Experimental Uncertainty (Error)

and Data Analysis

Advance Study Assignment
Please contact Dr. Reuven at [email protected] if you
have any questions
Read the Theory part of the experiment (pages 2-14) and answer the following questions.

1. Do experimental measurements give the true value of a physical quantity? Explain

2. Distinguish between random (statistical) error and systematic error. Give an example of each.

4. What is the difference between measurement accuracy and precision? Explain the general dependence of these
properties on the various types of errors.

5. What determines how many figures are significant in reported measurement values? What would be the effect of
reporting more or fewer figures or digits than are significant?

6. In expressing experimental error or uncertainty, when should (a) experimental error and (b) percent difference be
used?

7. How could the function y = 3t2 + 4 be plotted on a graph to produce a straight line? What would be the
numerical values of the slope and intercept of the line?

1
THEORY

A. Types of Experimental Uncertainty

Experimental uncertainty (error) generally can be classified as being of two types:
(1) random or statistical error
(2) systematic error
These are also referred to as
(1) indeterminate error and (2) determinate error, respectively.

Let’s take a closer look at each type of experimental uncertainty.

Random (Indeterminate) or Statistical Error

Random errors result from unknown and unpredictable variations that arise in all experimental measurement
situations. The term indeterminate refers to the fact that there is no way to determine the magnitude or sign (+,
too large; –, too small) of the error in any individual measurement. Conditions in which random errors can result
include:

1. Unpredictable fluctuations in temperature or line voltage.

2. Mechanical vibrations of an experimental setup.
3. Unbiased estimates of measurement readings by the observer.

Repeated measurements with random errors give slightly different values each time. The effect of random errors
may be reduced and minimized by improving and refining experimental techniques.

Systematic (Determinate) Errors

Systematic errors are associated with particular measurement instruments or techniques, such as an improperly
calibrated instrument or bias on the part of the observer. The term systematic implies that the same magnitude and
sign of experimental uncertainty are obtained when the measurement is repeated several times. Determinate means
that the magnitude and sign of the uncertainty can be determined if the error is identified. Conditions from which
systematic errors can result include

1. An improperly “zeroed” instrument, for example, an ammeter as shown in ● Fig. 1.1.

Figure 1.1 Systematic error. An improperly zeroed

instrument gives rise to systematic error. In this case
the ammeter, which has no current through it, would
systematically give an incorrect reading larger that the true
value. (After correcting the error by zeroing the meter,
which scale would you read when using the ammeter?)

2
2. A faulty instrument, such as a thermometer that reads 101 °C when immersed in boiling water at standard
atmospheric pressure. This thermometer is faulty because the reading should be 100 °C.

(a) Temperature measurement

3. Personal error, such as using a wrong constant in calculation or always taking a high or low reading of a scale
division. Reading a value from a measurement scale generally involves aligning a mark on the scale. The
alignment—and hence the value of the reading— can depend on the position of the eye (parallax). Examples of such
personal systematic error are shown in ● Fig. 1.2.

Figure 1.2 Personal error. Examples of personal error due

to parallax in reading (a) a thermometer and (b) a meter
stick. Readings may systematically be made either too
high or too low.

(b) Length measurement

Avoiding systematic errors depends on the skill of the observer to recognize the sources of such errors and to
prevent or correct them.

B. Accuracy and Precision

Accuracy and precision are commonly used synonymously, but in experimental measurements there is an
important distinction.
The accuracy of a measurement signifies how close it comes to the true (or accepted) value—that is, how nearly
correct it is.

3
Example 1.1 Two independent measurement results using the diameter d and circumference c of a circle in
the determination of the value of p are 3.140 and 3.143. (Recall that p 5 c/d.) The second result is more
accurate than the first because the true value of p, to four figures, is 3.142.

Precision refers to the agreement among repeated measurements—that is, the “spread” of the measurements or how
close they are together. The more precise a group of measurements, the closer together they are. However, a large
degree of precision does not necessarily imply accuracy, as illustrated in ● Fig. 1.3.

Figure 1.3 Accuracy and precision. The true value in this analogy is the bull’s eye. The degree of scattering is an
indication of precision—the closer together a dart grouping, the greater the precision. A group (or symmetric
grouping with an average) close to the true value represents accuracy.

