consultant). As a
consultant, we would do local market research and look at the market
research of other reputable sources.

Having said that, I would not base my decision on which to carry based on a
sampling any smaller than 1000 for local demographics.

And I trust you also know that 3 people is not a representative sampling.

Yes, and this is my point. It seems we are in agreement after all.

What constitutes a representative sample has been defined by people who
are paid to know these things:
http://en.wikipedia.org/wiki/Sampling_(statistics)

I have two problems with your link:
1. It takes me to a page which states that no result could be found for
your search.
2. Wikipedia is hardly the definitive word on anything since no one is
paid to check accuracy unless something is challeged.
You show me ANY entry at Wiki and 10 minutes later, I'll show you
some new *facts* at the same URL. So, I would hardly use anything from Wiki
as definitive proof of anything.

If 99% of the development jobs are going to VB.Net developers but they
aren't visible in the public landscape they are by definition "invisible".
It might be that Eiffel.Net or COBOL.Net has twice as many jobs as VB.Net
in that case right? Lack of public access to the data is a problem but
what are you hoping that people will do in that case, fund studies?

Uh, yes? And, I would *guess* that you could ask several professionals who
are paid (by a neutral party), say Human Resource (national head-hunters)
personnel and aggregate their results to get a national statistic on the
matter. But that is hardly what we have been given for this 3:1 statistic.

As I said before, you may be entirely right about the stat., but the limited
research on this is simply, not acceptable for a reliable reference.

What people are saying in their posts is "as far as I can see" which may
or may not be very far but it is the best that they can do.

Great, but that's not what what people have said (AFAICS). They've listed a
few online job sites and actually then said that 3 sites polled is enough to
make a definitive statement about a national/international trend. They
haven't said "I only checked 3 out of thousands of sites, but here's what
I've found."

It would be incumbent upon the VB.Net developers to point out _not_ that
the "flawed study" has flaws (so does any VB.Net job study) but rather to
point to results which suggest otherwise when using any other accepted
criteria.

Why? I'm not the one making broad claims that are not believeable. I'm not
presenting any statisics or findings at all. I'm simply questioning what
has been presented here. That doesn't shif any burden of proof back to me,
because I've made no statements to have to prove.

I think it is incumbant upon *anyone*, regardless of what language they use
or like to provide credible proofs when presenting a supposed fact.

So instead of mentioning 3 friends, do a Google search on "soft drink
popularity" for instance. Google could have it all wrong but clearly you
see the difference between citing 3 people you happen to know and citing a
statistically neutral group. Can I prove Google messages are
statistically neutral? I doubt it but if I had limited funds and had to
decide whether to only sell ginger ale I'd use it as an indicator.

Again, this is not the point. Let me clarify (again) what I take issue
with: It's not the source, per se, it's the way the "results" are
presented. If all you had time to do was an un-scientific Google search,
fine, no problem. But don't tell me that it was a comprehensive search and
represents a true, relaible statistic. As I said before, you may be right,
but don't claim you are right and prove it with flimsy observations. Say "I
only had time to do an un-scientific Google search, but here's what it
showed.", then let the reader make any conclusions they want.

To further your argument I'll offer the following: there are probably more
poor C# developers than poor VB.Net developers. The total count isn't the
only measure and certainly isn't the primary one. Consider there may be
far more bad singers in the lines at American Idol than in the
neighborhood Burger King. There are simultaneously however some great
ones destined to make some recording company millions. So does a record
exec watch American Idol or visit Burger King stores hoping to keep his
"bad" count down?

But that is not what I'm debating. I'm talking about the way the conclusion
is "packaged".

Personally I don't know the answers,

I don't either, so I can't agree or disagree with you because I'm not
informed enough about the scenario.

I'll guess that Arne doesn't either (and that he would be willing to admit
it)

You don't have to guess. Arne has not admit it when pressed on it.

but we're stuck having to make decisions. You can choose to sell ginger
ale based upon "no reasonable studies conclude it isn't the number one
seller" and others will make the mistake of thinking "we have to sell Coke
and Pepsi or we'll go bust." In either case we can visit again in say 3
years and see what's up. Arne may have landed a management position at a
firm with 100 C# developers and you may have a 25% increase in VB.Net
students. They aren't mutually exclusive conclusions.

Yep, but I would not make the mistake of characterizing such a small pool of
information as a good source to determine a trend. Arne has pretty much
done exactly that.

Scott (relaxing)

 

[Scott M.]

Bogus analogy.

The right analogy is that you ask 9379 people in a town in US
and 6867 of them liked Ginger Ale and 34986 people in UK
and 26956 liked Ginger Ale and 602 people in a town in German any 511
liked Ginger Ale - and then conclude that people like Ginger Ale.

Fair enough. Obvously I'm exaggerating a bit here. But, I do agree with
you that you could conclude that "people like Ginger Ale". But, I don't
think you can take such a small sampling (again 3 samples) and extrapolate
that into "all people prefer Ginger Ale by a 3:1 margin". A pollster could,
because they conduct scientifc polls targeting specific demographics that do
allow them to extrapolate out to accurate results for the given area they
are polling (with a small margin of error). AFAIK, you aren't a pollster or
statistician, so I don't believe you (or I for that matter), could take such
a small sampling and extrapolate anything reliable as a result.

I would conclude that people like Ginger Ale.

But, exactly (or reliably) how many?

 

[=?ISO-8859-1?Q?Arne_Vajh=F8j?=]

Yes, and this is my point. It seems we are in agreement after all.

But we are discussing whether over 40000 is a good sample.

I have two problems with your link:
1. It takes me to a page which states that no result could be found for
your search.

If you look at the URL then it should be rather obvious that your
browser dropped the trailing ')'. And if you insert that manually the
link works.

2. Wikipedia is hardly the definitive word on anything since no one is
paid to check accuracy unless something is challeged.
You show me ANY entry at Wiki and 10 minutes later, I'll show you
some new *facts* at the same URL. So, I would hardly use anything from Wiki
as definitive proof of anything.

All tests shows that Wikipedia is about as reliable as other
lexica.

