Holy Purolator Flip Flop!

[ZeeOSix]

Originally Posted By: Wilhelm_D
Originally Posted By: ZeeOSix
Do you really think only BITOG members who cut open Purolator filters magically get only the filters that have media tears in them?

I think I know the difference between a statistically valid sample and some anecdotes.


Well simple logic and the odds that only the bad Purolator oil filters fall into the hands of BITOG members who cut open oil filters is basically impossible.

 

[440Magnum]

Originally Posted By: ZeeOSix
Originally Posted By: Wilhelm_D
Originally Posted By: dnewton3
Honestly, I think Puro is beginning to show some trending for failures here.

Purolator makes and sells millions of oil filters under a variety of names.

A half dozen or so non-random samples on BITOG can't demonstrate a trend in that large a universe.

It does seem to illustrate the lemming effect in special interest discussion groups. It sort of reminds me of the vilification of Mobil some years ago based on one sample of one weight of one model of their Mobil 1 motor oil.

I recommend chill pills all around.


Basically, what you are saying is that you think of a very small percentage of bad Purolator filters in the world, it just so happens that most of them have been bought and used by BITOG members who just happen to cut them open. And the rest of the world who bought Purolator filters got all the good ones. Do you know the odds of that?

No, that's not what he's saying. He's saying that our BITOG sample size is WAY too small to be statistically valid. The way probability theory works, when you take too small a sample (in this case, BITOGers who've opened filters) of a very, very large population (all the Purolator filters produced), you can get results that would on the surface imply that something sudden happened, when in fact the exact same number of filters may actually be failing as were failing 2 years ago.


And conversely- two years ago our small sample size may have been UNDER-reporting the actual number that were failing! It works both ways.

I still think its a possible warning sign of a problem, don't get me wrong. But the field of probability is very, very counter-intuitive. Our brains are wired to think selfishly, not probabilistically- we weight the things that affect us directly or that we can see and touch much more heavily than we should when we subconsciously "calculate" whether we think a trend is happening or not.

 

[ZeeOSix]

Originally Posted By: ZeeOSix
I look at it like this. Of the 1 million Purolator oil filters made per year, lets say BITOG members buy 0.002% (2,000 filter), and of those members only 20% of them cut open the filters for inspection. That's 400 Purolator filters cut open per year. Of those cut open, 20 per year show failures. Now you have a sample failure rate of 20/400 = 5%. That's a huge failure rate, not one any company would want, or any customer would want to use.


Forgot to add, if you only looked at the models of Purolator filters with metal crimped media seams, you'd probably see the failure rate estimate be much higher.

As I've said before, I think this media tearing issue seems to be only on the filter models with the metal crimped seam. There is something unique about those filters that is associated with the root cause of this problem.

 

[Hollow]

Originally Posted By: ZeeOSix
Originally Posted By: Wilhelm_D
Originally Posted By: ZeeOSix
Do you really think only BITOG members who cut open Purolator filters magically get only the filters that have media tears in them?

I think I know the difference between a statistically valid sample and some anecdotes.


Well simple logic and the odds that only the bad Purolator oil filters fall into the hands of BITOG members who cut open oil filters is basically impossible.


That's...not what he is saying at all. He is saying there is no statistically significant data at this point.

 

[ZeeOSix]

Originally Posted By: 440Magnum
No, that's not what he's saying. He's saying that our BITOG sample size is WAY too small to be statistically valid. The way probability theory works, when you take too small a sample (in this case, BITOGers who've opened filters) of a very, very large population (all the Purolator filters produced), you can get results that would on the surface imply that something sudden happened, when in fact the exact same number of filters may actually be failing as were failing 2 years ago.


I think I was the first guy here in the last 4~5 years to post up a torn Purolator Classic ... that was back in May 2011.
First Report of Purolator Media Tear

There might have been a couple more posted after me around the same time frame, but recently (in the last 6 months) there seems to be a very noticeable uptick of reported media tears. The odds that BITOG members magically got the only 20 filters out of millions that had failures is literally astronomical.

The bottom line, regardless if it can statistically be proven or not, is there's a current problem going on with media tears in Purolator filters with the metal crimped media seam for some reason. Purolator has been made aware of this phenomenon, and hopefully they are looking into it and not just shrugging their shoulder and laughing saying those BITOG guys are the unluckiest dudes in the world.

 

[ZeeOSix]

Originally Posted By: Hollow
That's...not what he is saying at all. He is saying there is no statistically significant data at this point.


There never will be a viable "statistically significant sample" from the viewpoint of traditional statistics methodology. But there is no refuting that something is wrong with these Purolator filters that have media tears. It's been way too common of a finding lately by the small group of people in the world who cut open oil filters.

 

[ZeeOSix]

Originally Posted By: 440Magnum
And conversely- two years ago our small sample size may have been UNDER-reporting the actual number that were failing! It works both ways.


IMO, anyone here cutting open an oil filter that finds some kind of problem (like torn/failed media) is going to report it here. I can see if someone cuts open a filter and everything looks fine inside then they are not going to say anything on here. I've cut open many filters that look fine and I don't say jack about them here.

 

[qwerty1234]

I think the most disturbing part is the response of Purolator. The people who have written to them either don't receive a response or get the PC line of the company. I stocked up on a few of the Quaker State (Purolator made)Menard ones and now won't use'em.