Example 1.2 Two independent experiments give two sets of data with the expressed results and
uncertainties of 2.5 ± 0.1 cm and 2.5 ± 0.2 cm, respectively. The first result is more precise than the second
because the spread in the first set of measurements is between 2.4 and 2.6 cm, whereas the spread in the
second set of measurements is between 2.3 and 2.7 cm. That is, the measurements of the first experiment are
less uncertain than those of the second.

Obtaining greater accuracy for an experimental value depends in general on minimizing systematic errors.
Obtaining greater precision for an experimental value depends on minimizing random errors.

C. Least Count and Significant Figures

In general, there are exact numbers and measured numbers (or quantities). Factors such as the 100 used in
calculating percentage and the 2 in 2πr are exact numbers. Measured numbers, as the name implies, are those
obtained from measurement instruments and generally involve some error or uncertainty. In reporting
experimentally measured values, it is important to read instruments correctly. The degree of uncertainty of a number
read from a measurement instrument depends on the quality of the instrument and the fineness of its measuring
scale. When reading the value from a calibrated scale, only a certain number of figures or digits can properly be
obtained or read. That is, only a certain number of figures are significant. This depends on the least count of the
instrument scale, which is the smallest subdivision on the measurement scale. This is the unit of the smallest reading
that can be made without estimating.
For example, the least count of a meter stick is usually the millimeter (mm). We commonly say “the meter stick is
calibrated in centimeters (numbered major divisions) with a millimeter least count.” (See ● Fig. 1.4.)

Figure 1.4 Least count. Meter sticks are commonly calibrated in centimeters (cm), the numbered major divisions,
with a least count, or smallest subdivision, of millimeters (mm).

4
The significant figures (sometimes called significant digits) of a measured value include all the numbers that
can be read directly from the instrument scale, plus one doubtful or estimated number—the fractional part of the
least count smallest division.
For example, the length of the rod in Fig. 1.4 (as measured from the zero end) is 2.64 cm. The rod’s length is known
to be between 2.6 cm and 2.7 cm. The estimated fraction is taken to be 4/10 of the least count (mm), so the doubtful
figure is 4, giving 2.64 cm with three significant figures.
Thus, measured values contain inherent uncertainty or doubtfulness because of the estimated figure. However, the
greater the number of significant figures, the greater the reliability of the measurement the number represents.
For example, the length of an object may be read as 3.65 cm (three significant figures) on one instrument scale and
as 3.5605 cm (five significant figures) on another. The latter reading is from an instrument with a finer scale (why?)
and gives more information and reliability.
Zeroes and the decimal point must be properly dealt with in determining the number of significant figures in
a result. For example, how many significant figures does 0.0543 m have? What about 209.4 m and 2705.0 m? In
such cases, the following rules are generally used to determine significance:

1. Zeroes at the beginning of a number are not significant. They merely locate the decimal point.
For example, 0.0543 m has three significant figures (5, 4, and 3).
2. Zeroes within a number are significant.
For example, 209.4 m has four significant figures (2, 0, 9, and 4).
3. Zeroes at the end of a number after the decimal point are significant.
For example, 2705.0 has five significant figures (2, 7, 0, 5, and 0).
Some confusion may arise with whole numbers that have one or more zeroes at the end without a decimal point.
Consider, for example, 300 kg, where the zeroes (called trailing zeroes) may or may not be significant. In such
cases, it is not clear which zeroes serve only to locate the decimal point and which are actually part of the
measurement (and hence significant). That is, if the first zero from the left (300 kg) is the estimated digit in the
measurement, then only two digits are reliably known, and there are only two significant figures.
Similarly, if the last zero is the estimated digit (300 kg), then there are three significant figures.
This ambiguity is removed by using scientific (powers of 10) notation:

3.0 × 102 kg has two significant figures.

3.00 × 102 kg has three significant figures.

This procedure is also helpful in expressing the significant figures in large numbers.
For example, suppose that the average distance from Earth to the Sun, 93,000,000 miles, is known to only four
significant figures. This is easily expressed in powers of 10 notation: 9.300 × 107 mi.