Not perfect but not bad.

As I said before, you may be entirely right about the stat., but the limited
research on this is simply, not acceptable for a reliable reference.

Hello.

This is a news group not a scientific paper.

Why? I'm not the one making broad claims that are not believeable. I'm not
presenting any statisics or findings at all. I'm simply questioning what
has been presented here. That doesn't shif any burden of proof back to me,
because I've made no statements to have to prove.

Why not ?

I told what my experience was.

You asked where I had it from.

I told you.

You did not believe it.

I picked a few samples.

They matched almost perfectly with what I said.

You still did not believe it.

Someone else picked another sample. That also matched.

You still did not believe it.

I do not want to check every job site in the world to satisfy
some moron that either does not understand statistics or think
that I fixed the selection of samples.

You should pick a few site yourself and check.

I think it is incumbant upon *anyone*, regardless of what language they use
or like to provide credible proofs when presenting a supposed fact.

Again this is a news groups - you should not expect people to be willing
to spend many hours documenting everything they post.

Yep, but I would not make the mistake of characterizing such a small pool of
information as a good source to determine a trend. Arne has pretty much
done exactly that.

How many people do they interview to make a political poll ?

How many jobs did my samples include ?

Arne

 

[Scott M.]

"How so?"

In my understanding of English that means that you do
not understand what I am writing.

No, you are mistaken here. It means what is your proof? Explain your
reasoning for making such a statement (I understood what you were saying,
just not why you said it).

So you do not understand what I am saying, but you are sure that I do not
understand it myself ??

Again, I uderstood what you said, just not why you said it.

Ah. Then you do understand it.

Yes. Thanks.

It should be very obvious from the numbers that there are a huge
overlap.

Yes, I agree. So, this helps make my point. What if 100,000 of the C# jobs
are multiply listed on all 3 sites you visited, but only 30,000 of the
VB.NET jobs are multi-listed. You said it yourself the job list is *not*
distinctive. This only helps my point that you'll have to look at a lot
more sites to thin out this margin for overlapped listings.

Go into itjobwatch.co.uk and read where they have their data from.


Much bigger than what is needed to make a good sample.

So, how much? You mean you are able to determine what the non-online posted
jobs are and how many of them there are? How did you do that?

With a lot of jobs in them.

a lot = 3:1 definitively when a single job is probably cross-posted but not
necessarially to all of these 3 particular sites and other jobs may not be
posted at these sites and yet other jobs may not be posted online and yet
other jobs are only available to in-house employees?

The probability of getting those results with more VB.NET jobs
than C# job in total is so close to zero that it is almost
non existing.

Ooh! A new *fact*, and your source for this statement is?

Well - I have provided lots of facts.

Could you summarize your *facts* in a simple list for me since I'm too
stupid to find them among all the big words you smart people use? Because
all I've seen from you is an extremely small sample of potentially flawed,
but grossly incomplete at a minimum suppositions.

So far you have just produced some babble "I do not believe you".

It's true, I don't. But, I've gone a bit further than this by providing
reasonable questions asking for substantiating the claims you made. You
reply was "You're stupid and you babble". Why don't you just say "I know
you are, but what am I?" Oh, wait, you would only say that to the person
who insulted you and I didn't do that. Now, let's see...who did? Yes, that
sounds like a reasonable argument to support your *facts*.

You know, I've said a few times here that you may be entirely correct. It's
not like I just don't want to say that C# is sought after more than VB.NET.
I'm simply asking for some credible proof of your *facts*. I haven't
insulted you and I've tried to explain my side. But you continuously
provide weak data, unsubstantiated statements and insults.

-Scott

 

[=?ISO-8859-1?Q?Arne_Vajh=F8j?=]

No, you are mistaken here. It means what is your proof? Explain your
reasoning for making such a statement (I understood what you were saying,
just not why you said it).

Experience that the same jobs get posted at multiple sites.

The fact that the numbers say that if there are thousands of such
job sites most of the jobs has to be duplicates.

Yes, I agree. So, this helps make my point. What if 100,000 of the C# jobs
are multiply listed on all 3 sites you visited, but only 30,000 of the
VB.NET jobs are multi-listed. You said it yourself the job list is *not*
distinctive. This only helps my point that you'll have to look at a lot
more sites to thin out this margin for overlapped listings.

You are being absurd.

I deliberately choose one site per country. How many dupliactes
do you think there between for US, UK and Germany ? Not many I can
tell you !

Furthermore there are no reason to believe that the rate of
duplication depends on the language.

So, how much? You mean you are able to determine what the non-online posted
jobs are and how many of them there are? How did you do that?

Again.

I see no reason why jobs adds in paper should have a different
distribution between C# and VB.NET than online.

Ooh! A new *fact*, and your source for this statement is?

Do the math yourself.

If P(C#)=0.49 and P(VB.NET)=0.51 what is the probability of
getting >=6867 C# out of a sample of 9379 (dice numbers).

Could you summarize your *facts* in a simple list for me since I'm too
stupid to find them among all the big words you smart people use?

No.

I assume that you know how to read old posts.

You know, I've said a few times here that you may be entirely correct. It's
not like I just don't want to say that C# is sought after more than VB.NET.
I'm simply asking for some credible proof of your *facts*.

This is a news group.

I am not here to provide you proof.

I provide some info.

You can decide to believe it.

You can decide to not believe it.

You can make your own check.

There was this other guy that apparently is somewhat smarter
than you that instead of writing post after post just checked
at a job site he knew. And saw that it matched pretty good.

You could do the same, if you were interested in knowing.

But you are an idiot if you expect anyone to check all
job sites in the world (and not to forget: all the non-online
jobs as well).

Arne

 

[Scott M.]

But we are discussing whether over 40000 is a good sample.

But Arne, it's not 40,000 since we are not talking about distinct jobs and
we are not talking about the many jobs (I don't know how many) that are
available, but not on the 3 of thousands of sites, nor are we talking about
jobs not posted online at all, not are we talking about jobs available only
to in-house employees. So clearly, we are not talking about a sampling that
is anywhere near a true representaion of jobs out there.

