Road Kill

A. Summary

1. The System

The RoadKill Project monitors the number of animals killed by motor vehicles. The number of roadkill is potentially influenced by amount of traffic, type of road, speed limit, weather conditions, town population, as well as by migratory patterns of animals and local availability of habitat. For more information see Introduction.

2. Learning Goals

The RoadKill Project has no curriculum per se, so it offers no detailed learning goals. The RoadKill project is designed to involve students and teachers with scientific monitoring of an environmental parameter using telecommunications and to increase participant awareness of motor vehicle hazards with wildlife. The project does offer a project protocol that gives an overview of different levels of participation, procedures for data transfer and retrieval, possible activities, and student objectives.

Students are expected to:

...compare the data from participating schools.
...understand reciprocal effects between humans and wildlife.
...predict which type of animal will be most and least killed by motor vehicles.
...compare the relationship between geography and topography and the types and number of roadkill in different locales and states.
...understand the habitats and ecological importance of small and large mammals, reptiles, and birds.
...understand the migratory patterns of different animals.
...recognize animal species in their country.
(cited from project protocol, http://earth.simmons.edu/roadkill/rk_protocol.html)
The last three goals show that the project is much more ambitious than just monitoring killed animals.

For more information see Introduction and data analysis in the curriculum.

3. Available data

To enter the data archive or to send data to the project director a password is required.
The Roadkill Project provides two different types of data:  “ Dr.Splatt’s grouped roadkill data”, and “ Year-round roadkill data”. Both data sets include some information about the location and the conditions of the observed road and the number of dead animals observed divided into over 80 species. The data sets differ in the periods over which the data are collected. The “ Year-round roadkill data” contains student observations made on any day over the entire year, while “ Dr. Splatt’ sgrouped roadkill data” contains data collected during an eight-week period from March to May.
All data are collected by students. The database has quite a lot of errors (example), and comparability of the data is questionable. For example, one student tallies roadkill daily while another counts roadkill only once per week;  some roadkill erroneously are counted twice. Weather conditions as reported in the dataset also are suspect. If the students find a dead animal while the sun is shining, they do not know if there was rain falling when the animal was killed.

Comparability of data is rather questionable in other respects, too. Most of the schools did not collect data over the entire period of 8 weeks (the suggested time), and it is self evident that more roadkill will be seen in 8 weeks than in 4. The length of the road observed was not considered in the analyses either, but it would certainly affect the number of roadkill observed.

We recommend that classes discuss and agree on fixed rules of data collection before they begin their observations, to increase data accuracy and comparability. For more information see data and data archive .

4. Supports for data analysis

Software

The project recommends that for analysis, students use spreadsheet software, though no particular program is recommended.  Some data are provided in File Maker Pro format. This program, however is more a database and not a data analysis program.

Subject matter knowledge

The project does not provide subject matter, but merely lists links to" road kill web resources".

Data analytical knowledge, strategies

Although RoadKill provides no curriculum, it does offer some ideas about the data analytical knowledge that students need and also about which analytical strategy to use (see suggested classroom activities ( Using RoadKill Data and RoadKill: Graphing and Analysis of Data for Middle School Science and Math ). Students must filter the data they need from the large and complex data base before they can start to summarize, visualize, and interpret it. The units Using RoadKill Data and RoadKill: Graphing and Analysis of Data for Middle School Science and Math suggest  questions at the most basic level,  such as "what are the three most frequent road kills seen in your school?"  At the research level, students are asked to compare different distributions, such as frequency of animals killed on roads with different speed limits. However, there is no guidance on how the students can manage this important data analytical activity.

For more information see data analysis in the curriculum.

Subject matter questions to be answered by data analysis

Students are to find out which types of animals are killed in a certain region, and with which frequency. These numbers are considered as being affected by different wild lifes, road conditions, weather conditions and seasonal and daily changes of conditions that may help to explain the data the students have collected.

Exemplary data analyses, expected answers

The project does not offer any prototypical data analyses nor provide ideas about what kind of patterns students are likely to see in the data.

5. Our own example of data analysis

The first question we analyzed was "What are the three most frequent road kills seen in our school? In our state and area of the country? In the whole data base?". This question was suggested by the project authors. We reformulated this question in a more general way in that we look at the whole distribution and not only at the most frequent species. We used bar graphs and ordered bar graphs to answer the questions. The ordered bar graphs tend to form groups that we could use to classify the animals into frequency classes.

Afterwards we investigated the question: "What are the three most frequent road kills in each month?". We made bar charts for each month, which showed that for some species number of roadkill varied considerably while for other species the number remained relatively constant.  For us, this raised two new questions:

  1. How does the relative frequency: "roadkills of one species in one month/total roadkills of all species in that month" vary over the year?
  2. How does the relative frequency: "roadkills of one species in one month/total roadkills of that species in the whole year" vary.
Finally we investigated the project's data analysis questions on research level (see project protocol or introduction for levels of participation).
We used medians, means and boxplots to answer the following questions (results are given in parentheses):
  1. Does the number of roadkills depend on the road classification? (Nearly all values are 0; only the mean is influenced by some outliers. Therefore no predictions are possible.)
  2. Does the number of roadkills depend on the weather conditions? (Nearly all values are 0; only the mean is influenced by some outliers. Therefore no predictions are possible.)
  3. How does a speed limit effect the number of roadkills? (Fewer animals killed on average on roads with the lower speed limits.)
  4. Does the number of roadkill depend on the moonphase? (There is no dependence.)
In answering these questions, we had to confront problems in the data base. About 60% of the observations in the "Year-Round RoadKill data" did not include any killed animal, which means the students did not see any roadkill on those days and reported "no roadkill." and it is difficult to do comparisons with data containing so many zeros. In addition, data were often not collected under the same conditions so comparability is rather questionable (see above 3. available data).

For more information see exemplary data analysis.

6. Summary from the perspective of data analysis

In many ways, the Roadkill project is well suited to introducing students to an elementary use of data. They can learn to process, summarize, and graphically represent raw data.  In our opinion, the project's potential for student learning could be enhanced if the raw data were reviewed and edited before they were made available for downloading, in particular to eliminate many of the obvious errors. The activities that the project authors propose remain at a fairly elementary level and do not make full use of the data’s potential. In our analyses we intended to exploit the pot entail more fuly with slightly more advanced methods. The data available are not appropriate for an in-depth analyses intended to find out which external conditions make a difference, because the comparability is doubtful. For these reasons, the “ Year-Round RoadKill Data” do not seem useful. The analysis of data that a single school collected would allow more reliable conclusions.