Survey on Newsgroup Behavior

[Jens Arndt]

Hello everyone,

my name is Jens Arndt, and I am writing a Master's Thesis about
drivers of behavior in newsgroups. That's why I need your help. The
comp.lang.c newsgroup is one of the few that fulfill the predefined
requirements. Therefore, I would appreciate it if you could take 10 –
15 minutes to fill out my survey. You'll find it under the following
link:

http://wi1.bwl.uni-mannheim.de/phpesp/public/survey.php?name=complangc

The pages have been tested to be compatible with all recent browsers
I'm aware of. In case that you experience any problems please don't
hesitate to contact me (via (e-mail address removed)-mannheim.de or by
replying to this post).

To a large extend the survey asks for your personal opinions, so don't
think too much about it, be spontaneous. Also, as you'll see the
survey does not ask for any identification such as your e-mail
address. Consequently, all collected data is treated absolutely
anonymous and confidential.

Furthermore, once the survey has been successfully completed, you'll
find the results as a working paper under the following link:

http://www.bwl.uni-mannheim.de/wifo1/ger/p/4_6.php3?lst=wifo1

Thanks in advance for helping me with your participation.

Jens Arndt

 

[Dan Pop]

Hello everyone,

my name is Jens Arndt, and I am writing a Master's Thesis about
drivers of behavior in newsgroups. That's why I need your help. The
comp.lang.c newsgroup is one of the few that fulfill the predefined
requirements. Therefore, I would appreciate it if you could take 10 –
15 minutes to fill out my survey. You'll find it under the following
link:

http://wi1.bwl.uni-mannheim.de/phpesp/public/survey.php?name=complangc

Badly designed and badly implemented.

It is not possible to have a look at the whole survey before deciding
whether you want to participate or not. You can't see the next page
before completely answering the questions on the current one. Of course,
one can work around it (as I did), by giving bogus/random answers, but
this is not an elegant solution. Also, there should be no mandatory
questions on such a survey.

Some of the questions are next to impossible to answer, even
approximatively, for a newsgroup regular:

1. How many messages do you post on average during one week?

2. How many hours do you spent in the newsgroup on average during one
week?

4. Discussions in newsgroups are organized in threads. If you
participate in a discussion, how many messages are typically contained
in the thread you contribute to?

6. With how many fellow participants do you interact through
comp.lang.c during an average week?

Others cannot be answered by any of the options provided by the survey:

I feel confident about the other participants' skills and knowledge.

There are all kinds of people: people I trust, people I don't trust and
people I don't know well enough to know whether to trust or not. How
should I rate this statement?

The following are highly pointless:

I achieve monetary returns through participating in comp.lang.c.

Never noticed anyone (seriously) oferring advice for money.

Promises that are given to me by fellow members are usually kept.

In 12 years, I have received exactly one promise. It was kept
(thanks again, Larry Weiss).

Don't consider this list as exhaustive, it contains only things that
immediately caught my eye upon a very superficial reading.

Dan

 

[Randy Howard]

The following are highly pointless:

I achieve monetary returns through participating in comp.lang.c.

Never noticed anyone (seriously) oferring advice for money.

Agree with the majority of your comments. You could consider an
alternative impression of this one though. People may learn a great
deal through their participation in c.l.c, which could be manifested
as better job opportunities, or developing commercial software using
improved methods, etc. Again, nebulous, but at least feasible that
it might be happening.

 

[Richard Tobin]

Some of the questions are next to impossible to answer, even
approximatively, for a newsgroup regular:

1. How many messages do you post on average during one week?

[etc]

In any case, what is the point of asking for subjective answers to
questions whose answers can be objectively determined by analysis of
posting archives?

-- Richard

 

[Dave Vandervies]

Agree with the majority of your comments. You could consider an
alternative impression of this one though. People may learn a great
deal through their participation in c.l.c, which could be manifested
as better job opportunities, or developing commercial software using
improved methods, etc. Again, nebulous, but at least feasible that
it might be happening.

