I dont disagree that he's better than Houghton. I just dont agree with those that think Berros performances have been at the standard required for whatever reason.

I'd say expected rather than required.

He's been playing in a shiite team with a poor coach, no gameplan, playing in just about every position on the field in a side decimated by injuries, having injections just to get on the pitch. It's hardly surprising really

I have made the point on here before that one of the reasons Swain made so many tackles, both here and in the NRL, was that he was targeted due to his style of tackling. Like Houghton, he usually went low and often got caught under the carrier. He differs from Houghton in that this happened less often because of superior technique.



I did say I wasn't giving a complete picture and your point is valid. Teams will try to tire him as it's his energy and explosiveness out of dummy half that is his strength. Although you only have to look at the number of times teams scoot after Houghton has made the tackle and ended up under the ruck to know that he is targeted because of a weakness also.

You should go and explain all that stuff to agar

Last person that tried that was Jonny Sharp.

and he got a response!!!

But to understand what they mean you have to understand what they are measuring. And to understand what they are measuring you need to know how they are collected and what definitions are being worked to. I would be surprised if you'd attended board meetings at which no interest was paid to the collection of the statistics you are reporting and what it is they are measuring.



By writing something along the lines of what I wrote about Houghton being targeted in defence, something that uses specific, game-related actions to show that Berrigan is weak in specific areas or not contributing all that he could. Is Berrigan being caught on the wrong side of the ruck and being scooted against? Is he missing opportunities to get out from dummy half when there are markers not set and a broken line to run against? Is he making the wrong decision when we have a line set, with good numbers and choosing to go the wrong way against the orders of his halves? Is he failing to get involved in tackles that he could get involved in to slow the play because he's struggling for fitness so choosing just to retire the 10m instead?

That sort of stuff, rather than giving me a list of numbers with which I can't do anything. And that's not because the numbers are incorrect or measuring the wrong thing. It's because we simply don't know and as such I can't debate them. I can ask questions about their validity, but there won't be any answers forthcoming because the answers aren't known.



Fine with me. I only mentioned it because you mentioned your affiliation with statistics earlier.





From my point of view and in my experience (sorry to bring that up again), the methodology is at least as important as the interpretation. Just because the methodology is flawed (if it is) doesn't mean the flaw is consistent across all records. For example, what if the methodology for collecting the statistics for tackles made was heavily skewed in favour of counting a third man in flop as a tackle, but not counting a big hit that left the carrier on the floor to be finished off by another player as a tackle for the big hitter? And, for example (purely hypothetically), what if Fitzgibbon made a lot of tackles in which he was the third man in and relatively few big initial impacts, but Tickle made a lot of big initial hits yet not many 'finishers'? The statistics would be skewed heavily in Fitzgibbon's favour. Now that's an extremely simple example and involves cases that probably won't happen often enough to have a huge effect on the overall picture, but the point is that we just don't know. And as such, I really don't think using the statistics provided have any validity because they can't be consistent across all scenarios. The variable, subjective nature of the game means that measuring can never be fully consistent and fair to all.

Collection of stats and what they are measuring are two discreet things. (input and output as ive mentioned in the follow up post)

The executive board of organisations dont have the time to spend questioning the methodology of research. They employ people to get it right. What they are measuring and the context of the data is what the exec board need to know to make their exec decisions. They dont want to know for example how sales data is captured, just what it's telling us and how we can use that data to forecast end of year profit etc.

I agree fully that getting valid data is important but i focus more on trusting data and driving action rather than spending time questionning the validity. (maybe wrong in this instance)

I don't see that. To collect them you have to know what it is they are measuring, otherwise how do you know you are collecting the right thing?



But that still means they know what it is that was collected in the first place. They'll know it was sales data, they'll know what it was sales of, they'll know where it was collected, they'll know when it was collected, etc. If not, you've been involved with some very naive (or maybe just extremely trusting, to the point that they're staking the whole company on an underling doing their job correctly) board execs.



I just can't trust data unless I know exactly how it was collected or I personally know the collector to be entirely reliable and fastidious.

