MEASUREMENT SYSTEM ANALYSIS (MSA)

LOGO
VARIABLE MEASUREMENT SYSTEM STUDY: BIAS FORMAT NUMBER

Instrument Name: CALIPER CHECKER No. of Trials (n) : 15 No. of Sub Group (g) : 1
Instrument Code : Confidence : 0.95 α : 0.05 Date:
Referance Value: 600.0000
Trial # Observation Bias
1 599.999 -0.0010 Histogram of Bias Study
2 600.001 0.0010
3 600.000 0.0000 5

4 600.000 0.0000
4
5 600.001 0.0010

Frequency
6 600.001 0.0010 3
7 600.001 0.0010
8 599.999 -0.0010 2

9 600.000 0.0000
1
10 600.002 0.0020
11 600.000 0.0000 0
12 600.001 0.0010 599.99 599.99 599.99 599.99 599.99 600.00 600.00 600.00 600.00 600.00 600.00 600.00 600.00
75 81 87 93 99 05 11 17 23 29 35 41 47
13 599.999 -0.0010 Observed value
14 600.000 0.0000
15 600.001 0.0010
Average 600.0003 0.0003

max (xi) - min(xi)

σrepeatability= d2*
= 0.00084

σb = σr/(n)1/2= 0.00022

t = bias / σb = 1.52910

ANALYSIS OF BIAS STUDY

n (m) Mean Standard Standard Error of

deviation σr Mean σb
Measured
15 600.0003 0.00084 0.00022
Value

Referance value = 600.00 α = 0.05 g= 1 d2* = 3.5533

t Significant t value 95% Confidence Interval of the Bias

df Bias
statistic ( 2 - tailed)
Lower Upper
Measured
1.5291 10.8 2.206 0.0003 -0.0001 0.0008
Value

Acceptance Criteria
1. Bias is acceptable at the α level if zero falls with in the 1 - α confidence bounds around the bias value:
Bias - [σb (tv , 1 - α / 2)] <= 0<= Bias + [σb (tv , 1 - α / 2)]
2. Bias should be less than the permissible error defined in calibration procedure
Conclusion:
Since zero bias falls within the confidence interval of bias , the measurement bias is acceptable

Study Conduced by: Approved by:

 RESTRICTED AREA

Calculation

1. σrepeatability
σrepeatability= max (xi) - min(xi) d2*

max (xi) = 600.002

min (xi) = 599.999
d2* = 3.5533
d2* is taken from Appendix C with g =1 and m= n ( n =15)
where g = No. of Subgrops & m = Sub group Size

σr = 0.00084

2. Standard Error of Mean σb

σr
σb=
(n)1/2
n = 15
(n) 1/2= 3.87298334621
σb = 0.00022

3. t statistics

t = bias / σb

bias = average of obseved bias = 0.0003

t = 1.52910

df = v = 10.8

" v " is taken from Appendix C with g =1 and m = n=15

where v is the degrre of freedom
 (tv , 1 - α / 2) =Significant t Value= 2.206

Significant t value is taken from t distribution table with v =10 & α = 0.025
From t distribution Table with v =10 & α = 0.025
Significiant t Value = 2.228
But actual value of v = 10.8 (from Appendix C)
But in t distribution table value of v is available either 10 or 11, so take v value as 10 and find the difference of
11 and divide the difference by 10 since there are 10 equal division in between 10 & 11 and multiply the result b
value of 0.80. Then subtract the value of 0.8 from v value of 10 to get exact v value of 10.8 since the t value decr
v (degrre of freedom) value increase.
From t distribution Table
t value for v=10 & α / 2 = 0.025 2.228
t value for v=11 & α / 2 = 0.025 2.201
Difference 0.027
t value difference of 1 division 0.027
t value difference of 0.1 division 0.0027
t value difference of 0.8 divisions 0.0216
There for significant t value with v = 2.228-0.0216= 2.2064
=10.8 and α = 0.025

The value of α is 0.05 but this value is the the total value of +ve and -ve side. So the actual value of α is ± 0.025
value is 0.025. This mean value is using for taking the significant t value from t distribution table
where α = Risk factor
t value using excel formula
t10,0.05 = 2.22813885199 t11,0.05 = 2.20098516
Now follow the above step to get the t value of t10.8 ,0.05

4. Acceptance Crateria of Bias

Bias is acceptable at the α level if zeor falls with in the 1 - α confidence bounds around the bias value:

Bias - [σb (tv , 1 - α / 2)] <= zero<= Bias + [σb (tv , 1 - α / 2)]

Bias - [σb (tv , 1 - α / 2)] = -0.0001 Bias + [σb (tv , 1 - α / 2)] =

A
as 10 and find the difference of v values of 10 and
0 & 11 and multiply the result by 8 to get the
ue of 10.8 since the t value decrease as the v

064

the actual value of α is ± 0.025 i.e the mean

distribution table

ula
formula tinv(0.05,10)
where 0.05 = α and 10 = v

round the bias value:

Bias + [σb (tv , 1 - α / 2)] = 0.0008

Bias Study

We are considering 95 % confidence level i.e 5% risk

In that 5% risk 2.5% in both the sides (i.e. + ve & - ve Side)
For MSA calculation we are considering α = 0.05 for 95% confidence level
While we are using T distribution curve in that case we will use 0.025 as α value.
Because 0.05 is in both the side , here we will consider only single side (mean value of α)

For Bias Study

g=1 i.e No of sub group
m= no of trials
V= degrees of freedom & we will get the value of v from Appendix C d2* table.

Step-1
Take master sample or any production part & calculate the reference value by taking the average of minimum10 values
using any precised instrument with the help of single appraiser who is the master of that instrument..

Step-2
Measure the sample minimum 10 times in normal manner with the help of single appraiser.

Step-3
Analysis--if there is any abnolmality in the reading then there is presence of some special causes, if not continue with the a
Special caution ought to be excercised for any interpretation or analysis when n<30.

Step-4
Caluculate the average of observed measurement value
n
Σ Xi
I=1

X bar=----------
n

Step-5

Bias = Xbar - reference value

Step-6
Compute repeatability standard deviation

σrepeatability=------------------------------------
max Xi - min Xi
---
d2*
Step-7
Sangram:
Calculate Standard Error of Mean

σ
σb = ---------------------------------------
r

d2*

Step-7
Sangram:
Calcuale t statistic for the bias this apprach uses

Bias
t = ---------------------------------------
σb

Step-8 REMARKS

Bias is acceptable at the α level if Zero falls within the 1-α confidence bounds around the bias value

Bias - [σb(tv,1-α/2)] < =zero <= Bias + [σb(tv,1-α/2)]

 e of minimum10 values
ment..
.

s, if not continue with the analysis.