It seems to me that "through participating in..." implies a more direct
connection than that.


dave

 

[Dik T. Winter]

>
> Badly designed and badly implemented.
>
> It is not possible to have a look at the whole survey before deciding
> whether you want to participate or not.

Indeed. That is why I do not participate, seeing that I can indeed not
answer most questions on page 1 (I have not even looked beyond that).

 

[Mark Piffer]

Some of the questions are next to impossible to answer, even
approximatively, for a newsgroup regular:

1. How many messages do you post on average during one week?

[etc]

In any case, what is the point of asking for subjective answers to
questions whose answers can be objectively determined by analysis of
posting archives?

-- Richard

[OT]
He asked for drivers of behaviour, so it is reasonable trying to find
out one's subjective opinion about engagement within this group. This
may reveal deviations from the measurable facts, i.e. you are
perceiving your "life" in c.l.c in a very different way than what an
objective analysis would tell.

Mark

PS: Dan, instead of wasting your time bashing a harmless surveyor you
could bash me on my opinions about the macro in the "Re: GetByte(int
x,int n)" tread. I still didn't get a definitve slap in the fa...err
answer.
[/OT]

 

[Jens Arndt]

Okay,

I had to decide whether I
(a) ask you for your e-mail or names or any identification, so that I
could look up all your data, or
(b) not to do it (make the survey anonymous), with the implication
that I could not access the posting archives.

I decided for (b). Don't ask me if it was the right decision.

Jens

Some of the questions are next to impossible to answer, even
approximatively, for a newsgroup regular:

1. How many messages do you post on average during one week?

[etc]

In any case, what is the point of asking for subjective answers to
questions whose answers can be objectively determined by analysis of
posting archives?

-- Richard

 

[Jens Arndt]

Badly designed and badly implemented.

It is not possible to have a look at the whole survey before deciding
whether you want to participate or not. You can't see the next page
before completely answering the questions on the current one. Of course,
one can work around it (as I did), by giving bogus/random answers, but
this is not an elegant solution. Also, there should be no mandatory
questions on such a survey.

Some of the questions are next to impossible to answer, even
approximatively, for a newsgroup regular:

1. How many messages do you post on average during one week?

2. How many hours do you spent in the newsgroup on average during one
week?

4. Discussions in newsgroups are organized in threads. If you
participate in a discussion, how many messages are typically contained
in the thread you contribute to?

6. With how many fellow participants do you interact through
comp.lang.c during an average week?

I agree that most people probably don't have a ready-made answer to
these questions. However, I had to ask one way or another. If you have
any ideas how else I could ask these questions, expecting a better
answer please let me know.

Others cannot be answered by any of the options provided by the survey:

I feel confident about the other participants' skills and knowledge.

There are all kinds of people: people I trust, people I don't trust and
people I don't know well enough to know whether to trust or not. How
should I rate this statement?

The questions asks you to rate your feelings. Don't start making your
own list, just say what you feel.

The following are highly pointless:

I achieve monetary returns through participating in comp.lang.c.

Never noticed anyone (seriously) oferring advice for money.

Then just say 'strongly disagree'. Pretty simple.

Promises that are given to me by fellow members are usually kept.

In 12 years, I have received exactly one promise. It was kept
(thanks again, Larry Weiss).

Don't consider this list as exhaustive, it contains only things that
immediately caught my eye upon a very superficial reading.

Dan

I know that most if not all surveys do not perfectly fit for all the
respondents, but that's one of the inherent weaknesses of survey
research. If I could, I would have scheduled an hour-long interview
with all of you, but you'll agree that that's not possible.

So I decided to do a survey research, and that's what I did. I do not
raise the claim to be an expert in the field, so any critique is taken
serious. But please let me know how to do it better.

Jens

 

[Jens Arndt]

It seems to me that "through participating in..." implies a more direct
connection than that.

The survey askes for your feeling and/or opinions. If you do not
believe that you achieve monetary returns from participating just
klick on 'strongly disagree'.