My point probably not very well explained is that the metric itself and the methodology are defined at input stage. The interpretation of the findings from a metric perspective are at output once theyve seen what the measure is saying including commentary for context and rationale etc. You dont always know youre measuring the right thing until the output stage.



They'll know that its sales data yes, and what it sales of, both are outputs. They wont care about where it was collected and they wont care about when it was collected. It will roll up into a monthly or quarterly report that they receive from a trusted source. They aren't naive at all, they simply pay some of the most respected people in the business to take care of the input stuff for them. One of the key strengths a senior manager can have is to not need to know the detail to allow them to focus on what matters. Thats why most senior managers try to surround themselves with people that complement their weaknesses or they ensure the organisational hierarchy is designed to ensure the right quality data is created at the right time.



I guess the 2nd part of this about reliable and trustworthy is satisfied in the example above for the company i work for. The exec dont personally know the people but i guess the 6 figure salaries been paid to the people pulling the MI and reports together means they come with a certain amount of respect and credibility.



My interpretation (going back to the very original point) of the superleague stats was that someone is spending a lot of time to collate a lot of data and i have absolutely no evidence to disregard it or dis-trust it. Secondary data like that has to just be trusted sometimes. (which is the stance from experience that i tend to take unless the data doesnt feel like it makes sense)

Its fairly easy to spot where data is inaccurate. In masses of data like on the SL site there will be tell tales signs which immediately can make it questionnable.

I think any differences of opinion in the above are almost entirely cosmetic and due to labelling discrepancies more than anything else. (That's my way of saying there's not much to disagree with there.)



Here's where there's a difference.



I don't distrust it based on not trusting the gatherers. I can't entirely trust it either, but I'm sure whoever is responsible is reliable at collecting what they are attempting to collect. My issue is with what they are attempting to collect, because, unlike sales data, they aren't collecting something that has an absolute, unequivocal definition. As I mentioned earlier with the Fitzgibbon/Tickle/tackling hypothetical, consistency of collation isn't always enough when there are as many variables present as there are in a game of RL. It's not that the data is inaccurate, it's that it has to inflexible to attempt to remain consistent and that just doesn't work, as far as I'm concerned, when there are so many variable, unmeasurable, inconsistent factors involved every minute.

Again, a quick example, assuming the statistics collected are entirely consistent yet remain fair to all. Say tomorrow Berrigan was attempting to scoot against a Leeds pack consisting of Leuluai, Diskin, Peacock, JJB, Ablett and Sinfield. And he's working of plays from Dowes and O'Meley. 20 minutes later, Houghton comes on and is working against a Leeds pack of Bailey, Buderus, Burgess, JJB, Lauiti'iti, Sinfield off plays from Cusack and Moa. Houghton probably gets a quick play against a poorer defence. Or maybe the starting Leeds front row was having a torrid time of it and the replacements are doing a much better job. Who knows? Those collecting the statistics certainly don't. And it's that type of factor (of which there are an innumerable amount every game) that isn't taken into account.

There are just too many uncontrollable variables that can't be factored in for me to be able to regard the statistics provided with anything more than disdain.

Fully get the level of doubt about the variables. The problem is in every single instance practically there are always going to be variables that are uncontollable and often unknown. (or overly complex to feed into any kind of statistical analysis easily)

We tend to go with working assumptions based on what we know. My working assumption about the superleague stats is that they are consistently collated for ease of use. As long as the data is collated consistently it gives a level of comparability even if the data is slightly incorrect. The debate i had on here a couple of years ago was that the figures may be 2 or 3 wrong, but as long as that variance exists in all data collated similarly then the comparability exists which is where the real value comes in this instance.

I did my Masters dissertation on the Return on Investment for a Human Resources function. Until you sit down and map out all of the possible inputs and outputs to something like you wouldnt believe the size of that kind of task. (especially with HR generally been non-statistical and often intangible)

What i needed to do to ensure that the task could actually be completed was to make some working assumptions that may mean the overall result are slightly inaccurate but as long as they have a statistical significance thats relatively high sometimes thats good enough.



(i know i said i wasnt going to mention education, it just slipped out)