If you look at the URL then it should be rather obvious that your
browser dropped the trailing ')'. And if you insert that manually the
link works.


All tests shows that Wikipedia is about as reliable as other
lexica.

Ooh, another *fact*! Dare I ask what your source is for that statment? Did
you know that 2 weeks ago, Wikipedia's entry for the American comedian
"Sinbad" indicated he recently died (not kidding, was on the news with an
interview with the "dead" comedian). I don't remember seeing /hearing
anything in the real media, incicating his death. Why, because anyone can
change Wiki, but reputable news sources have to stake their reputation on
their reporting. Wiki's reputation relies on its accuracy, which is less
than stellar. That's not to say the main stream media always gets it right
(election night 2000), but they certainly do make more of an effort to
*fact-check* what they report.

Oh, by the way, here's some more Wiki-Facts:

Stephen Colbert (American comedian) challenged his viewers recently to go to
Wiki and change the definition of "truth" within 5 minutes of his challenge
to a particular phrase. I can't remember the exact phrase but guess what? 5
minutes later, the definition of "truth" had been changed.

Or, how about that Microsoft paid an "indepenent" consultant to alter Wiki
posted data with new *facts* disputing a recent report of the overall
benefits of .NET vs. Java?

Not perfect but not bad.


Hello.

This is a news group not a scientific paper.

Hello? You haven't gotten what I'm saying at all. Because it is a
newsgroup, don't present data as if it were from a white-paper when it's
just your own, non-scientific research. Simply say "Here's what I found at
a few sites.", don't pass 3 sites (with all the deficiencies I've previously
pointed out) as the basis for a "credible trend".

Why not ?

(See 4 lines up)

I told what my experience was.

You asked where I had it from.

I told you.

You did not believe it.

I picked a few samples.

They matched almost perfectly with what I said.

You still did not believe it.

Someone else picked another sample. That also matched.

You still did not believe it.

I do not want to check every job site in the world to satisfy
some moron that either does not understand statistics or think
that I fixed the selection of samples.

Hey, thanks for the continued insults. I never asked you to check every
site or look in any place. Nor, did I suggest you fixed any result.
Suggesting that I did, just diverts us from the topic and away from the
truth. I asked you to substantiate your *facts*, you responded with insults
and more unreliable sources.

You should pick a few site yourself and check.

Why? I'm the one saying that you can't get a reliable statistic from job
sites in the first place.

Again this is a news groups - you should not expect people to be willing
to spend many hours documenting everything they post.

I sure do when they post something as a definitive fact that is in dispute.
I'm sorry you have a lower standard for accepting information that is
unsubstantiated.

How many people do they interview to make a political poll ?

You accused my earlier of not understanding statistics. I am not a
statistician, but I do know a few things, like: It depends on what is being
polled geographically and demographically.

If you are trying to find out which candidate is in the lead for a state
senate seat, you need a sample of as little as 5% of registered voters in
specific areas that are chosen to get a accurate idea of realistic results.
You also would look at previous voter turn out in these areas as well as
median income for these areas, demographics on mean level of education,
gender, etc.

If you poll too many (an amount over 25%) believe it or not, the accuracy of
the poll decreases. But the key here is to know what to look at, you can't
just poll *any* 5% - 25% of any of the people. It must be 5% - 25% of the
geographically and demographically chosen pool.

By asking your question, in such a simple way (as to make it seem that it is
only the number of people polled), it tells me that I know a lot more about
statistics than you. Perhaps, that is why you don't/can't/won't understand
my point.

How many jobs did my samples include ?

As I've just said, and have been saying, it's not the amount that matters,
it's the source(s) that matter.3 randomly selected sites with overlapping
results that doesn't include all the other sources (non-online) of data on
this, does not come close to getting an accurate sample.

But, that's fine, just simply say "Here's what I found.", don't say "This is
enough for a credible trend to be surmized."

That's all I'm saying. And, I'm doing it without insulting you and using
real facts (like how a poll is really conducted) and sticking to my point
(instead of diverting attention from the topic at hand to the purpose of a
newsgroup).

 

[Scott M.]

Experience that the same jobs get posted at multiple sites.

I agree. Now, how does this help you make your point? Because it really
seems to bolster my point that you are seeing the same jobs more than once.
Which, to use a very good analogy, is like voting more than once and trying
to get an accurate count of how many people voted. It's also like believing
a web site hit counter that uses Page_Load to determine when to increase the
count, rather than Session_Start.

The fact that the numbers say that if there are thousands of such
job sites most of the jobs has to be duplicates.

Again thanks for making my point. How does this improve the accuracy of
your *resulting facts* then?

You are being absurd.

What?! How so? Do you know anyting about polling and statistical sampling.
What you call absurd is the basis for statistical analysis used by
professional statisticians. Your well thought proof to disprove me is: "You
are being absurd"?

Would you be willing to explain what you find absurd in the point that you
need a much larger sample than 40000 when dealing with many thousands of
cross-posted (duplicate) jobs? Because, I'd really like to hear you
rationally dispute that with any *facts* you have on it.

I deliberately choose one site per country. How many dupliactes
do you think there between for US, UK and Germany ? Not many I can
tell you !

You are just digging your hole deeper. If you "can tell" me, then please
do, how many? And, of course, you'll need to tell me where your answer came
from, as I'm having a hard time keeping track of all the *facts* you have
presented.

Furthermore there are no reason to believe that the rate of
duplication depends on the language.

True, but you have provided no *facts* to dispute the possibility, and
that's all I proposed, a possibility, I did not present that as a *fact*.

Again.

Can you explain what "Again" is supposed to equate to in regards to my
question above?

I see no reason why jobs adds in paper should have a different
distribution between C# and VB.NET than online.

Why not? WHAT ARE YOUR FACTS THAT MAKE YOU FEEL THAT WAY?

Do the math yourself.

If P(C#)=0.49 and P(VB.NET)=0.51 what is the probability of
getting >=6867 C# out of a sample of 9379 (dice numbers).

I'm not sure how you prove your statement, by starting with *If* followed by
numbers that you have yet to substantiate.