Jens

 

[Tim Rentsch]

So I decided to do a survey research, and that's what I did. I do not
raise the claim to be an expert in the field, so any critique is taken
serious. But please let me know how to do it better.

Make the survey pages so that all questions can be read before any are
answered. (I gave up after seeing that completing page 1 was a
prerequisite for page 2.)

Either make questions optional, or allow "don't know" or "no answer"
as an answer.

Finite choice questions should have finite choice selector boxes.
If the set of choices is large, the selector boxes could provide
appropriate binning, eg, "less than 1 per month", "less than 1
per week", "1 to 3 per week", "4 to 10 per week", "more than 10
per week".

Last but not least, do a little statistics gathering in preparation.
I would think it wouldn't be too hard to get information about, say,
the last six months of postings; tabulate that information, broken
down by person posting (that is, the email address used), and make
that available in a prelude page before getting to the survey. Don't
make everyone taking the survey do the mechanical work that you can do
once and very likely at higher quality (for those questions that have
observable answers based on publically maintained history).

 

[Arthur J. O'Dwyer]

Hear hear!

Isn't this a good thing, scientifically speaking? The surveyor wants
to avoid self-selection in his sample population as much as possible.
He isn't willing to go whole hog like the professional pollsters and
contact random individuals, but at least he can make sure nobody's going
to look at the survey and then decide, "You know what? I don't really
want to do this after all."

Definitely. Mandatory questions are counterproductive, because it means
there's no way to enter an unambiguous "don't know," "N/A," or "do not
wish to provide an answer." So the data for every single mandatory
question is polluted by a certain amount of noise, generally weighted
toward the ridiculous (e.g., someone might enter "Very low" instead of
"WTH is 'interactivity'?" for question 5).

Right. And many of the questions are actually purely statistical;
for example, numbers 1, 4, 5, 6,

The questions asks you to rate your feelings. Don't start making your
own list, just say what you feel.

Please don't start becoming an idiot on us. (Perhaps I should say,
don't /finish/ becoming an idiot --- with this survey, you're well on
the path to idiocy already!) That question is like asking, "I trust
other people to tell me the truth." Obviously, it's true for some
people and some situations, and false for others. No "rate your feelings"
involved.
Have you stopped beating your wife yet? (Agree/Neutral/Disagree)
And then some questions betray a deep-seated misunderstanding of the
whole purpose of technical newsgroups, such as this one from page 7:

(Rate truth of the statement from 1 to 7)
I would never move to a different newsgroup, even if it
would be similar to comp.lang.c.

Usenet isn't a chat room. I doubt anybody here reads only this newsgroup
and no others. (Again, Google will tell you for sure whether that's the
case, and you won't need to ask anybody.)

So I decided to do a survey research, and that's what I did. I do not
raise the claim to be an expert in the field, so any critique is taken
serious. But please let me know how to do it better.

Remove all the dumb questions; define your terms before using them;
and then answer all the objective numerical questions yourself, by
consulting the newsgroup archive at Google Groups. You will learn a
lot more about newsgroup behavior from observation (via the archive)
than you ever will from a bunch of dumb multiple-choice questions on
a web poll. And it will be more objective, too. Small margins of
error are good. Repeatability is good. Web polls have neither;
archive statistics have both.

HTH,
-Arthur

 

[Daniel Haude]

On Sat, 30 Oct 2004 13:07:16 -0400 (EDT),

Have you stopped beating your wife yet? (Agree/Neutral/Disagree)

Thanks for making my day!

--Daniel

 

[Nils Petter Vaskinn]

comp.lang.c newsgroup is one of the few that fulfill the predefined
requirements.

Your statistics will be flawed.

1. You will be able to say nothing about the drivers of behavior in
newsgroups. You will only be able to say something about drivers of
behavior for members of those specific groups.

2. There may be a connection between the willingness to participate in
online surveys and different behaviors in newsgroups. The self selevtion
ruins the value of the statistics.

