Error Analysis
Repeating Measurements
Calculation of Mean and Standard
Deviation
The Gaussian distribution
Propagation of Errors
Significant Figures
Review quiz
Kirk is sitting in the right-hand passenger seat of a car. The car makes a right-hand turn at constant speed. If Kirk stays in his seat as the car turns, there is
A. no force on Kirk.
B. a horizontal force directed forward on Kirk.
C. a horizontal force directed to the left on Kirk.
D. a horizontal force directed to the right on
Kirk.
E. a horizontal force in a direction between forward and left on Kirk.
About the Mechanics Test
Test average was 64%.
You will receive your marked test near the END of tutorial this week.
If you find a mistake in the marking you must notify
Dr. Savaria in MP129 before next Friday, November
16 by 5:00PM.
This guy is responsible for calculating your mark!
Test 1 Histogram
Percentage with A 19%
Percentage with B 16%
Percentage with C 28%
Percentage with D 23%
Percentage with F 14%
Two Kinds of Statements
1. Exact
2 + 3 = 5 (math)
K = ½ m v 2 (definition)
2. Approximate
F spring
= – k x ( any physical law) g = 9.80 m/s 2 ( all numerical measures of the universe)
Today: approximate statements
Period of a Pendulum
Procedure: Measure the time for 5 oscillations, t
5
.
The period is calculated as T = t
5
/ 5.
Did Harlow do anything wrong when measuring t
5
?
A. No
B.
Yes, he should have counted “Zero” when he started the stopwatch.
C. Yes, he should have started the stopwatch when it was at the bottom of its swing, not at the top.
The t
5
data
7.53 s
7.38 s
7.47 s
7.43 s
Repeated Measurements of Period
Consider a single measurement, in a group of measurements that follow a normal distribution. What is the probability that this measurement lies within + or – one standard deviation σ of the mean?
A. 0%
B. 50%
C. 68%
D. 95%
E. 100%
Here were Harlow’s measurements of t
5
:
7.53 s
7.38 s
7.47 s
7.43 s
Which of the following might be a good estimate for the error in
Harlow’s first measurement of
7.53 seconds?
A. 0.005 s
B. 0.05 s
C. 0.5 s
D. 5 s
E. Impossible to determine
Histogram: 4 Measurements
7
6
5
7.53 s
7.38 s
7.47 s
7.43 s
2
1
4
3
0
7.2
7.25
7.3
7.35
7.4
7.45
7.5
7.55
7.6
7.65
7.7
Measured Time (half second bins)
Histogram: 8 Measurements
7
6
5
4
3
2
1
7.53 s
7.38 s
7.47 s
7.43 s
7.44 s
7.56 s
7.48 s
7.40 s
0
7.2
7.25
7.3
7.35
7.4
7.45
7.5
7.55
7.6
7.65
7.7
Measured Time (half second bins)
Histogram: 12 Measurements
7
6
5
4
3
2
1
0
7.2
7.25
7.3
7.35
7.4
7.45
7.5
7.55
7.6
7.65
7.7
Measured Time (hafl second bins)
Histogram: 16 Measurements
5
4
3
2
7
6
1
0
7.2
7.25
7.3
7.35
7.4
7.45
7.5
7.55
7.6
7.65
7.7
Measured Time (half second bins)
Histogram: 16 Measurents
4
3
7
6
5
Gaussian Curve
(best fit)
Stopwatch
Measurements
2
1
0
7.2
7.25
7.3
7.35
7.4
7.45
7.5
7.55
7.6
7.65
7.7
Measured Time (half second bins)
The Gaussian
68% of data between the dotted lines on the graph.
Heights of some People
(London, 1886)
inches
Random Walk
Where does an object end up, if it takes
N steps randomly left or right?
The final distribution is described by a
Gaussian function!
The t
5
data
7.53 s
7.38 s
7.47 s
7.43 s
+ 0.06 s
+ 0.06 s
+ 0.06 s
+ 0.06 s
Numerically:
Propagation of Errors
z = A x Δz = A Δx
Repeated Measurements
Repeated n times
Each individual measurement has an error of precision
D x
Significant Figures
Discussed in Section 1.9 of Knight Ch.1
Rules for significant figures follow from error propagation
Assume error in a quoted value is half the value of the last digit.
Errors should be quoted to 1 or 2 significant figures
Error should be in final displayed digit in number.
Example: If a calculated result is (7.056 +/-
0.705) m, it is better to report (7.1 +/- 0.7) m.