If a die has nothing but one's on all six sides, then the probability of
rolling anyting other than a one is zero. A true statement based on a made
up scenario.

No.

Because, time and again, you haven't presented any facts, only opinions. As
you say, a newsgroup is a great forum for ideas, opinions and facts. But you
are presenting your ideas and opinions as Gospel without any credible facts
to substantiate yourself. In fact, now you have begun to present more
unsubstantiated conclusions as the *facts* that support your first
unsubstantiated *facts* (see dice scenario above).

I assume that you know how to read old posts.


This is a news group.

I am not here to provide you proof.

I provide some info.

You can decide to believe it.

You can decide to not believe it.

You can make your own check.

There was this other guy that apparently is somewhat smarter
than you that instead of writing post after post just checked
at a job site he knew. And saw that it matched pretty good.

So two studies of bad data that match make a proof?

You could do the same, if you were interested in knowing.

But you are an idiot if you expect anyone to check all
job sites in the world (and not to forget: all the non-online
jobs as well).

Good thing I haven't asked anyone to do that then, huh? But, thanks for the
newest insult.

Arne, all this boils down to is don't package something as a fact, when it's
not. Call a spade, a spade and don't choke on your garbage just because you
don't want to admit I have a point. It's not about which language is used
more, it's about how you came to that conclusion and just being honest about
that.

 

[Scott M.]

Think about how a scientist would provide a supposed *fact*. They tend to
like to be able to prove (using scientific methods) the suppositions they
make. They do this by doing painstaking research, followed by experiment
and then write a paper that is then peer-reviewed.

Lawyers and detective (who are supposed to deal in *facts* ) need to be able
to *prove* the suppositions they present as *facts*. They do this with
logical and rational thinking and try to find undisputable *facts* to
support thier suppositions. Oh yeah, they don't insult the jury when the
jury doesn't believe thier *facts* due to insufficient evidence.

Now, I know what we're talking about is far less important than these
professions, but my point is there is a proven method for supporting a
staement of supposed *fact* and you haven't come close to it.

By the way, statistical analysis is a scientific profession and the bar for
providing a reliable statistic and proving its reliability isn't anywhere
near what you have provided. We are talking about polling here after all.

 

[=?ISO-8859-1?Q?Arne_Vajh=F8j?=]

Ooh, another *fact*! Dare I ask what your source is for that statment?

Since you apparently either are not capable of or to lazy to Google:

http://www.wired.com/culture/lifestyle/news/2005/12/69844

Did
you know that 2 weeks ago, Wikipedia's entry for the American comedian
"Sinbad" indicated he recently died (not kidding, was on the news with an
interview with the "dead" comedian). I don't remember seeing /hearing
anything in the real media, incicating his death. Why, because anyone can
change Wiki, but reputable news sources have to stake their reputation on
their reporting. Wiki's reputation relies on its accuracy, which is less
than stellar. That's not to say the main stream media always gets it right
(election night 2000), but they certainly do make more of an effort to
*fact-check* what they report.

Oh, by the way, here's some more Wiki-Facts:

Stephen Colbert (American comedian) challenged his viewers recently to go to
Wiki and change the definition of "truth" within 5 minutes of his challenge
to a particular phrase. I can't remember the exact phrase but guess what? 5
minutes later, the definition of "truth" had been changed.

There are such cases.

That does not necessarily make Wikipedia less reliable than many
other sources.

A few find it funny to put in incorrect fact, but the millions
of reviewers usually get it fixes rather quickly.

Or, how about that Microsoft paid an "indepenent" consultant to alter Wiki
posted data with new *facts* disputing a recent report of the overall
benefits of .NET vs. Java?

I think you may be referring to the OOXML-ODF thing and not .NET-Java.

Hello? You haven't gotten what I'm saying at all. Because it is a
newsgroup, don't present data as if it were from a white-paper when it's
just your own, non-scientific research.

It should be rather obvious from the post that it was my research.

And it should be rather obvious too that it was based on samples.

Simply say "Here's what I found at
a few sites.", don't pass 3 sites (with all the deficiencies I've previously
pointed out) as the basis for a "credible trend".

I choose my way to post.

If you do not like it, then use a killfile.

Hey, thanks for the continued insults. I never asked you to check every
site or look in any place. Nor, did I suggest you fixed any result.

Unless the sample selection was fixed, then the samples are more
than statistically significant.

You accused my earlier of not understanding statistics. I am not a
statistician, but I do know a few things, like: It depends on what is being
polled geographically and demographically.

If you are trying to find out which candidate is in the lead for a state
senate seat, you need a sample of as little as 5% of registered voters in
specific areas that are chosen to get a accurate idea of realistic results.
You also would look at previous voter turn out in these areas as well as
median income for these areas, demographics on mean level of education,
gender, etc.

If you poll too many (an amount over 25%) believe it or not, the accuracy of
the poll decreases. But the key here is to know what to look at, you can't
just poll *any* 5% - 25% of any of the people. It must be 5% - 25% of the
geographically and demographically chosen pool.

By asking your question, in such a simple way (as to make it seem that it is
only the number of people polled), it tells me that I know a lot more about
statistics than you. Perhaps, that is why you don't/can't/won't understand
my point.

Hm.

Who was it that about 20 lines above criticized the poll due to its
size ?

You like the scientific approach: how do you score scientifically
if you first claim that a sample is too small and when pointed
out that it is very big compared to most sample then change
argument and now claim that it is skewed ?

Arne

 

[=?ISO-8859-1?Q?Arne_Vajh=F8j?=]

I agree. Now, how does this help you make your point? Because it really
seems to bolster my point that you are seeing the same jobs more than once.
Which, to use a very good analogy, is like voting more than once and trying
to get an accurate count of how many people voted.

It indicates that the 3 sites has a much higher share than you
indicate when you talk about 3 out of thousands.

Because my samples was chosen to have minimal overlap due to being
in different countries.

Again thanks for making my point. How does this improve the accuracy of
your *resulting facts* then?

It reduces the total number of jobs the samples should be compared to.

