Precision Testing Using Only 2 Data Points is Inadequate to Calculate Precision
To educate the Court as to the inadequacy of calculation of the instrument's precision by the local police service, close in time to the subject tests.
To obtain admissions from the CFS expert as to proper methods for calculating precision.
To confirm Mr. Kupferschmidt's evidence that the local police had stopped doing linearity checks by the time of this periodic inspection.
Q. ..... If we go back and we count back 50 control tests. A. Yes. Q. Prior to my client’s subject tests but included the values from my client’s subject tests, and we go back to March 15th, 2006. And of course, we have all that data in the Cobra data. A. In the pages, yes.
Q. So, that’s going to be back – I need to find March 15th – I’m sorry, March – March 6th, 2015. A. One question. Was this how the data was provided to you or is this...
Q. The problem... A. ...how it was printed from a spreadsheet? Q. The problem as I indicated to you, is the way that it was disclosed, the data was not in chronological order. A. Okay. Q. And the difference has to do – you’ll understand this; I don’t know that the Court will understand this, but there’s a difference between Cobra 4 and Cobra 5 reporting of data. A. Yes.
Q. Cobra 4 was always chronological when we saw the reports. Cobra 5 is not necessarily chronological. A. Yes. Q. So, the user has to take the data, do an unlocking mechanism on the data and then sort it chronologically by the fields for date and time. A. Yes. Q. Is that your understanding of what.... A. I have had to do.... Q. Of what.... A. I’ve had to do the same thing. Q. Of the wonderful task we have to do to Cobra 5 data. A. I’ve had to do it myself. Q. All right, so.... A. When I’ve been provided the data. Q. 2015, March the 6th. If we look at 2015, March the 6th, and then we start comparing cal-checks. And the first one, March the 6th, 2015 that I’ve pointed to you at 15:15:53 is a result of 95. A. Yes. Q. Right? Which compares with the top of the spreadsheet I’m showing you. Cal-check of 95. A. Yes. Q. And that was actually during a stability test. Right? So, there’s no temperature showing. At 15:15:53. A. So, that’s the same time as this periodic inspection worksheet. Q. All right. A. On page 44 of Exhibit 12. Q. Yes. All right, but in any event, we have a cal-check, a control check at 100 solution of 95. We don’t know the temperature, but it’s done during a stability test by an inspector, so I would hope the temperature of the solution is 34 degrees Celsius. Does that make sense?
A. Sorry. Can I just check one other thing?Does this go back to October 7th, 2014?
Q. No. It starts....
A. It doesn’t.
Q. It starts – I’m starting at March the 6th,2015 because the methodology that I’m suggesting could be followed in this case, is one where we go back 50 control tests.
A. Right. But my concern is that the data for this is not complete. There should be more numerical values here.
Q. For 95 – of – on – what date are you looking at?
A. Well, I’m – there’s no data for the –there’s no printouts for the periodic inspections done on May20th, 2015.
A. April 14th, 2015 or May 6th, 2015.
Q. Right. And I agree with you.
Q. That is problematic and that’s one of the subject matters of this application.
A. So, my concern is, is that there’s an I-T-P check that was done and a wet bath simulator that was done on page 44, on the date of 6th of March, 2015. But that’s not reflected in this data here. And you’re using the data from those points, but there’s only what appears to be two points,so I’m concerned that there’s....
Q. There’s an I-T-P immediately on the page
before, page number... A. Thirty-seven?
Q. ...37. There’s the I-T-P check. A. Again, it doesn’t – even though the pages are numerically accurate, page 37 to 38, the data does not appear to be there in its entirety. Q. What’s missing? A. There should be other data points. Right? So, if you refer to the wet bath simulator cal-check on page 43, and the I-T-P. Q. Yes. A. There are 10 values. Q. Yes. A. There should be 10 values that are printed here that is used to derive the average and the standard deviation which are recorded here, but for some reason.... Q. Unless – unless of course, York Regional Police is not doing it properly and someone’s taken it upon themselves to produce 10 – ah, produce 2 stability tests rather than 10, for the calculation of accuracy and precision that they record on the inspection report. I mean, maybe – I’m just asking, from your experience, when you’re running IT- P stability tests or – or control check stability tests using the software on the Intoxilyzer 8000C, first thing I’d ask you is how many tests do you run? Right? A. That’s.... Q. Could be 2, could be 10, could be 50? A. That's correct. Q. So, what’s going on more recently, in other words, with respect to the dates of – I forget what the dates are for those inspections. March, April and May, York Regional Police has switched from used to doing 10 data points
to calculate standard deviation or to calculate accuracy. Instead of using 10 they’re using 2, which I suggest to you, reflects another problem with the methodology being used by York Regional Police. And that’s why the Cobra data shows only two data points used to calculate accuracy and precision. A. I can’t tell from looking at this data whether or not the program was set to do 2 or whether it was set to do 10, as it usually does. Q. All right, take a look.... A. Or whether there was a change – I mean, there’s no way to say from here. Q. Take a look at... A. Oh. Q. ...6th of March 2015. Conduct... A. Oh. Q. ...precision testing. Testing of precision. A. Two external calibration checks. Q. Using two external calibration checks. I mean, I’m just – I’m just wondering. As a scientist, how on earth can anyone do a determination of precision based on two external calibration checks? You need more data. A. That is correct. The form changed from October 7th, 2014 to a two-page document, to a one-page document on the 6th of March 2015 and the results of the accuracy and precision testing that is done was changed from 10 for external calibration checks and 10 internal test procedure checks, to 2. Q. And the linearity checks are gone, as pointed out by Mr. Kupferschmidt. A. Correct. Yes. I also think I pointed that out.