D. Computations with Measured Values

Calculations are often performed with measured values, and error and uncertainty are “propagated” by the
mathematical operations—for example, multiplication or division. That is, errors are carried through to the results
by the mathematical operations. The error can be better expressed by statistical methods;
however, a widely used procedure for estimating the uncertainty of a mathematical result involves the use of
significant figures.
The number of significant figures in a measured value gives an indication of the uncertainty or reliability of a
measurement. Hence, you might expect that the result of a mathematical operation can be no more reliable than
the quantity with the least reliability, or smallest number of significant figures, used in the calculation.

That is, reliability cannot be gained through a mathematical operation.

It is important to report the results of mathematical operations with the proper number of significant figures.
This is accomplished by using rules for
(1) multiplication and division
(2) addition and subtraction.
To obtain the proper number of significant figures, one rounds the results off. The general rules used for
mathematical operations and rounding follow.

5
Significant Figures in Calculations
1. When multiplying and dividing quantities, leave as many significant figures in the answer as there are in the
quantity with the least number of significant figures.
2. When adding or subtracting quantities, leave the same number of decimal places (rounded) in the answer as there
are in the quantity with the least number of decimal places.

Rules for Rounding

1. If the first digit to be dropped is less than 5, leave the preceding digit as is.
2. If the first digit to be dropped is 5 or greater, increase the preceding digit by one. Notice that in this method, five
digits (0, 1, 2, 3, and 4) are rounded down and five digits (5, 6, 7, 8, and 9) are rounded up.

What the rules for determining significant figures mean is that the result of a calculation can be no more accurate
than the least accurate quantity used. That is,

you cannot gain accuracy in performing mathematical operations.

These rules come into play frequently when doing mathematical operations with a calculator that may give a string
of digits. ● Fig. 1.5 shows the result of the division of 374 by 29.

Figure 1.5 Insignificant figures. The calculator shows

the result of the division operation 374/29. Because
there are only two significant figures in the 29, a
reported result should have no more than two
significant figures, and the calculator display value
should be rounded off to 13.
The result must be rounded off to two significant
figures—that is, to 13. (Why?)

Example 1.3 Applying the rules.

Multiplication:
2.5 m × 1.308 m = 3.3 m2
(2 sf) (4 sf) (2 sf)

Division:
(4 sf)
882.0 s
 3600 s  3.60 102 s
0.245
(3 sf) (represented to three significant figures; why?)

6
Addition:
46.4
 1.37
0.505
______
48.275  48.3
(rounding off) (46.4 has the least number of decimal places)

Subtraction:
163
 4.5
______
158.5  159
(rounding off) (163 has the least number of decimal places, none)

E. Expressing Experimental Error and Uncertainty

Percent Error
The object of some experiments is to determine the value of a well-known physical quantity—for example, the value
of π.
The accepted or “true” value of such a quantity found in textbooks and physics handbooks is the most accurate
value (usually rounded off to a certain number of significant figures) obtained through sophisticated experiments or
mathematical methods.
The absolute difference between the experimental value E and the accepted value A, written |E - A| , is the
positive difference in the values,
for example, |2 - 4| = 2 and |4 – 2| = 2.
Simply subtract the smaller value from the larger, and take the result as positive.
For a set of measurements, E is taken as the average value of the experimental measurements.
The fractional error is the ratio of the absolute difference and the accepted value:

The fractional error is commonly expressed as a percentage to give the percent error of an experimental value.

7
Example 1.4 A cylindrical object is measured to have a diameter d of 5.25 cm and a circumference c of 16.38
cm. What are the experimental value of p and the percent error of the experimental value if the accepted
value of p to two decimal places is 3.14?

Solution with d = 5.25 cm and c = 16.38 cm,

Then E = 3.12 and A = 3.14, so

If the uncertainty in experimentally measured values as expressed by the percent error is large, you should check for
possible sources of error. If found, additional measurements should then be made to reduce the uncertainty. Your
instructor may wish to set a maximum percent error for experimental results.

Percent Difference
It is sometimes instructive to compare the results of two measurements when there is no known or accepted value.
The comparison is expressed as a percent difference, which is the ratio of the absolute difference between the
experimental values E and E and the average or mean value of the two results, expressed as a percent.
2 1

Dividing by the average or mean value of the experimental values is logical, because there is no way of deciding
which of the two results is better.

Example 1.5 What is the percent difference between two measured values of 4.6 cm and 5.0 cm?