What?! How so? Do you know anyting about polling and statistical sampling.
What you call absurd is the basis for statistical analysis used by
professional statisticians. Your well thought proof to disprove me is: "You
are being absurd"?

Would you be willing to explain what you find absurd in the point that you
need a much larger sample than 40000 when dealing with many thousands of
cross-posted (duplicate) jobs? Because, I'd really like to hear you
rationally dispute that with any *facts* you have on it.

Not much point.

In another post you changed your mind and stated that it was not the
count but the sampling.

You are just digging your hole deeper. If you "can tell" me, then please
do, how many? And, of course, you'll need to tell me where your answer came
from, as I'm having a hard time keeping track of all the *facts* you have
presented.

You are welcome to believe that it is a common case that jobs are
posted in both US, UK and Germany.

But it should be obvious that it is not the case.

True, but you have provided no *facts* to dispute the possibility, and
that's all I proposed, a possibility, I did not present that as a *fact*.

Almost anything is possible.

I just go for what is likely.

Can you explain what "Again" is supposed to equate to in regards to my
question above?

I wrote about things being independent and again below.

Why not? WHAT ARE YOUR FACTS THAT MAKE YOU FEEL THAT WAY?

It does not make any sense that there should be a correlation
between media and and language.

I'm not sure how you prove your statement, by starting with *If* followed by
numbers that you have yet to substantiate.

You can try it with any probability distribution where P(C#)>P(VB.NET).

I just choose the one most negative to me.

If a die has nothing but one's on all six sides, then the probability of
rolling anyting other than a one is zero. A true statement based on a made
up scenario.

Yes. But not much relevance.

Because, time and again, you haven't presented any facts, only opinions.

I think that is a lie.

I gave you numbers.

You just don't think the numbers are representative, but that does
not change that they are fact.

As
you say, a newsgroup is a great forum for ideas, opinions and facts. But you
are presenting your ideas and opinions as Gospel without any credible facts
to substantiate yourself. In fact, now you have begun to present more
unsubstantiated conclusions as the *facts* that support your first
unsubstantiated *facts* (see dice scenario above).

You apparently did not understand the probability challenge.

Or did not want to, because it makes many of your previous
posts look rather silly.

Arne

 

[Tom Leylan]

But Tom, my friend, please see inline and remember that I'm not in a
condition that I need to relax from. I'm simply making a point (to which
you seem more receptive of than earlier.

I think you must have me confused with somebody else The reason I hang
around the VB.Net group answering VB.Net questions isn't because I don't use
it...

Yes, I do "know" that Coke/Pepsi are the largest non-alcoholic carbinated
beverages. But I know this because of the consitent results of the
reliable market research in "Nation's Restaurant News" (in another life I
was a full-service restaurant manager and then a restaurant consultant).
As a consultant, we would do local market research and look at the market
research of other reputable sources.

Which other reputable sources? What makes them reputable? Just because you
post the words "reputable" doesn't make the reputable. See how easy it is?

I have two problems with your link:
1. It takes me to a page which states that no result could be found
for your search.
2. Wikipedia is hardly the definitive word on anything since no one is
paid to check accuracy unless something is challeged.
You show me ANY entry at Wiki and 10 minutes later, I'll show you
some new *facts* at the same URL. So, I would hardly use anything from
Wiki as definitive proof of anything.

I cut and pasted the link, perhaps the conversion to ASCII introduced a
problem. Here it is again:
http://en.wikipedia.org/wiki/Sampling_(statistics) .

Your opinion on the accuracy of Wikipedia seems to be an unsubstantiated
opinion as formal studies have concluded the opposite. Follow the links
(these should work) and you'll notice that studies using reputable samplings
have demonstrated it is nearly as accurate as the Encylopedia Britannica.
You don't accept a search of 3, 4 or 5 large job sites about job statistics
but guessing is acceptable when it comes to Wikipedia. Note who claims the
test is flawed in the BBC report, the Encylopedia Britannica, who would have
guessed?

http://news.bbc.co.uk/1/hi/technology/4530930.stm
http://news.bbc.co.uk/2/hi/technology/4840340.stm

http://arstechnica.com/news.ars/post/20061127-8296.html

Uh, yes? And, I would *guess* that you could ask several professionals
who are paid (by a neutral party), say Human Resource (national
head-hunters) personnel and aggregate their results to get a national
statistic on the matter. But that is hardly what we have been given for
this 3:1 statistic.

Do you seriously think anybody posting here is going to fund a study? Would
you if I claimed your opinion on Wikipedia was unfounded? I'm going to
guess not. Can you post a link to Human Resource? I found HR.com but that
doesn't look like the one you're talking about. Perhaps somebody near one
of their offices will phone and post the facts. Something tells me you'll
dismiss the numbers with "sure one phone call but what about the other
offices" if things don't go your way.

Great, but that's not what what people have said (AFAICS). They've listed
a few online job sites and actually then said that 3 sites polled is
enough to make a definitive statement about a national/international
trend. They haven't said "I only checked 3 out of thousands of sites, but
here's what I've found."

Yes it is. The messages started up much like your opinion about Wikipedia.
When challenged it evolved to various ad-hoc queries which brought us to the
present "as far as I can tell" status. Unlike your Wikipedia opinion which
remains in the "I just made this up" category.

If you are trying to suggest that the 1000 job sites all have unique
listings it won't sell. If you are suggesting that a job board with 35 jobs
on it (the newest one posted 90 days ago) should be considered the equal of
Monster or Dice the CEO's of those companies would like to chat with you.
And it wouldn't be hard to imagine that "VBJobs.com" might have more VB jobs
posted.

The posting as I recall consisted of the results of a search on a few large
sites. It wasn't suggested that there were no VB jobs, we know that isn't
true. You're in essence trying to imply that posting "Coke sells 1.3
billion beverages a day" is akin to writing "nobody likes ginger ale."

Why? I'm not the one making broad claims that are not believeable. I'm
not presenting any statisics or findings at all. I'm simply questioning
what has been presented here. That doesn't shif any burden of proof back
to me, because I've made no statements to have to prove.

