Comments
3 Approaches to Justifying Analytics Results
View Comments: Newest First | Oldest First | Threaded View
Page 1 / 3   >   >>
Re: Foundationalism
  • 1/22/2014 12:33:20 PM
NO RATINGS

SRS1,

That is correct. In fact all three views have flaws. As foundationalism seeks to serve as a better option than an infinite chain or an infinite loop, it does demand that one make an asssertion assumed to be true.  For example, we used to assume that the sun revolved around the earth, that the earth was flat and that ulcers were caused by only stress.  Today we believe that the earth moves around the sun, that the earth is round and bacteria contribute to ulcers. Foundationalism creates a starting point that is assumed to be true.

The age-old questions of 'how can we know anything' and 'how can we prove anything' are tough nuts to crack.  Hence as analysts, we may not have the perfect answer, but we should be prepared to defend what we believe - flawed as our evidence may be.

It could very well be that the smartest analyst is not the one who knows anything but the one who can logically and ethically defend what s/he has found.

 

Foundationalism
  • 1/21/2014 6:59:25 PM
NO RATINGS

But isn't foundationalism not acceptable because its allowing an arbitrary reason at the base. Meaning that there's a reason for which there are no further reasons making it even slightly better to accept than any of its contraries?

Re: Coherent... to who?
  • 1/14/2014 2:49:38 PM
NO RATINGS

OK Bryan, I see where you're coming from now. We often talk about the balance between gut decision-making and data-based decision-making. Perhaps the need for the gut instinct is what comes into play when the results are "believable" if not precisely right!

Re: Coherent... to who?
  • 1/14/2014 12:11:37 PM
NO RATINGS

I tossed in believable because I was thinking about those honest moments when the results are inconclusive and you have to present plausible explanations.  There are times when you see causal relationships that do not make immediate sense. For example, we noticed that our church web site hits went up between Christmas and the first 3 weeks of January.  We typically were in post holiday break mode so it did make immediate sense why there was increased activity (Easter and Christmas - yes, post-Christmas and January - no).  What we figured out (but could not prove statistically) was that people hit the reset buttons on there lives during this period (new diets, gym memberships and other resolutions).  Our programming runs on a academic calendar (nothing new during the summer or holiday periods).  The explanation that we became comfortable with is that folks were thinking about starting the new year by going to church. Could not prove it, but it was the most believable explanation for the web site hit pattern. We have since started emphasizing themes of 'a new you for a new year' in marketing existing programs during this period for those seeking church engagement.

This may not be the best example, but there are those moments in analytics when the stakeholders do not have answers - good guesses - but not clear cut answers.  I totally agree with you that we should not tempt folks to steer the interpretation of the results set.  But there are those times when plausible becomes a proxy for provable. We don't talk about those moments because 'we really don't know for sure' is not a response that keeps the money tree shaking.  Not promoting fudging, just laying out the harsh realities of the business.  Probably plays a legitimate part in sustaining the analytics mantra 'more research is needed'.

Re: Coherent... to who?
  • 1/14/2014 9:45:11 AM
NO RATINGS

I like leaving it at "the results must be right." To suggest that they only have to be believable seems to open up the door to fudging and the other nonsense that might be too tempting not to use for those who have a bias about the data and the decision it's intended to inform.

Re: Coherent... to who?
  • 1/14/2014 9:41:35 AM
NO RATINGS

Beth, glad to hear that they owned up to the error.  Mistakes are like tumors - the longer you don't acknowledge them, the worse they become.  Almost goes back to the basic question of this post "When doing research and analysis, how do you know that you are right, why should anyone believe you and what is the scope of damage to others if you were wrong?"  We may get fast results with pretty output, but at the the end of the day, the results have to be right or at least believable. 

Re: Coherent... to who?
  • 1/13/2014 9:42:19 PM
NO RATINGS

Coincidentally, I received a "hey, we were wrong" email just a few hours ago, asking for a retraction of survey results released late last week -- and this was from a company that prides itself on its advanced analytics capabilities. I can't help but wonder how it was that it released wrong info in the first place -- and whose head was rolling as a result.

Re: Coherent... to who?
  • 1/13/2014 9:33:17 PM
NO RATINGS

Hi Bryan, I think this all sounds very reasonable. Unfortunately, all it takes is one unreasonable, perhaps insecure, manager to throw it all to the wind!

Re: Coherent... to who?
  • 1/13/2014 3:48:53 PM
NO RATINGS
1 saves

Broasdway,

It all depends - private and public universities have different funding streams and continuing development incentives. For example, I once worked at Johns Hopkisn University (late 1980s to early 1990s).  That is a private university. Incoming assistant professors would get their salaries carried for 2 years. After that, if they could not generate their salaries and the overhead expenses via external funding then they were out of the door - period. After that, they had a certain number of years to make associate or their spot was at risk. But the bottom line was to get tenure there and to get a salary, there was an expectation to bring in your own money.  Survival of the fittest - maybe if you rose to the top of the ranks, you might get an endowed chair. But even then, there was an expectation that you bring money into the house. Being dependent on private funding makes non-revenue producers dead weight.

Public universities run off of tax dollars, so the pressure to be a cash cow is not as bad. And yes, perfomance can slip when you do not have to grind out the grant proposals.

So I would agree with you but only for public universities. In that light, private universities are like the private sector - 'It does not matter if you are cheetah or an antelope, when the sun rises, you better start running". Public entities tend to have more mercy but private entities are very competitive and tend to destroy the weak.

So even in academia, based on the funding source, the pressure to produce can skew research interests toward organizations that will fund similar interests. It is not just the search for truth via analytics, it is also the search for the mortgage money.

Re: Coherent... to who?
  • 1/13/2014 3:01:04 PM
NO RATINGS

In academia, once you get your tenure, which is relatively early in a faculty member's career, performance doesn't matter anymore, right? Now as a staff member, at a well-run educational institute, results do drive promotion and pay.

Page 1 / 3   >   >>


INFORMATION RESOURCES
ANALYTICS IN ACTION
CARTERTOONS
VIEW ALL +
QUICK POLL
VIEW ALL +