BuzzerBeater Forums

This discussion has come up before, only last time it started to turn into a philosophical debate over whether or not induction or deduction should be used to build the theory. For the record, I'm against the use of inductive observations like those suggested here, and I think LA-Wolph just touched on one of several flaws. Yes, you are also leaving out probably a number of other important pieces of data. A quick example would be the number of domestic vs. foreign players. Large data samples wouldn't just help or increase precision. They avoid the additional assumption that effects are not gradual. Good luck with building the house of cards.

Last edited by bonespawn at 11/13/2010 3:41:55 PM

That's the thing, with a true random sample, those wouldn't matter. You would need a control group for those who did not change PR managers, but that's not too difficult.

The reason for introducing random sampling is to reduce the likelihood of bias. In the case of disease research, you are looking for anything that will result in changing of condition, and when you find it you may have something. PR effects, attendance levels, merchandising and such are not just on/off states. They are values that fluctuate with or without changes to PR manager staff levels and are influenced by other factors. You are making an apples to oranges comparison.

Last edited by bonespawn at 11/13/2010 10:03:33 PM

In order for that type of research to work the sample most be pulled from a homogeneous population, you grab a group, split it in two or more and subject the different groups to a change to observe the effect of the change.

Not really. It would defeat the whole purpose in having a double-blind study if you manipulated variables in your control group.

It works in a large volume to look at the overall effect, but all of the oddballs also need to be looked out and accounted for, for example-

I'll skip quoting your verbose analogies and commentary, and I think you went a little overboard in trying to make a a very simple point (that anomalous data can result in confirmation bias). Point taken.

There are many different approaches to experimentation in medicine. Some are quantitative and others are qualitative. A descriptive study for example, does not deal with the intentional manipulation of variables at all. I'm not going to debate the usefulness of a particular method of experimentation or how to structure it, because I think that is already jumping ahead. To start with, the basic assumption has been made here that inductive reasoning is more useful or appropriate.

As the saying goes, "If you only have a hammer"... PR manager's aren't a neurological disease. They are a bloody algorithm in a computer program with a finite number of parameters. One of the biggest reasons why deduction is commonly used in math is because there is a small number of unknowns. In medical science and many other fields you are dealing with a large or seemingly infinite number of unknowns where inductive observations are more useful.

I gavea long list of factors which impact attendance. Unfortunatley there are a lot of them. So it is getting similar to trying to put a pulse on an actual organism.

In the same way folded paper is similar to an F-22 Raptor, sure.

I am not against trying to figure it out anyway, but since as you point out it is a finite mathematical problem, we might as well as identify and account for the variables.

Might as well pick out what animals we are going to test on. That is also very important

I think you guys scared the OP off.

Or bored him to death.

I know full well that it is an easy study.
If I get this information from 200 managers, I'll have a pretty good idea.

My intention isn't to find out everything perfectly, but the most accurate way possible.

As you say, there are many influencing factors, but how more data ask fewer teams are going to send me your details and prefer to gather data from many teams that are in short supply, instead of receiving data from a few teams.

To conduct a survey, so there is minimal relative error that meet the most people possible.

It is better to interview 2,000,000 people more or less than 200,000 respondents who interviewed very well.

My intention is to find the% of a public relations from one level to another.

Develop my study as follows:
1. I ask many USER data.
2. Together all data and keep it in my working folder.
3. Do a study on the book and I can publish the study and make a hypothesis about the outcome.
[For example, I get this data:
·100 teams to increase the technical staff have been increased by 2%
·40 teams have been increased by 3%
·20 teams have been increased by 1%
·40 teams have been down by 2%].
4. Find out if this hypothesis is true.
5. I ask again, data teams are interviewed to see if the same profits or losses previously.
6.Finally if my hypothesis is confirmed, the study ended.
7. If my hypothesis isn't confirmed, I will repeat some steps.

In the other study, which is to find the% of more and less to lose / gain in the pavilion as the global rankings'm studying computers that have not changed their public relations.

In This thread:
(163800.1)