A tale of 3 identical VOA samples tested by three different test laboratories...

Joined
Jul 16, 2021
Messages
337
I thought some may find this interesting.

I sent off three different 0w30 oils each to three different laboratories.

Each oil was sample was from the same container of each oil.

The laboratories i used were:

Blackstone
Wearcheck
Polaris

The three virgin samples sent to each lab were:
Mobil 1 0w30 AFE
Mopar MaxPro+ 0w30
Amsoil SSO 0w30

While i cannot say which if any laboratory provided accurate results i can at least show the reliability between laboratories.

Here are the results...

oil analysis picture.jpg
 
Its a bit unsettling that Polaris lists the oxidation at 47 abs/cm on virgin amsoil. Clearly this must be a significant error.
 
Some of these common numbers have as much as a 20% difference across the tests. That's significant variation. You'd almost have to do this test with several labs and then just take the averages.
 
Some of these common numbers have as much as a 20% difference across the tests. That's significant variation. You'd almost have to do this test with several labs and then just take the averages.

The problem with taking averages is it assumes averaging the numbers gets you closer to accurate results but theres no way of knowing that. The true results could even be higher or lower than the highest or lowest results.
 
I thought some may find this interesting.

I sent off three different 0w30 oils each to three different laboratories.

Each oil was sample was from the same container of each oil.

The laboratories i used were:

Blackstone
Wearcheck
Polaris

The three virgin samples sent to each lab were:
Mobil 1 0w30 AFE
Mopar MaxPro+ 0w30
Amsoil SSO 0w30

While i cannot say which if any laboratory provided accurate results i can at least show the reliability between laboratories.

Here are the results...
Don't you mean to say,
"I can at least show the variability between laboratories?"
 
Don't you mean to say,
"I can at least show the variability between laboratories?"
You can have high reliability in data or low reliability in data. You can have high accuracy in data or low accuracy in data.
 
You can have high reliability in data or low reliability in data. You can have high accuracy in data or low accuracy in data.

Reliability - the degree to which the result of a measurement, calculation, or specification can be depended on to be accurate.

Reliability is dependent on the accuracy to which the measurements and procedures were made. Since you don't know the degree of accuracy to which the measurements or procedures were made, you can not determine any measure of reliability.
 
Reliability - the degree to which the result of a measurement, calculation, or specification can be depended on to be accurate.

Reliability is dependent on the accuracy to which the measurements and procedures were made. Since you don't know the degree of accuracy to which the measurements or procedures were made, you can not determine any measure of reliability.

Reliability in scientific experiments have nothing to do with accuracy. Reliability is the extent to which results repeatably produce the same result. This is regardless of their accuracy.

A speedometer that consistently reads 30mph high in any and all conditions is dead reliable. Its just not accurate.




The results from the above laboratories lack perfect reliability but any one of them could be accurate or none of them could be accurate. We dont know.
 
Last edited:
Reliability in scientific experiments have nothing to do with accuracy. Reliability is the extent to which results repeatably produce the same result. This is regardless of their accuracy.
As per the above definition I gave (Reliability - the degree to which the result of a measurement, calculation, or specification can be depended on to be accurate.), reliability is highly dependent on the accuracy of the measurements, calculations, specifications, or procedures.

I think your chart above shows variability, but since you admitted that you "cannot say which if any laboratory provided accurate results," you cannot show any reliability in the results, only the variability between the labs.
 
The definitions of reliability in your links seem more similar (not exactly) to "precision" vs. "accuracy" arguments.
 
Last edited:
The definition you gave is not correct. See the links i posted. Reliability has nothing to do with accuracy in science. Scientific study 101. Accuracy, reliability, validity. Three different things.

Precision is yet another construct.

In lay peoples terminology they say a watch is "precise" or "accurate" when in science what they really mean is its reliable.

If you set a reliable watch incorrectly by 5 minutes and it remains EXACTLY 5 minutes off for the rest of its life it will never be accurate during that time but it will be perfectly reliable. Absolutely dead on reliable... but heres the kicker...potentially not very precise either.

precision has to do with the amount of resolution in the measure. An anolog clock or watch with a second hand may be precise to 1 second where a stop watch may be precise to a thousandth of a second.
 
Last edited:
  • Like
Reactions: MM8
Precision has a bit more meaning than just the resolution of the measurement: https://techqualitypedia.com/accuracy-and-precision/
"Accuracy is measuring near the target/true value while Precision is getting consistent results of repeated measurements."

But, beyond this academic discussion, I think that Molakule is correct that the data you presented does not indicate any level of reliability.
 
Last edited:
Precision has a bit more meaning than just the resolution of the measurement: https://techqualitypedia.com/accuracy-and-precision/

But, beyond this academic discussion, I think that Molakule is correct that the data you presented does not indicate any level of reliability.

Exactly why i indicated the results were indicators of reliability and not accuracy. Three separate lab results of the same sample is a test of reliability not necessarily accuracy. As i said you can have good reliability or poor reliability. The results indicate somewhat poor reliability.
 
O.k.. I had to re-read your original post to see that you stated " i can at least show the reliability between laboratories."

Key word being not of each lab itself, but between labs.

I do soil testing and a big problem is that labs across the country use different analytical methods with different scales. 100 ppm phosphorous in one region might be very high while in another region it would indicate low or medium, if that makes sense. I think that TBN is like this - results cannot be compared between labs unless you know what analytic method they are using.

EDIT to response below: agreed, which is why I originally questioned your use of reliability - I missed that crucial word "between".
 
Last edited:
O.k.. I had to re-read your original post to see that you stated " i can at least show the reliability between laboratories."

Key word being not of each lab itself, but between labs.

You cant have reliability with just one lab/result ;)

Reliability requires multiple measures. Accuracy does not.

Has to be between labs.
 
Back
Top