Computer Science in Science PD: Using Computer Models in Science - Discussion


Experiments in class are wonderful, but there are certainly issues that have to be addressed. Experiments in class can be time consuming, expensive, and difficult to gather needed materials. Building a computational model would most likely eliminate most of these issues. I can see time issues being eliminated, for the most part. Human error isn’t as much of an issue with running computational models.


A lab inquiry and computational models follow the same basic steps. However, in a lab setting, the time given to perform the inquiry is over a shorter period of time. Computer modeling is faster, providing more data which can help the students to interpret their data and revise a hypothesis if necessary.


All valid points. I can see computer modeling will be a productive tool.*


I agree. I think that blended lab setting that incorporates both traditional labs and computer modeling would offer the greatest benefit to students.


My main focus in teaching experimental design to 6th graders, is to teach them to identify independent and dependent variables. This is a very difficult concept to understand when we are simply reading scenarios from a paper, but is essential with both a “hands on” experiment or a computer modeled experiment. Learning about IV/DV using traditional hands on experiments is costly and time consuming however. I think that computer models would allow students to practice this in a much more authentic way, allowing the the learning process to be more of a discovery than an idea that is force fed. Students could create their own experiment, with the teacher acting only as a facilitator and not having to approve steps or explain errors, and thereby discover what the independent and dependent variables are (or end up being) in their models. By modifying the model a number of different ways, they could discover the relationship between IV and DV and end with a deeper understanding of the need for a controlled experiment. In this way, computer models are not only beneficial for modeling experiments, but also a beneficial tool for the learning process.


Computational models allow students to complete a design they have made without human errors and biases is the one way how is differs from the experimental design. Both design and model allows students to test out the hypothesis made, but the model method allows for students to see the results much quicker and with more preciseness.


It is very similar the scientific processes are the same. What my classroom is lacking that computational models have is the ability to always complete a lab in time. With the quickness and accuracy the program offers there could be time left for further exploration of students who may move quicker than others.


There are many limitations to the kind or the extend an experiment can be conducted by students due to time constraint. Conducting experiencments in computational science will allow students to run the simulation/experiment multiple times since the simulation can speed up the time. Since time is less of a factor, students can also run the simulation trying out many variables and parameters to test their hypothesis.


In my class, we follow the scientific method for experiments. Both computational models and traditional experimental design include creating a question based on an observation, conducting research, forming a hypothesis, collecting and analyzing data, and drawing conclusions. However, in computational models the hypothesis may be formed after running a simulation, rather than prior to conducting an experiment or study. Computational models can allow a researcher to test virtually any variable–no matter how dangerous it might be in reality. The experimental design that I teach includes a section on feasibility (length of time, cost, equipment availability, safety, etc.) I have to deny approval to many students each year for their science fair experiment due to a lack of feasibility. Computational simulations would allow students to research almost any interest.


Agreed. Nothing is more frustrating for both a teacher and students than having an experiment fail due to unforeseen variables or running out of time/resources. Computational models would allow for running centuries worth of repeated trials in seconds. If a variable was overlooked, it could be easily changed and a new simulation run–for no cost, and probably within the same day.


Good point. In my classroom we are often lacking on time to complete a deep discussion of conclusion or analysis of results. Computer models could also free up more time for this type of post-lab discussion.


We use pHET as well - the students enjoy the ability to make choices regarding the variables and seeing how those choices affect the outcome.


When comparing experiments conducted in class with computational models the similarities are they are both are designed have controls and to reduce interfering variables. The ease of exact replication is a big advantage to computer models, however sometimes models have too many variables and the students manipulate all of them with the result that they are uncertain as to which variable produced the effect. In either case students need direction and oversight.


Regarding conducting experiments in class, part of the procedure is identifying variables in the experiment. This includes things we can and cannot control, things that may or may not be considered independent/dependent variables, or things that we realize could change the outcome and therefore need to be controlled. Some of these variables are things that are involved in physically collecting data like parallax and errors related to physically using a stopwatch/measuring carefully. The vocabulary used for the computational models sounds different, but the idea is the same. However, with computational models, once you add in the code you don’t have to worry about whether you’re looking from the same angle as your partner or if you pushed the car when you should have just released it…


I appreciate the speed with which they can see their results. Also, if things need to be changed, it’s a code change, not a whole new set of painstaking data collection!


The greatest advantage to students learning using computational models is that it gives them the ability to learn from the outcomes, change variables, hypothesis, etc. based on what they learned and repeat the experiment over and over. Each time they gain new insight which enhances their learning. It gives students the opportunity to take control and responsibility for their learning. With lab experiments they may get to repeat the experiment only 1 or 2 times and with the computer models repeating the experiments multiple times becomes possible.


I think this is a great tool. I fight time constraints on most every experiment we do. By the time we get everything situated, while leaving time to clean-up, there is definitely not time to effectively perform an experiment.


that’s a very good point - I am anxious to incorporate this in my classroom.


More similarities here than differences for the level of science I teach and the controlled experiment content we try to accomplish each year. Common ideas include; developing our question, the idea of identifying control variables and changing one independent variable at a time is stressed heavily at our level. Also the idea of running multiple trials to gather a sufficient amount of data is similar to those run in computational models. One idea in the video that was interesting that is usually not part of the way we conduct experimentation is that parameters could be changed in a computational model and then a hypothesis could come. Often our hypothesis is derived from previous experiments/experiences.


Both experimental design and computational models require controlled variables and hypotheses.