Solution With E = 4.6 cm and E = 5.0 cm,

1 2

As in the case of percent error, when the percent difference is large, it is advisable to check the experiment for errors
and possibly make more measurements. In many instances there will be more than two measurement values.

8
When there are three or more measurements, the percent difference is found by dividing the absolute
value of the difference of the extreme values (that is, the values with greatest difference) by the
average or mean value of all the measurements

Average (Mean) Value

Most experimental measurements are repeated several times, and it is very unlikely that identical results will be
obtained for all trials. For a set of measurements with predominantly random errors (that is, the measurements are
all equally trustworthy or probable), it can be shown mathematically that the true value is most probably given by
the average or mean value. The average or mean value x of a set of N measurements is

where the summation sign Σ is a shorthand notation indicating the sum of N measurements from x1 to xN. ( x is
commonly referred to simply as the mean.)

Example 1.6 What is the average or mean value of the set of numbers 5.42, 6.18, 5.70, 6.01, and 6.32?

F. Graphical Representation of Data

It is often convenient to represent experimental data in graphical form, not only for reporting but also to obtain
information.

Graphing Procedures
Quantities are commonly plotted using rectangular Cartesian axes ( X and Y ). The horizontal axis (X) is called the
abscissa, and the vertical axis (Y ), the ordinate. The location of a point on the graph is defined by its
coordinates x and y, written (x, y), referenced to the origin O, the intersection of the X and Y axes.
When plotting data, choose axis scales that are easy to plot and read. The graph in ● Fig. 1.6A shows an example
of scales that are too small. This “bunches up” the data, making the graph too small, and the major horizontal scale
values make it difficult to read intermediate values. Also, the dots or data points should not be connected.

9
Figure 1.6A Poor graphing. An example of an improperly labeled and plotted graph.

Choose scales so that most of the graph paper is used. The graph in ● Fig. 1.6B shows data plotted with more
appropriate scales
Also note in Fig. 1.6A that scale units on the axes are not given. For example, you don’t know whether the units of
displacement are feet, meters, kilometers, or whatever.
Scale units should always be included, as in Fig. 1.6B. It is also acceptable, and saves time, to use standard
unit abbreviations, such as N for newton and m for meter. This will be done on subsequent graphs.
With the data points plotted, draw a smooth line described by the data points. Smooth means that the line does not
have to pass exactly through each point but connects the general areas of significance of the data points ( not
connecting the data points as in Fig. 1.6A). The graph in Fig. 1.6B with an approximately equal number of points
on each side of the line gives a “line of best fi t.”

10
Figure 1.6B Proper graphing. An example of a properly labeled and plotted graph.

In cases where several determinations of each experimental quantity are made, the average value is plotted and the
mean deviation or the standard deviation may be plotted as error bars.

For example, the data for the period of a mass oscillating on a spring given in Table 1.1 are plotted in ● Fig. 1.7,
period (T) versus mass (m).

11
Figure 1.7 Error bars. An example of graphically presented data with error bars. An error bar indicates the precision
of a measurement. In this case, the error bars represent mean deviations.

(The d is the mean deviation, shown here for an illustration of error bars.) A smooth line is drawn so as to pass
within the error bars.

Graphs should have the following elements (see Fig. 1.7):

1. Each axis labeled with the quantity plotted.

2. The units of the quantities plotted.
3. The title of the graph on the graph paper (commonly
listed as the y-coordinate versus the x-coordinate).
4. Your name and the date.

Straight-Line Graphs
Two quantities (x and y) are often linearly related; that is, there is an algebraic relationship of the form y = mx + b,
where m and b are constants. When the values of such quantities are plotted, the graph is a straight line, as shown in
● Fig. 1.8.
The m in the algebraic relationship is called the slope of the line and is equal to the ratio of the intervals Dy/Dx.
Any set of intervals may be used to determine the slope of a straight-line graph;

12
Figure 1.8 Straight-line slope. Examples of intervals for determining the slope of a straight line. The slope is the
ratio of Dy/Dx (or Dd/Dt). Any set of intervals may be used, but the endpoints of an interval should be relatively far
apart, as for Dy2/Dx2.

for example, in Fig. 1.8,

Points should be chosen relatively far apart on the line. For best results, points corresponding to data points should
not be chosen, even if they appear to lie on the line.
The b in the algebraic relationship is called the y- intercept and is equal to the value of the y-coordinate where the
graph line intercepts the Y-axis. In Fig. 1.8, b = 3 cm.
Notice from the relationship that y = mx + b, so that when x = 0, then y = b. If the intercept is at the origin (0, 0),
then b = 0.
The equation of the line in the graph in Fig. 1.8 is d = 7.5t + 3. The general equation for uniform motion has the
form d = vt + do. Hence, the initial displacement do = 3 cm and the speed v = 7.5 cm/s.

