Review Team (February 2016)

VitD-18 (February 2016) FOR ERP USE ONLY DO NOT DISTRIBUTE

Table 2: Accuracy—Bias compared to reference material

Measured Results (1)

Certified Reference Material

Bias

95% Confidence interval

CRM Product

Certified value

Certified range

Results (ug/hg)

(3)

k (2)

Mean

SD

Analyte

Uncertainty

Date

DF CRV

Nominal % Bias

(4)

DF (5)

p -value (6)

T stat

09 Dec 2015 14 Dec 2015 16 Dec 2015 21 Dec 2015 21 Dec 2015 29 Dec 2015 29 Dec 2015 31 Dec 2015 31 Dec 2015 05 Jan 2016 05 Jan 2016 07 Jan 2016 07 Jan 2016

10.9 10.6 11.0

9.8 9.4

10.0

NIST1849a

Vit D3

11.1

1.7

9.4

- 12.8

2

60

10.1

0.53

9.8 - 10.4

-9.1%

1.165

63.92

0.25

9.7

10.0

9.5 9.5 9.9

10.3 10.6

sd

d

t crit

n

Footnotes:

(1) H 0 : = 0, The null hypothesis is that there is no difference between the measured results and the certified value: (2) k = coverage factor used for clculating expanded uncertainty. These are given in the NIST certificate, although it is often 2. (3) DF CRV = calculated for k, using a 2-sided t distribution at D =0.05. An empirical model for doing this calculation was obtained by fitting 1/DF vs log t 0.05 d

to a 4th order polynomial for integral values of DF from 2-

100 DF. The maximum relative error over the fitted range was 0.06%.

(4)

where: = mean of the differences between measured results and reference value sd = standard deviation of the experimental results n = number of replicate analyses of SRM U CRV = CRV uncertainty d

(5) DF was calculated using Satterthwaite approximation:

where: n CRV -1= DF crv = mean of the differences between measured results and certified value s m = standard deviation of differences between measured results and certified value n m = number of replicate analyses of CRM U CRV = uncertainty of certified reference value k = coverage factor used for calculating expanded uncertainty (6) p-value is he probability (at the stated level of confidence) of obtaining the test statistic assuming that the null hypothesis is true. p -value < 0.05 indicate a difference of statistical signifance between measured results and CRV. d

SPIFAN ERP Checklist v1.0