--
Nils Petter Vaskinn (n + surname at broadpark dot no)

"They can sense totalitarianism approaching from a distance,
as animals can sense an approaching thunderstorm."
Paul Graham, about Hackers (http://paulgraham.com/gba.html)

 

[Jens Arndt]

1. I agree that the survey is not perfect, but I guess that there's no
perfect theory out there. Sure, I'll not be able to say anything about
all the newsgroups that are out there, but don't you think that
analyzing one newsgroup can contribute to the overall understanding of
newsgroup behavior in general?

2. There definitely is a bias introduced through the self-selection of
participation. But since I try to figure out what makes friendly,
constructive members participate in the way they do, I believe that
this bias does not ruin the whole study.

Jens

 

[Jens Arndt]

First of all I want to thank everyone who filled out the survey so
far. I guess I had to expect that not everyone is happy about this
thing, but I nevertheless appreciate everyone who participated.

For those who did not yet particiapte, I hope that you'll find the
time to fill out the survey. I would really be thankful about it.

For those who decided not to participate, I'm sorry that I could not
convince you to help me with this thing, and I hope that I didn't
steal too much of your time.

Thank you.

Jens

 

[Dan Pop]

Agree with the majority of your comments. You could consider an
alternative impression of this one though. People may learn a great
deal through their participation in c.l.c, which could be manifested
as better job opportunities, or developing commercial software using
improved methods, etc. Again, nebulous, but at least feasible that
it might be happening.

Nope, there are many other questions dealing exactly with this kind of
indirect monetary returns one could achieve from participation. This one
can be only interpreted as dealing with direct monetary returns.

Dan

 

[Dan Pop]

I agree that most people probably don't have a ready-made answer to
these questions. However, I had to ask one way or another. If you have
any ideas how else I could ask these questions, expecting a better
answer please let me know.

Let the responder choose between a few ranges of values, rather than try
to evaluate a single value. Don't ask irrelevant and hard to answer
questions, like 4 and 6 above (could you answer them, WRT your favourite
newsgroup?).

Keep in mind that answering a survey must be an entertaining experience
for the responder. If it's too difficult or too boring or the questions
don't make much sense to them, people will simply give up, way before
reaching the end.

The questions asks you to rate your feelings. Don't start making your
own list, just say what you feel.

What I feel about such a question is that it is highly idiotic, for
reasons already explained. Until you add this particularly rating to
your survey, there is no way I can reply to it.

Then just say 'strongly disagree'. Pretty simple.

You're missing the point: why include questions which are bound to get
answered identically by all the participants?

I know that most if not all surveys do not perfectly fit for all the
respondents, but that's one of the inherent weaknesses of survey
research. If I could, I would have scheduled an hour-long interview
with all of you, but you'll agree that that's not possible.

Why would any of us be interested in participating to such an interview?

So I decided to do a survey research, and that's what I did. I do not
raise the claim to be an expert in the field, so any critique is taken
serious. But please let me know how to do it better.

Make it fun for the participant, rather than boring or stressing him.
Be your first guinea pig: it is not ready for production until you find
it fun yourself. Then, try it on a few of your friends and see if they
like it, too. Carefully consider their objections. Because you can't
offer anything in return to the participants, this is the *only* way to
have a reasonable number of people taking part to your survey. Believe
it or not, the average clc-er is not a masochist.

Dan

 

[Mike Wahler]

Hello everyone,

my name is Jens Arndt, and I am writing a Master's Thesis about
drivers of behavior in newsgroups.

In <[email protected]>

(e-mail address removed)-mannheim.de (Jens Arndt)

Make it fun for the participant, rather than boring or stressing him.
Be your first guinea pig: it is not ready for production until you find
it fun yourself. Then, try it on a few of your friends and see if they
like it, too. Carefully consider their objections. Because you can't
offer anything in return to the participants, this is the *only* way to
have a reasonable number of people taking part to your survey. Believe
it or not, the average clc-er is not a masochist.

In addition to Dan's valid points, I'll add that I find
it difficult to believe that a university would award
a Master's Degree for such "research". About the only
conclusions I think could be drawn from it are that "everyone
is different", and "everyone has an opinion". Who doesn't
know that already?

-Mike