13
Straight-Line Graphs For Non-Linear Functions (Linearizing)
Some forms of nonlinear functions that are common in physics can be represented as straight lines on a Cartesian
graph. This is done by plotting nonlinear values.
For example, if y = ax + b is plotted on a regular y-versus-x graph, a parabola would be obtained. But if the
2

substitution x = x' were used, the equation becomes y = ax' + b which has the form of a straight line.
2

This means plotting y versus x', would give a straight line. Since x' = x , the squared values of x must be plotted.
2

That is, square all the values of x in the data table, and plot these numbers with the corresponding y values.

Other functions can be linearized by this procedure, including an exponential function:

y = Aeax
In this case, taking the natural logarithm of both sides:
ln y = ln A + ln eax

or, since ln eax = ax

ln y = ax + ln A
Plotting the values of the natural (base e) logarithm versus x gives a straight line with slope a and an intercept ln A.

Similarly, for
y = axn

using the common (base 10) logarithm,

log y = log a + log xn

and, since log xn = nlogx

log y = n log x + log a

Plotting the values of log y versus log x gives a straight line with slope n and intercept log a.

14
Laboratory Report
EXPERIMENTAL PROCEDURE
Complete the exercises in the Laboratory Report, showing calculations and attaching graphs as required. (Note: In
this experiment and throughout, attach an additional sheet for calculations if necessary.)

INTRODUCTION AND OBJECTIVES

Laboratory investigations involve taking measurements of physical quantities, and the process of taking any
measurement always involves some experimental uncertainty or error. Suppose you and another person
independently took several measurements of the length of an object. It is highly unlikely that you both would come
up with exactly the same results. Or you may be experimentally verifying the value of a known quantity and want to
express uncertainty, perhaps on a graph. Therefore, questions such as the following arise:

• Whose data are better, or how does one express the degree of uncertainty or error in experimental measurements?
• How do you compare your experimental result with an accepted value?
• How does one graphically analyze and report experimental data?

In this introductory study experiment, types of experimental uncertainties will be examined, along with some
methods of error and data analysis that may be used in subsequent experiments.
After performing the experiment and analyzing the data, you should be able to do the following:

1. Categorize the types of experimental uncertainty (error), and explain how they may be reduced.
2. Distinguish between measurement accuracy and precision, and understand how they may be improved
experimentally.
3. Define the term least count and explain the meaning and importance of significant figures (or digits) in
reporting measurement values.
4. Express experimental results and uncertainty in appropriate numerical values so that someone reading your report
will have an estimate of the reliability of the data.
5. Represent measurement data in graphical form so as to illustrate experimental data and uncertainty visually

EQUIPMENT NEEDED

• Rod or other linear object less than 1 m in length

• A measuring tape or ruler with calibrations of meter, centimeter, and millimeter, respectively
• Pencil and ruler
• Calculator
• 3 sheets of Cartesian graph paper
• French curve (optional)

1. Least Counts (see p. 4: Fig. 1.4)

(a) Use the measuring tape to measure the length of the object chosen above four times: in meters,
centimeters, and millimeters, respectively and record with the appropriate number of significant figures in
Data Table 1.

DATA TABLE 1
Purpose: To express least counts and measurements. (Skip the dm column)

(b) If the actual length of the object is given as 0.9 m, Find the percent errors for the four measurements
and record in in Data Table 2.

15
DATA TABLE 2
Purpose: To express the percent errors.

2. Significant Figures
(a) Express the numbers listed in Data Table 3 to three significant figures, writing the numbers in the first
column in normal notation and the numbers in the second column in powers of 10 (scientific) notation.

DATA TABLE 3
Purpose: To practice expressing significant figures.

