trials (P < 0.05) using the Watson two-sample test (for statistical
details, see Batschelet, 1981). A tutorial in R is provided in the Online
Materials, including an example data set drawn from the magnetoreception literature. The tutorial can be completed by the students in
advance of the lab activity if desired. The lab handout (Appendix)
has questions for the students to answer regarding the results of their
Part 7: Conclusion & Comprehension
Following statistical analyses, students should answer the “
Conclusion and Comprehension” questions in their lab handout, in which
we ask students to talk to each other about the observations of pill
bug behavior that they made during the lab, brainstorm what confounding variables may have resulted in different results between different groups, and suggest further experiments that they could
perform. It may also be useful for the groups to combine their data
and calculate the statistics for the class as a whole using the mean
direction for each individual pill bug to avoid issues of pseudoreplication (see below). Lastly, students brainstorm ways in which an
understanding of the magnetic sense might be useful to society, for
example in developing advanced navigation technologies.
At the end of the activity, students should be able to
• articulate a basic understanding of the proposed mechanisms of
magnetoreception and the ways in which animals use a magnetic sense,
• assess whether the control and pulse groups were oriented,
• determine whether the control and pulsed groups were different from each other,
• suggest other cues that may be influencing the pill bug’s orientation, and
• outline other orientation experiments to test other possible cues.
The assessment can take place through a variety of assignments.
The simplest is to have students complete the handout (Appendix),
but other techniques such as oral presentations, posters, and journal entries submitted to the instructor can be included.
Results & Discussion
We present a set of experimental results obtained from six groups
(16 students) in a 100-level undergraduate sensory biology course
in which the students were either advanced high school students
or college freshmen. The mean current measured from the pulse
magnetizers was 3.5 A, which corresponded with a mean magnetic
field strength of ~3.0 × 10−3 T (3 mT). To put this number in perspective, this field strength is approximately 50–100× greater than
Earth’s magnetic field (0.03–0.06 mT) and approaching that of an
ordinary refrigerator magnet (5 mT). Five of the six groups found
no evidence of orientation in their pill bug both before and after
the magnetic pulse, and the mean directions did not differ. However, one group found significant orientation in their control trials,
which also differed from their pulse-magnetized trials (Figure 4).
This activity thus illustrates the difficult nature of performing behavior experiments and should motivate discussion between instructor
and students regarding potential confounding cues and methods
to control for them (see below).
Confounding Factors, Pseudoreplication &
After data collection, the students engaged in lively discussion based
on the questions provided in the “Conclusion and Comprehension”
section of the handout (Appendix). They proposed that confounding
factors such as magnetic anomalies in the room due to the building’s
metal structure and electronic equipment may have affected the pill
bugs’ orientation. Other examples of confounding variables not controlled included hunger level (a reasonable prediction is that hungrier
individuals may be more inclined to searching behavior), “personality”
(in many species, some individuals are consistently bolder than others
and thus more willing to move), lighting, and temperature.
Because each group performs their data analysis independently,
teachers can lead the class in combining all groups’ data in a class data
set. This should begin with a short discussion about pseudoreplication. For example, the class could brainstorm what “
pseudoreplication” means and how it might apply to this activity. Teachers must
emphasize that, when the class combines their data, they will have
to account for pseudoreplication. In particular, if each group has performed 10 trials with their pill bug, they will not submit 10 different
values to the class data set, but rather only one: the mean angle. To
illustrate how failure to account for pseudoreplication can change
results, teachers could have students submit (a) all of their data and
Figure 4. Example orientation data from a single pill bug
before (control; black) and after (pulsed; gray) exposure to a
pulsed magnetic field. Some points overlap. Arrows indicate
mean direction within each group. The control trials were
significantly oriented (Rayleigh test, P = 7.0 × 10−4) and differed
from the pulsed trials (Watson test, 0.01 < P < 0.05).