Anth 473 / 573: Archaeological Measurement and Lab Analysis
Spring 2003
Lab Exercise 1: Precision and Accuracy
Due Friday February 7 at the beginning of lab
As we discussed in class, the concepts of precision and accuracy are fundamental to evaluating any form of
measurement. These are complementary
concepts that describe different aspects of reliability
– that is, the ways in which repeated measurements relate to each other. Precision
refers to the closeness of repeated measurements to each other, while accuracy refers to the closeness of
measurements to a “true value”.
In this exercise, you will use the concepts of precision and
accuracy to evaluate estimates (measurements) of attributes of flakes,
ceramics, and soils. We will be
creating several different kinds of estimates and you will compare your
estimates to those of others. When the exercise is complete, we will have a
series of estimates of attributes that reflect:
1.
multiple measurements of an attribute taken with the same
instrument
2.
multiple measurements of an attribute taken with different
instruments
3.
multiple measurements of different attributes taken with
different instruments
4.
all of the measurements by the students, the TA and the
professor
The objects from which some attributes will be measured are
located at five stations in the lab.
You will make your measurements individually and record those
measurements on data sheets that we will provide. Scott will collect all the data sheets and will make tables that
will be posted on the class web site Friday afternoon. You will need to download the tabulated data
in order to complete the lab exercise.
You will measure the following attributes:
- Color. You will measure the dry and moist colors of a soil sample
using a Munsell soil color chart. This involves comparing the sample to a
series of standardized color tiles until you have found the closest match.
Write the appropriate Munsell color codes on your data sheet.
- Soil Texture. You will estimate the texture of a soil
sample using a standard field method, often (colorfully) referred to as a
“spit and roll test”. The
procedure involves observing several attributes of the sample when dry and
when moist in order to estimate the proportions of sand, silt and clay
sized particles. Scott will
demonstrate the method in lab.
Record your texture estimates on the data sheet. We will compare our estimates to the
texture Scott determined using a settling tube method.
- Rim arc and estimated orifice diameter. You will measure the degrees of arc of
two potsherds using a paper template.
Using the same template and sherds, you will also estimate the
orifice diameter of the pots they came from. Record your measurements on the data sheet.
- Weight. You will measure the weight of a potsherd using two instruments:
a triple-beam balance and an analytical electronic balance. Record triple-beam measurements to the
nearest tenth of a gram (0.1 g), and electronic measurements to the
nearest thousandth of a gram (0.001 g).
Weigh the sherd three times on each scale, and record your
measurements on the data sheet. Do not take your measurements at the
same time. Weigh the sherd on each
scale, then move to another station before returning to weigh the sherd
for the second time; repeat the process for the third measurement.
- Length and Thickness. Measure the length and thickness of an
obsidian flake and a potsherd using both a caliper and a ruler. Record caliper measures to the nearest
tenth of a millimeter (0.1 mm) and ruler measurements to the nearest
millimeter. Again, take three
measurements of each attribute on each object and, as above do not take all measurements at the
same time.
Written assignment:
Use the standard lab report format as much as possible to
answer the following questions. Please
type your answers and attach a copy of the data. To answer the questions, it may be helpful to construct
histograms or calculate means and standard deviations. For example, the latter will work well for
describing variation in estimates of length, width and weight. Any calculations
or other data manipulation you perform should also be included, either in the
report itself or attached with the data.
- For
instances where you measured the same attribute using more than one
measurement instrument (such as length, thickness, and weight), do you see
a difference in the degree of precision
represented by different instruments?
Which of the instruments is most precise for a particular
estimate? Which is least
precise? What might account for
the differences (or lack of difference) in precision? Include all instruments in your
discussion.
- Do you
note any difference in accuracy
for these different instruments?
If there were differences in accuracy, could you determine which is
more accurate? Why or why not? Include all instruments in your
discussion.
- In
measuring the thickness and length of the obsidian flake and potsherd, you
used two instruments (caliper and ruler).
For either instrument, does precision vary depending which
dimension is being measured? In other
words, does either instrument have a different degree of precision when
used to measure length than when used to measure thickness? Why or why not? Does precision vary depending on which
object is measured? If so, what
accounts for this variation?
- How
many different Munsell color values did the class identify for the soil
sample? What factors may account
for the assignment of different color values to the same sample? If more than one value each for moist
and dry was determined, what does this suggest about the precision of the
Munsell chart as a tool for estimating color? Are there different levels of precision in estimating the
color of moist vs. dry samples? If
so, what might account for this?
Can you assess the accuracy of the Munsell charts with this data
set? Comment briefly on the usefulness of Munsell charts – does their
level of reliability affect how they should be used? How could you minimize reliability
problems if you used Munsell charts in your research?
- Did
the class generate different estimates of texture for the soil
samples? If so, why? What does this suggest about the
precision of standard field textural determinations? Can you assess the accuracy of this
method? How could you minimize
reliability problems if you use soil textural determinations in your own
research?
- Looking
at the orifice diameter estimates made by the entire class, are there
differences in precision between estimates made using the large sherd and
the small sherd? If so, what might
account for these differences? How
many sources of bias and/ or measurement error can you identify that are
associated with this method of measurement? How might you minimize these if you used this method in your
own research?
N.B. These
questions are guidelines for your discussion, and you should try to answer
each. DO NOT, however, turn in a list
of questions and answers – DO your best to include the answers to the questions
in a discussion built around the general lab writeup structure.