You do if you care. You may as well chat about the the faked moon landing,
Bigfoot and the Bermuda triangle. The contention that none of these things
is true should easily be countered by your challenge that we have no proof
they are fake. We can site 12 scientists and you can post "what about the
10,000 other BigFoot sites"? Another successful challenge. If you "care"
then you should offer some counter fact, if you're just debating what are
you hoping to accomplish?

Again let me point out that nobody is stating that "you ought to learn C#
because there are no VB.Net jobs" because "when 3 large job sites were
searched here are the results". Let me ask you a question? If it is easy,
like a phone call to Human Resources or a search of one of the many other
job sites why haven't you simply posted your results? You have to admit you
could have done that in the time it took to write your response.

So why? Perhaps in part because it looks like you're just playing the
devil's advocate. People who aren't tend to post "well I checked the
following 2 sites and they are 3 to 1 in favor of VB.Net jobs." Which would
lead credence to the theory that the original numbers are flawed... what we
have now is your guess that the trend won't hold.

I think it is incumbant upon *anyone*, regardless of what language they
use or like to provide credible proofs when presenting a supposed fact.

You're in the minority

Again, this is not the point. Let me clarify (again) what I take issue
with: It's not the source, per se, it's the way the "results" are
presented. If all you had time to do was an un-scientific Google search,
fine, no problem. But don't tell me that it was a comprehensive search
and represents a true, relaible statistic. As I said before, you may be
right, but don't claim you are right and prove it with flimsy
observations. Say "I only had time to do an un-scientific Google search,
but here's what it showed.", then let the reader make any conclusions they
want.

That's what happened.

But that is not what I'm debating. I'm talking about the way the
conclusion is "packaged".

Packaging is what it is about. Take for instance your remark about
Wikipedia. Seriously were you trying to dismiss it as a source of
information or were you trying to enlighten readers about potential
distortions? Had the quote come from the Encyclopedia Britannica would you
have dismissed it as readily without having read any of the studies?

You were packaging your answer for maximum impact. I said "Wikipedia" you
countered with "unbelievable source" despite what experts have said.
Incredibly the article was about what constitutes a statistically relevant
sample hardly a controversial subject. Perhaps I can point you to their
definition of PI and you could suggest it's all bogus and made up by people
with an ax to grind

Yep, but I would not make the mistake of characterizing such a small pool
of information as a good source to determine a trend. Arne has pretty
much done exactly that.

The trend is "3 sites checked 3 sites confirming". If 2 more large sites
are checked and the results are the same then it would be 5 for 5. The
trend would be for every large site checked the numbers remained the same.
That of course doesn't mean there isn't a large site with all the VB.Net
jobs but that could remain undiscovered even if 2 dozen large sites were
checked.

As people have more time perhaps they will check some sites and post the
results here as well.

Tom

 

[=?ISO-8859-1?Q?Arne_Vajh=F8j?=]

Fair enough. Obvously I'm exaggerating a bit here. But, I do agree with
you that you could conclude that "people like Ginger Ale". But, I don't
think you can take such a small sampling (again 3 samples) and extrapolate
that into "all people prefer Ginger Ale by a 3:1 margin". A pollster could,
because they conduct scientifc polls targeting specific demographics that do
allow them to extrapolate out to accurate results for the given area they
are polling (with a small margin of error). AFAIK, you aren't a pollster or
statistician, so I don't believe you (or I for that matter), could take such
a small sampling and extrapolate anything reliable as a result.

I did target specific demographics. Did you forget ?

But, exactly (or reliably) how many?

3:1

Arne

 

[Scott M.]

Ooh, another *fact*! Dare I ask what your source is for that statment?

Since you apparently either are not capable of or to lazy to Google:

http://www.wired.com/culture/lifestyle/news/2005/12/69844

Ok, nice substantiation. That point conceeded. And thanks for the nice
insults right off the bat. I had thought that I repeatedly indicated that I
wasn't going to dig up proof for your statements, since I have made no
claims. But, you seem to be incapable of getting that. You are really good
at slinging mud though.

There are such cases.

That does not necessarily make Wikipedia less reliable than many
other sources.

A few find it funny to put in incorrect fact, but the millions
of reviewers usually get it fixes rather quickly.


I think you may be referring to the OOXML-ODF thing and not .NET-Java.


It should be rather obvious from the post that it was my research.

How so? Your words exactly were that 3 site, do a trend make. You never
said, "this is what I found, take it for what it is".

And it should be rather obvious too that it was based on samples.

As stated and run into the ground by now, bad samples that do not prove
anything statically.

I choose my way to post.

You sure do.

If you do not like it, then use a killfile.

No need. I will always post a rebuttle when more info. is needed. If you
don't like it, don't reply.

Unless the sample selection was fixed, then the samples are more
than statistically significant.

You aren't even reading my replies anymore. That statement is completely
false. Now, you are saying that any data (however incomplete) is
statistacally significant? Ok, I once ate toast with butter and the
following day, I developed a cold. So, I think it is clear that toast with
butter causes colds.

Hm.

Who was it that about 20 lines above criticized the poll due to its
size ?

That is but one component of a poll. The above 3 paragraphs try to indicate
to you that there is much more to consider, but you just don't want to
accept anyting outside of your tunnel vision. I told you that your 40000
number was not really 40000 because of the multi posting of jobs and that
just looking at 3 sites out of a targeted demographic that could include
thousands of sites, makes your 40000 not an accurate number and that you'll
need a much larger and divers pool of data to work with. What exactly is
there in that you don't get?

You like the scientific approach: how do you score scientifically
if you first claim that a sample is too small and when pointed
out that it is very big compared to most sample then change
argument and now claim that it is skewed ?

Well, first off, I guess I'd respond by saying that I never said anything of
the sort, so the rest of your comments are irrelevant?

Where, exactly, did I say it was very big compared to most samples? I said
no such thing. And, when you attempted to I corrected you. I'm saying it's
skewed BECAUSE it's too small.

 

[Cor Ligthert [MVP]]

Scott,

Do you think that this is the right newsgroup for this discussion?