 

[MuzzleFlash40]

Well regardless of the "statistics" or whatever, it is enough for me to take a step back from motorcraft. I held out for a long time on here while people were reporting failure after failure. Even though there weren't any reported tears with the motorcraft line. I'm not going to take a chance with my cars to find out otherwise. Will I come back to them? Absolutely, if they admit there may be a problem and fix it. Some on here may not agree. But it is my wallet and my sanity.

So I just returned all my motorcrafts and ordered Wix replacements. Found a pretty good deal on them. $5.84 shipping included.

 

[MrQuackers]

I will use my Classics and P1 filters on low OCI and begin stocking Fram Tough Guards or better.

 

[hatt]

BITOG's statistical sample is just as valid as any other. Bad filters should be a very rare occurrence. The fact that one group is finding bad filters on a regular basis is significant. And to top that there's likely only two group who cut open filters with any sort of regularity. BITOG and filter manufacturers. So 50% of the people testing are showing negative results. The other group isn't going to be posting results so they don't matter anyway. "Oh hey we've found a bunch of our filters that failed" said no one ever.

 

[Hyde244]

+1 to 440Magnum. Appreciate the great explanation.

^ It isn't the randomness of BITOGers opening oil filters that is the concern (it is arguably a very good random sample), it is the small sample size when trying to be compared to an enormously large population.

 

[ZeeOSix]

Originally Posted By: Hyde244
It isn't the randomness of BITOGers opening oil filters that is the concern (it is arguably a very good random sample), it is the small sample size when trying to be compared to an enormously large population.


Please explain to me why the samples seen here on BITOG would be a totally skewed representation in the wrong direction of what's going on with the seen media tearing. IMO, there has been enough data to theorize that the tearing issue is not only happening to BITOG members who have discover the failure.

 

[qwerty1234]

Why can't Purolator be sued for gross negligence? I guess you need proof? There should be some federal organization examining faulty oil filters. I wonder how many engines have been ruined?

 

[901Memphis]

A filter will still filter with a rip in the media. It shouldn't Realistically affect engine life but I have never seen a rip in a Fram Ultra so I choose them!

 

[Hyde244]

Originally Posted By: ZeeOSix
Please explain to me why the samples seen here on BITOG would be a totally skewed representation in the wrong direction of what's going on with the seen media tearing. IMO, there has been enough data to theorize that the tearing issue is not only happening to BITOG members who have discover the failure.
Sure thing - I'll pull from a post in the other thread, discussing why the larger the sample the better:

1. The first reason to understand why a large sample size is beneficial is simple. Larger samples more closely approximate the population. Because the primary goal of inferential statistics is to generalize from a sample to a population, it is less of an inference if the sample size is large.

2. A second reason is kind of the opposite. Small samples are bad. Why? If we pick a small sample, we run a greater risk of the small sample being unusual just by chance. Choosing 5 people to represent the entire U.S., even if they are chosen completely at random, will often result if a sample that is very unrepresentative of the population. Imagine how easy it would be to, just by chance, select 5 Republicans and no Democrats for instance.

If you are genuinely interested in learning more about Statistics and how probability (a.k.a. odds) are determined, I'd encourage you to check out the Khan Academy which has some great overview videos on stats. Here is a video that discusses the impact having a small sample size has in relation to generalizing to an entire population. http://www.khanacademy.org/math/probabil...dence-intervals

 

[901Memphis]

All relevant statistics aside... My crystal ball tells me there are lots of rips out there.

That being said i think it would be nice to see how the small rip in the media affects filtration; i personally don't believe it has a huge impact.

 

[SilverC6]

Originally Posted By: 901Memphis
All relevant statistics aside... My crystal ball tells me there are lots of rips out there.

That being said i think it would be nice to see how the small rip in the media affects filtration; i personally don't believe it has a huge impact.


I think your crystal ball is right and your conclusion also accurate.

It's just an oil filter.

How many times does the oil pump cycle the oil through the filter on a 10 minute trip to the store?

The oil is going to get some filtration on the first pass or the third or the fifth.

A big non-issue.

 

[ZeeOSix]

Originally Posted By: Hyde244
If you are genuinely interested in learning more about Statistics and how probability (a.k.a. odds) are determined, I'd encourage you to check out the Khan Academy which has some great overview videos on stats. Here is a video that discusses the impact having a small sample size has in relation to generalizing to an entire population. http://www.khanacademy.org/math/probabil...dence-intervals


So tell me why if 100 people here on BITOG each cut open one Purolator oil filter, and 10 of those people found the filter's media was torn, why wouldn't you think 100 other random people not members of BITOG cutting open one Purolator oil filter each wouldn't find a similar failure rate.

What makes all the bad Purolator oil filters land in the hands of BITOG members who just happen to cut open oil filters? Where does the "statistics" take that into account?

 

[ZeeOSix]

Originally Posted By: SilverC6
A big non-issue.


IMO, not totally the case. If torn media goes down into the engine's oiling system, it certainly could be an issue. Since Purolator uses a metal center core, that doesn't seem to be happening. Still possible that some small fibers do get flushed past the filter and in to the engine, as any kind of fibrous material that rips/tears will slough off some particles.