This document summarizes some of the most frequently made mistakes when lab
reports are written:
- Figures not labeled and/or not referred to in the text. Each Figure must
be label with a figure number and a figure caption. Do not include every
graph you collect during a measurement; only include those that support your
conclusions and those that illustrate the quality of the data and the process
used to obtain your results.
- Absence of error estimates. Each measurement you carry out is limited in
accuracy due to measurement errors. You will need to estimate your measurement
error and report your results with an accuracy that reflects the quality
of your measurement. Keep the following issues in mind:
- Use the correct number of significant figures. If you measure the width
of the tables in B&L 407 with a ruler that is divided in centimeters,
you might be able to measure the width of the table with an accuracy
of 0.5 cm. The results of such a measurement could be reported as
(55.5 cm +/- 0.5) cm. A result reported as (55.489874632 +/- 0.5)
cm would be incorrect since the number of significant figures used
to report the measured distance implies that the accuracy would be
0.000000001 cm, which is very much smaller than the error in your
measurement (note: if the error in your measurement is 0.5 cm, do
you expect to be able to tell the difference between 55.489874632
cm and 55.489874633 cm?).
- The easiest way to estimate your measurement error is to repeat your
measurement several times. The spread in the results reflects the accuracy
of your measurement. For example, if I repeat the measurement of the
width of a table 3 times I might get the following results: 55.8 cm,
55.1 cm, and 55.6 cm. The average distance is (55.8 + 55.1 + 55.6)/3
= 55.5 cm. The difference between the maximum and minimum values is 55.8
- 55.1 = 0.7 cm. A good estimate for your measurement error would be
half of this difference, which is 0.4 cm. To report the results of you
measurement you would specify (55.5 +/- 0.4) cm.
- The actual calculation of your errors needs to be carried out using
statistically correct procedures (see for example the Intro
to Error Analysis document and our text book).
- The error in a measurement is not the difference between the
measured value and the theoretical value. Measurement errors are independent
of the theory that might be used to describe the experiment.
- Absence of units. When the results of a measurement are listed, make sure
to include the units which are being used. A statement that the length of
an object is 5 is meaningless. In this course we will be using metric units
for most of our experiments:
- The length of an object will be specified in terms of meters (m).
- The mass of an object will be specified in terms of kilograms (kg).
- The time will be specified in terms of seconds (s).
- Absence of calibration data. In most experiments you will need to confirm
the calibration of your equipment. You will need to list the calibration data
you took (using a table is frequently the easiest way to summarize the calibration
data). A statement "we verified the calibration of our equipment"
is not sufficient.
- Absence of experimental data. In each experiment you will carry out a number
of measurements which will provide the data to be examined in your report.
Your report should include all the data that are used to obtain your conclusions.
For example, if you repeat the measurement of the acceleration of an object
5 times, you will need to list the outcome of each of these 5 measurements
rather than just the average of these data. The results of the individual
measurements will allow me to judge the accuracy of your measurement and whether
or not the assigned error of the average is reasonable. The data can be listed
in tabular or graphical form.
- The rejection of data that do not agree with the theory to be tested. The
result of each measurement is significant. No data can be rejected because
they do not agree with the predictions of a theoretical model. You can only
reject data when there are experimental reasons to do so (for example, the
ruler was not aligned properly during the calibration measurement, we forgot
to record the mass of the cart used in the experiment, etc.).
- Conclusions are not supported by the data. In many cases, students make
the assumption that the theory to be tested is correct. Any observed differences
between theory and data are attributed to errors. "I must have made a
mistake in my analysis" is a conclusion that is frequently seen in lab
reports. Note that in many cases, the theory is developed for ideal conditions
(such as the absence of friction) and differences between theoretical predictions
and observations are due to normal measurement errors, and the presence of
effects like friction. Please note that it makes only sense to compare data
with predictions when the measurement errors have been estimated correctly.
A difference between experiment and theory that falls within the error bars
is nothing to be concerned about.
Last updated on Friday, September 16, 2005 10:19