(b) A rectangular block of wood is measured to have the dimensions 11.2 cm × 3.4 cm × 4.10 cm. Compute
the volume of the block, showing explicitly (by underlining) how doubtful figures are carried through the
calculation, and report the final answer with the correct number of significant figures.

Calculations Computed volume

(show work) (in powers of 10 notation) ___________________ (units)

(c) In an experiment to determine the value of π, a cylinder is measured to have an average value of 4.25
cm for its diameter and an average value of 13.39 cm for its circumference. What is the experimental value
of π to the correct number of significant figures?

Calculations
(show work) Experimental value of π ___________________(units)

16
3. Expressing Experimental Error
(a) If the accepted value of π is 3.1416, what are the fractional error and the percent error of the
experimental value found in previous part 2(c)?

Calculations
(show work) Fractional error ___________________
Percent error ___________________

(b) In an experiment to measure the acceleration due to gravity (g), two values, 9.96 m/s and 9.72 m/s , are
2 2

determined. If the accepted value is 9.80 m/s , find:

2

(1) the percent difference of the measurements,

(2) the percent error of each measurement
(3) the percent error of their mean.

Calculations
(show work) Percent difference ___________________
Percent error of E1 ___________________
Percent error of E2 ___________________
Percent error of mean ___________________

(c) Data Table 4 shows data taken in a free-fall experiment. Measurements were made of the distance of
fall (y) at each of four precisely measured times. Complete the table by filling the yaverage and the t2
columns. What are the units of t2?
Use only the proper number of significant figures in your table entries, even if you carry extra digits during
your intermediate calculations.

DATA TABLE 4
Purpose: To practice analyzing data.

yaverage

(d) Using Logger Pro, plot a graph of yaverage against the time t for the free-fall data in part (c).
(Remember that t = 0 is a known point). Paste the graph below

17
Can the plotted graph modeled with a straight line?

1 2
(e) The theoretical equation of for an object in free-fall starting from rest is y  gt , where g is the
2
acceleration due to gravity. The graph of such equation is a parabola (y = ax2, where x → t and a → 1 g )
2
Using Logger Pro, plot a graph of yaverage against t . Paste the graph below
2

18
 Can the plotted graph modeled with a straight line?

If yes, determine the slope of the line and, using the theoretical equation above, calculate the experimental
value of g from the slope value.

Calculations
(show work)
Experimental value of g from graph _____________________(units)

(f) If the accepted value: g is 9.8 m/s , calculate the percent error of the experimental value of g above
2

Calculations
(show work)
Percent error ________________

(g) When stretching a spring, the theoretical relationship between the stretching force F and the stretching
displacement x of the spring has the form F = kx, where the constant k is called the spring constant
and is a measure of the “stiffness” of the spring.
Notice that this equation has the form of a straight line.
Find the value of the spring constant k of the spring used in determining the experimental data plotted in
the Fig. 1.6B graph on page 11. (Note: Because k = F/x, the units of k in the graph are N/m.)

Calculations
(show work)
Value of spring constant of
spring in Fig. 1.6B graph ____________________(units)

(h) When a mass m suspended on a spring that has a spring constant k of is oscillating vertically, the
m
theoretical relationship for the period of oscillation T is T  2 .
k
Using LoggerPro, plot the data for T and m in Table 1.1 on page 11 so as to obtain a straight-line graph,
and determine the value of the spring constant used in the experiment.
[Hint: Linearize the above equation by squaring both sides, and plot the values for T 2 against m. The
4 2
slope of the graph should equal ]
k
Show the final form of the equation and calculations.

Calculations
(show work)
Value of spring constant of spring in Table. 1.1 ____________________(units)

(i) If the data in sections (g) and (h) above were for the same spring. Compute the percent difference for the
values of the spring constants obtained in each section.

19
QUESTIONS

1. Read the measurements on the rulers in ● Fig. 1.9. What is the difference between the three readings?

2. Were the three measurements of the block in part 2. (b) on page 16 all done with the same instrument? Explain.

3. Referring to the dart analogy in Fig. 1.3 on page 4, draw a dart grouping that would represent poor precision but
good accuracy with an average value.

4. Do percent error and percent difference give indications of accuracy or precision? Explain.

5. Suppose you were the first to measure the value of some physical constant experimentally. How would you
provide an estimate of the experimental uncertainty?

20