Statistical Data

in Chemical
Analysis
SIR KIM MARTIN D. BELIGON
INSTRUCTOR
CAS-DPS
CONFIDENCE LIMIT
TWO COMMON APPLICATIONS OF STATISTICAL TESTS OF ANALYTICAL RESULTS

1. Defining a numerical interval around the mean of a set of replicate analytical results within
which the population mean can be expected to lie with a certain probability (Confidence Interval)

2. Determining the number of replicate measurements required to ensure at a given probability

that an experimental mean falls within a certain confidence interval.

3. Using the least-square method for constructing a calibration curves.

CONFIDENCE LIMIT
Define as numerical interval around an experimentally determined mean (x)
within which the population mean (µ) lies with a given degree of probability.

CONFIDENCE INTERVAL
A numerical magnitude of the confidence limit. The size of the confidence
interval, which is computed from the sample standard deviation, depends on
how accurately we know (s), that is, how closely we think our sample standard
deviation is to the population standard deviation (σ).
Finding the CI When s Is a Good
Estimate of σ (z-test)
The relative frequency is plotted as a function
of the quantity z, which is the deviation from
the mean normalized to the population standard
deviation.
These relationships allows us to define a range of
values around a measurement within which the true
mean is likely to lie with a certain probability
provided that we have a reasonable estimate of
σ.
 𝑧𝜎
CL or µ= 𝑥ҧ ±
𝑁
z-Table
Confidence Levels for Various Values of z

Confidence Level, % z
50 0.67
68 1.00
80 1.29
90 1.64
95 1.96
95.4 2.00
99 2.58
99.7 3.00
99.9 3.29
Example
Calculate the 80% and 95% confidence limit from the data below. Assume that in each part s=0.1
is a good estimate of σ.

Specimen 1 ppm Hg
Sample 1 1.80
Sample 2 1.58
Sample 3 1.64
Example
How many replicate measurements of a specimen 1 are needed to decrease the 95% confidence
interval to ± 0.07 ppm Hg?

𝑧𝜎
CI= ±
𝑁
Finding the CI When σ Is
Unknown (t-Test/Student t-Test)
With some limitations in time or the amount of available sample that prevent us from accurately
estimating 𝜎. A single test of replicate measurements must provide not only a mean but also an
estimate of precision.
To account for the variability of s, we use the important statistical parameter t, which is defined in
the same way as z except that s is substituted for 𝜎.

𝑡𝑠
t=
𝑥−𝜇
thus CL or µ= 𝑥ҧ ±
𝑠 𝑁
Example
You have obtained the following data for the alcohol content of a sample of blood. Calculate the
95% confidence limits for the mean assuming (a)that you know nothing about the precision of the

method.

% C2H5OH
0.084
0.089
0.079
DETECTION OF GROSS ERRORS
(Q-test)
|𝑋𝑞−𝑋𝑛|
Qexp= 𝑊 If Qexp > Qcrit
the questionable result
Where:
can be rejected with the
W= Range
Xq= questionable result
indicated degree of
Xn= nearest neighbor
confidence.
Example
Five dominations of the Vitamin C content of a citrus fruit drink gave the following results:
0.218, 0.219, 0.230, 0.215, and 0.220 mg/mL. Apply the Q-test to see if the 0.220 value can be
discarded at 90% confidence.
 THE LEAST-SQUARE METHOD
(Two-Dimensional Data)
(Linear Regression Model)
Many analytical methods are based on a calibration curve in which a measured quantity (y) is
plotted as a function of the known concentration (x) of a series of standards.
Due to some indeterminate errors in the measurement process, not all data falls exactly on the line.

Regression Analysis is a statistical technique that provides the means for objectively obtaining such a line and
also for specifying the uncertainties associated with its subsequent use.

ASSUMPTIONS OF LEAST_SQUARES METHOD

1. There is actually a linear relationship between measured variable (y) and the analyte concentration (x).
Represented as: y= mx + b

2. That any deviation of individual points from the straight line results from an error in the measurement. That
is, we must assume that there is no error in the x values of the points.
Computing the Regression Coefficients
and Finding the Least-Squares Line
3 Quantities:

2
෌𝑥𝑖
𝐒𝐱𝐱 = ෌ 𝑥𝑖2 −
𝑁
2
෌𝒚𝑖
𝐒𝐲𝐲 = ෌ 𝒚2𝑖 −
𝑁

σ𝒙𝒊 σ𝒚𝒊
𝐒𝐱𝐲 = σ 𝒙𝒊𝒚𝒊 −
𝑁
 6 useful quantities can be derived
from Sxx, Syy, Sxy
𝑠𝑥𝑦 𝑆𝑟2
1. Slope of the line: 𝑚= 4. Standard deviation of a slope Sm: 𝑆𝑚 =
𝑠𝑥𝑥 𝑆𝑥𝑥

1
2. Intercept b: b= 𝑦ത − 𝑚𝑥ҧ 5. Standard Deviation of intercept Sb: 𝑠𝑏 = 𝑆𝑟
𝑁 − 𝛴𝑥𝑖 2 /𝛴𝑥𝑖 2

𝑆𝑟 1 1 𝑦ത 𝐶 −𝑦ത 2
3. Standard deviation about regression Sr: 6. Standard deviation obtained from calibration curve: 𝑆𝐶 = + +
𝑚 𝑀 𝑁 𝑚2 𝑆𝑥𝑥

𝑆𝑦𝑦 −𝑚2 𝑆𝑥𝑥

Sr=
𝑁−2
Example
x y xy x2 y2
1.00 2.96 2.96 1.00 8.76
2.00 5.05 10.10 4.00 25.50
3.00 7.03 21.09 9.00 49.42
4.00 8.92 35.68 16.00 79.57
5.00 10.94 54.70 25.00 119.68
∑x= 15.00 ∑y= 34.90 ∑xy= 124.53 ∑x2= 55.00 ∑y2= 282.93

2
෌𝑥𝑖
Sxx= ෌ 𝑥𝑖2 − =
𝑁

2
෌𝒚𝑖
Syy = ෌ 𝒚2𝑖 − =
𝑁

σ𝒙𝒊 σ𝒚𝒊
𝐒𝐱𝐲 = σ 𝒙𝒊𝒚𝒊 − =
𝑁
 𝑠𝑥𝑦
𝑚= = b= 𝑦ത − 𝑚𝑥ҧ =
𝑠𝑥𝑥

Therefore the equation for the least-square line is : y=

𝑆𝑦𝑦 −𝑚2 𝑆𝑥𝑥

The standard deviation of y-values, is given by: S r= 𝑁−2
=

𝑆𝑟2
The standard deviation of slope: 𝑆𝑚 = =
𝑆𝑥𝑥

1
The standard deviation of the intercept: 𝑠𝑏 = 𝑆𝑟 =
𝑁 − 𝛴𝑥𝑖 2 /𝛴𝑥𝑖 2