Cor

 

[Scott M.]

I think you must have me confused with somebody else The reason I
hang around the VB.Net group answering VB.Net questions isn't because I
don't use it...

Not sure what you are referring to here.

Which other reputable sources?

Here are some of the common sources:

Local and US Chambers of Commerce.
Local and US restaurant & hospitality associations
Trade publications (ie. Nation's Restaurant News).
Hire a market research company to perform statistical analysis.

What makes them reputable?

They are recognized (industry-wide) as reputable and reliable and they are
usually the only ones that do the kind of research required.

Just because you post the words "reputable" doesn't make the reputable.
See how easy it is?

Well, I've answered your questions, so I would say that if you have the
facts to back up a statement, it is pretty easy. All I've gotten from Arne
is opinion stated as fact with no simple and clear answers to the questions
I've provided.

I cut and pasted the link, perhaps the conversion to ASCII introduced a
problem. Here it is again:
http://en.wikipedia.org/wiki/Sampling_(statistics) .

Your opinion on the accuracy of Wikipedia seems to be an unsubstantiated
opinion as formal studies have concluded the opposite. Follow the links
(these should work) and you'll notice that studies using reputable
samplings have demonstrated it is nearly as accurate as the Encylopedia
Britannica. You don't accept a search of 3, 4 or 5 large job sites about
job statistics but guessing is acceptable when it comes to Wikipedia.
Note who claims the test is flawed in the BBC report, the Encylopedia
Britannica, who would have guessed?

http://news.bbc.co.uk/1/hi/technology/4530930.stm
http://news.bbc.co.uk/2/hi/technology/4840340.stm

http://arstechnica.com/news.ars/post/20061127-8296.html

This is the only thing Aren has provided as proof. I accept it and I
conceed that point.

Do you seriously think anybody posting here is going to fund a study?

No, but (for the last time as apparently no one wants to read my simple
point), don't presents facts as definitive, it they aren't. I never said
someone should fund a study for a newsgroup post, that was not the question,
nor was that my reply. Many studies are funded every day, by people that
want to know a reliable statistcial result.

Would you if I claimed your opinion on Wikipedia was unfounded? I'm going
to guess not.

You guessed wrong.

Can you post a link to Human Resource? I found HR.com but that doesn't
look like the one you're talking about.

I wasn't talking about any web site. I was talking about HR firms, such as
head-hunters. My point was they are not considered in any of the "research"
provided by anyone, which make the research incomplete and its data suspect
at best.

Perhaps somebody near one of their offices will phone and post the facts.
Something tells me you'll dismiss the numbers with "sure one phone call
but what about the other offices" if things don't go your way.

Nope, I'd say that until we got the other sources of data as well
(newspaper, etc.) and enough of each to provide a good sampling of data.

Yes it is. The messages started up much like your opinion about
Wikipedia. When challenged it evolved to various ad-hoc queries which
brought us to the present "as far as I can tell" status. Unlike your
Wikipedia opinion which remains in the "I just made this up" category.

Not really. Directly after Arne posted his ill-gotten statistic, I
challenge it and his response was that looking at 3 sites out of thousands
and not at any other non-web site data was enough to justify his statistic
as correct. That's what I've been debating ever since.

If you are trying to suggest that the 1000 job sites all have unique
listings it won't sell. If you are suggesting that a job board with 35
jobs on it (the newest one posted 90 days ago) should be considered the
equal of Monster or Dice the CEO's of those companies would like to chat
with you. And it wouldn't be hard to imagine that "VBJobs.com" might have
more VB jobs posted.

When did I suggest anything of the sort? If anything, I've said the exact
opposite of that.

The posting as I recall consisted of the results of a search on a few
large sites. It wasn't suggested that there were no VB jobs, we know that
isn't true. You're in essence trying to imply that posting "Coke sells
1.3 billion beverages a day" is akin to writing "nobody likes ginger ale."

Again, that's not what I said in any way, shape or form.

You do if you care. You may as well chat about the the faked moon
landing, Bigfoot and the Bermuda triangle. The contention that none of
these things is true should easily be countered by your challenge that we
have no proof they are fake. We can site 12 scientists and you can post
"what about the 10,000 other BigFoot sites"? Another successful
challenge. If you "care" then you should offer some counter fact, if
you're just debating what are you hoping to accomplish?

Because I'm not debating the statements being made and I've made no
statement of fact on the whole language question. I'm questioning the
methods of the data gathering. If you say something, the burden of proof is
on you to be able to back it up. I felt I backed up my Wiki statement, but
you have proved my wrong and I accept that. No one, though, challenged by
on the Wiki statement to go and prove it. If they did, I would/should be
the one to do the research, not you.

Again let me point out that nobody is stating that "you ought to learn C#
because there are no VB.Net jobs" because "when 3 large job sites were
searched here are the results". Let me ask you a question? If it is
easy, like a phone call to Human Resources or a search of one of the many
other job sites why haven't you simply posted your results? You have to
admit you could have done that in the time it took to write your response.

Well, first it's not as easy as you imply I said (which I didn't). If
anything I've said that it is much harder that a quick phone call and a
quick Google search. But again, I'm not making the claims here, so the
burden of proof isn't mine.

So why? Perhaps in part because it looks like you're just playing the
devil's advocate. People who aren't tend to post "well I checked the
following 2 sites and they are 3 to 1 in favor of VB.Net jobs." Which
would lead credence to the theory that the original numbers are flawed...
what we have now is your guess that the trend won't hold.

No, again you are mis-reading (or not reading) what I've said. What we have
is my doubtfullness of the supposition based on the limited data sources
used and in the non-scientific (and incorrect) statistical analysis made.

You're in the minority

Huh? You're saying that depending on what side of the issue someone falls,
only one group is responsible for providing credible proofs to back up their
statements? You can't be saying that, that's ridiculous.

That's what happened.


Packaging is what it is about. Take for instance your remark about
Wikipedia. Seriously were you trying to dismiss it as a source of
information or were you trying to enlighten readers about potential
distortions? Had the quote come from the Encyclopedia Britannica would
you have dismissed it as readily without having read any of the studies?

