BuzzerBeater Forums

I've changed PR managers a lot over the seasons and every time I increase the PR level I get higher merchandise that week, and everytime I lower my PR manager level my merchandise drops. So that indicates to me that merchandise is definitely affected by PR managers. I don't think there is a way to prove if it is directly affecting or indirectly affecting.

Any numbers I provide aren't going to be helpful bceause I constantly change my roster.
For example I hired a level 4 PR manager this week (I had a level 1 before this), and my merchandise increased by 3k. However, I also bought 4.5 million dollars in players this week, so that could also account for the merchandise rise.

Your strategy puzzles me. When you were tanking your games you had a level 1 PR and now that your planning on winning games with those new players you go and upgrade your PR to level 4. I was always taught to do the exact opposite.

Get a good PR when you expect to lose a lot of games to offset the decrease in arena attendance.
Get a level 1 PR if you are expecting a 22-0 season since your arena will be packed to the rafters anyways.

Well I upgraded the PR manager for a few reasons.
1. I wanted the Specialty to make games easier to win.
2. My arena attendance is low at the moment due to tanking so I'm hoping that winning tonights game plus winning the cup plus my PR manager at level 4 will help me get close to selling out.
Perhaps I should have kept my PR manager at level 4 the whole season, but I figured that the 20k I would save on salary each week is roughly the same as the amount the PR manager would increase my arena attendance by. So I didn't see a point in it.

OK, I dug up some data here to explain my observation last season. Keep in mind that this isn't an attempt to prove anything, and is merely to explain my observation and why I suspect that there could be an additional impact to merchandising unrelated to fan surveys and tied to the PR Manager. There are a number of other known and unknown possibilities I have considered, and this is merely one of them. Here goes...

In my case, 2 weeks had passed since my last player purchase on 7/1, and I was seeing consistent rises in MR for many weeks prior in Division V, and then continuing after my promote to DIV up until my staff event. I don't have records going back that far for merchandising, but I know for a fact that I had gradual/constant increases to MR prior to this point.

I downgraded from a level 3 to a level 2 PR on Sunday 7/18. I saw my first drop to MR on 7/19 the following day (the financial update). I don't have proof or a record of the fan surveys, but I can tell you they were also the same on this week as they were the prior week. I had equal numbers of wins/losses from week to week, etc.

Even if I was incorrect in my recollection of fan survey data, my fan surveys only went through the roof after. I promoted the previous season, went 6 rounds deep into the cup tournament, won TV game and began to really win games and beat higher ranked teams. Despite all the improvements week to week and my fan surveys hitting 4.5-5 balls, I continued to experience MR drops. These drops spanned from 7/26 - 9/13 with an average decrease of $345.00/wk.

My thought process here, is if fan surveys are supposed to improve my MR, why did great increases only result in continued drops. When I had worse fan surveys at the beginning of the season I had increases, not decreases. Anyway, on 9/19 I made my next staff change, and this was the first time I saw any change in MR trends. Following this, my data breaks up even further because of too many additional variables being introduced. I don't think this information is enough to draw any conclusions, but all I'm saying is I think this was a good enough observation to question what I really know about PR managers.

Public Relations Managers help market the team to the fans

Market the team meaning also selling jerseys and such, so merchandising. I'm pretty sure having a star player from your country is a big merchandising boost, NT selection is also a huge boost.

As for the hypothesis, this is not a theory I'm testing, it's just a vague idea so there is no need to formulate an accurate hypothesis... Basically it's an input-output analysis with only a few variables (15 and collected during 12 weeks so 180 elements). This is a modelling exercise with not enough variables so it's bound to be innacurate and hard to analyse. And I understand your point (I think), you're saying that an experiment must test a theory, and for this test to be efficient you have to formulate correctly your hypothesis in order to collect only the right data. Sure thing but I'm not exactly experimenting I'm trying to understand a system and find the typical response to a known input (PRM level).
So yes the data collected won't be enough, and some variables may be applied while it shouldn't, nonetheless it should provide some answers.
Imagine that with a lvl 4 PRM manager my merchandise revenue does not increase significantly (let's say less than + 20%) while my roster stays the same and my results are constant then we'll know that "merchandising-wise" it's not profitable to acquire a lvl 4 PRM.

Market the team meaning also selling jerseys and such, so merchandising.

I can't disprove this, but I think it is a stretch to say that this marketing directly translates into merchandising revenues.

And I understand your point (I think), you're saying that an experiment must test a theory, and for this test to be efficient you have to formulate correctly your hypothesis in order to collect only the right data

Yes, thank you. That is what I was meaning to say, except a good experiment can also test a question or validate previous results. It seems to me you may also be asking your question in a certain way and looking for a particular result. I recommend reading about confirmation bias effects on information processing.

This is a modelling exercise with not enough variables so it's bound to be innacurate and hard to analyse.

So yes the data collected won't be enough, and some variables may be applied while it shouldn't, nonetheless it should provide some answers.

So you expect this to be an inaccurate and difficult exercise with limited data and question your own methods with regard to applying variables, but anticipate reliable results?

"On two occasions I have been asked,—'Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?' ... I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question."
—Charles Babbage

Imagine that with a lvl 4 PRM manager my merchandise revenue does not increase significantly (let's say less than + 20%) while my roster stays the same and my results are constant then we'll know that "merchandising-wise" it's not profitable to acquire a lvl 4 PRM.

I think this may lead you to suspect something as I have. Although I noticed a trend in my own observations I'm not going to jump to a conclusion that there are no other possible solutions. I'm not going to assume that inaccuracies in my own data couldn't possibly cloak something important. Perhaps the results are random, and you just happened to get the same number several times in a row? This often happens with random number generation algorithms. There are many other explanations I can think of even based on either a positive or negative result from this exercise. I'm not saying this because I am trying to deter you from investigating or destruct your efforts. I do feel that we need reliable information on the subject.


Last edited by bonespawn at 10/26/2010 11:58:33 AM

So you expect this to be an inaccurate and difficult exercise with limited data and question your own methods with regard to applying variables, but anticipate reliable results?

Well I'm just pointing out the obvious, I mean the only information we have on merchandising is merchandising revenue... However if we're looking for a trend, as you said, it may corroborate yours, or not. My main objective is collecting data and trying to get something out of it. If I find something that seems meaningless, well I'll just say it is indeed meaningless and I won't try to arrange the results in order to prove something. If I find a strong tendency (like the +20% I mentioned earlier) well maybe it becomes necessary to investigate more and to conduct a real experiment (like asking to several GMs to record specific data, the specific part has yet to be determined). I believe the preliminary "random" data collecting part is usefull because if nothing dramatic comes out of it, there's no need to spend time and money on a thorough experiment.

I believe the preliminary "random" data collecting part is usefull because if nothing dramatic comes out of it, there's no need to spend time and money on a thorough experiment.

Ah, but it was one of your most notable countrymen, René Descartes, who supported advancements in deductive reasoning and discarded perception as unreliable. The problem of induction is a philosophical argument that has been going on since ancient times. I'm pretty sure my side is winning, but I don't think we will end this debate tonight on buzzerbeater