I've already conceeded I was wrong about the particular Wiki topics that
study looked at. And your continued suggestion that I will only believe the
*facts* I like is unwarrented. Arne has provided no facts to support his
3:1. That's all I'm looking for. I've said over and over that he may be
right, just show me how you support your conclusion and do it with an
accepted method for coming up with a valid statistic in this matter. He
never considered other sites (so he does not have enough web site data to
form a reliable sample and he has not used any other source of data (which
leaves out huge data pools).

You were packaging your answer for maximum impact. I said "Wikipedia" you
countered with "unbelievable source" despite what experts have said.
Incredibly the article was about what constitutes a statistically relevant
sample hardly a controversial subject. Perhaps I can point you to their
definition of PI and you could suggest it's all bogus and made up by
people with an ax to grind

Ok, so you've got me on Wiki. I think we can let that one go and address
the many other points I've made. But, again, you are making suggestions
about how I would react to hypothetical situations and, quite frankly, I've
given you no reason to think I'm biased about what a proper study would
find.

By the way, does the article on Wiki say that determining a statistically
relevant sample "hardly a controversial subject"? I don't think it does. In
fact, it has quite a bit to say about going about getting a good sample
(http://en.wikipedia.org/wiki/Sampling_method). I'm afraid Arne hasn't come
close to following this methodology.

The trend is "3 sites checked 3 sites confirming". If 2 more large sites
are checked and the results are the same then it would be 5 for 5.

No, it would be 5 for 5 sites, not including all the other sources in the
"sample frame". And, as pointed out on your Wiki, if you don't have a
relevant sample frame, your analysis is not relevant.

The trend would be for every large site checked the numbers remained the
same. That of course doesn't mean there isn't a large site with all the
VB.Net jobs but that could remain undiscovered even if 2 dozen large sites
were checked.

You are making this way more compicated than it needs to be. A 3:1
statistic was offered as a valid value for C# to VB.NET jobs currently
available. An insufficient sample was used to get this and the presenter of
this information will not conceed this point nor will he change the way in
which his result was presented. That's the bottom line.

Show me a result from a good sample or change the way you package your
result. Until you do that, your stats. are meaningless. That's all I'm
saying.

As people have more time perhaps they will check some sites and post the
results here as well.

Yes. And maybe they'll look not just at sites, since if that's all they
did, we'd still not have a good sample.

 

[Cor Ligthert [MVP]]

Arne,

I did not follow the whole thread , are you proofing that VB.Net is
easier to learn than C# or that more people are doing it, and therefore jobs
can easier be fulfilled?

Cor

 

[Scott M.]

I did target specific demographics. Did you forget ?

How could I remember something that didn't happen? You looked at a limited
sample from 3 locations (that's geographic, not demographic). What were the
demographics you looked at?

3:1

I wouldn't have expected you to say anyting else!

 

[Scott M.]

I agree. Now, how does this help you make your point? Because it really

It indicates that the 3 sites has a much higher share than you
indicate when you talk about 3 out of thousands.

Check your math Arne, it does the exact opposite of that. If we have
multiple listings for the same job, then we have an even smaller pool of
distinct jobs, meaning we now need to look in other (more) places to get the
pool size up since we don't want to count the same job twice.

Because my samples was chosen to have minimal overlap due to being
in different countries.

Perhaps, but even still, they do not represent all the places where jobs can
be found: not nearly all the sites and none of the non-online places.
Either way, your sample is too small.

It reduces the total number of jobs the samples should be compared to.

Correct! Meaning we don't have a large enough sample.

That's what I've been saying all along. We do need a much larger sample
that includes more than just one source for the data. Good thing you
finally agree with me (now, I can go to bed!)

Not much point.

In another post you changed your mind and stated that it was not the
count but the sampling.

No, I never said that. I've consistently said your count is low AND your
sources limited to online data only.

So, which is it? Are there lots of duplicates (which you just agreed there
were, here in this post) or are their not a lot of duplicates? Either way,
for the reasons I've been repeating over and over, you still don't have a
good sample.

You are welcome to believe that it is a common case that jobs are
posted in both US, UK and Germany.

Do you get that it's irrelvant. You can stop dwelling on duplicates or not
(since you can't seem to make up your mind on it). Either way, for the
reasons I've been repeating over and over, you still don't have a good
sample.

But it should be obvious that it is not the case.


Almost anything is possible.

But, I don't walk around stating these things as *facts*.

I just go for what is likely.

THANK YOU, you admit your 3:1 is not a fact (and I say it's unrealistic to
expect that it is accurate given the flaws in your sample data).

I wrote about things being independent and again below.


It does not make any sense that there should be a correlation
between media and and language.

But you don't *know* that there is or isn't, do you? As we've discussed in
newsgroups plenty of times, there are those that feel that you should look
for experience C# pros. differently and in different places than VB pros.
I've heard people say VB pros. are a dime a dozen. If that is actually the
case, you may not have to look very far or hard or the same way you look for
C# pros. This is my *impression* of one of the differences in recruiting
for one language over another. So, yes, it is possible that you might find
differences in the media type being looked at.

You can try it with any probability distribution where P(C#)>P(VB.NET).

Sure, but you are manufacturing an equation to suit your supposition. I
gave you my dice supposition, but it hardly makes it reliable.

I just choose the one most negative to me.


Yes. But not much relevance.

About the same as your manufactured probabilty in your manufactured
if...then.

I think that is a lie.

But you won't be clear as to why.

I gave you numbers.

Made up (your probability scenario above) or insufficient ones (flawed
sample group), yes.

You just don't think the numbers are representative, but that does
not change that they are fact.

Well, that about sums it up. I can't ague with a brick wall.

You apparently did not understand the probability challenge.

Or did not want to, because it makes many of your previous
posts look rather silly.

Or, it was an irrellevant side topic that has no relationship to how you got
your sample data and every supposition you made after based upon it.
Ever heard: "Garbage In...Garbage Out"?

I'm done with this Arne, it's clear I'm wasting keystrokes here.