Let’s begin our last lesson on this topic by reviewing.
We have discussed the beginning of the process of science is with observations. The early scientists, the early sea captains, all began here.
Ocean currents, planetary motion, and magnetism were all observations made by these ancients.
This followed with logical explanations. Great thought went into this, but there would eventually be an explanation that was logically sound to explain the observations they saw. And, for a while, this was science. If it made sense… it was so. In many cases their logic was wrong. But in some, such as Copernicus’s concept of the sun being the center of the system, they were correct.
Eventually they would come up with the idea of testing these logically explanations. William Gilbert and Galileo Galilei were two of the first to add this step to the process – an experiment.
And then there was the important role of technology. To test these logical explanations, you need the technology to do so. So, science became – Science and Engineering. With such things as the Bunsen burner, the telescope, the microscope, and even new technologies for navigating a ship across the ocean, science advanced.
In this lesson we are going to look at the process of testing / experimentation closer. As science advanced it also became evident that the process of testing a logical explanation needed to advance as well. Could you design an experiment that would support a logical explanation from the start? Yes… in a sense – cheat. For example, let’s say that it is logical that boys are taller than girls in high school. To test this explanation, you would measure the height of boys and girls. You decide to select the boys’ basketball team and the girls’ gymnastic team as your test subjects. You have stacked the cards already. Your explanation is already supported but not because boys are taller, but because you selected the boys that would support your statement. And certainly, some scientists have done this. No one wants to be wrong. But other scientists could see through it very easily and challenge the results – the beginning of the peer review system, which is VERY important in this process.
Then there is the question of how many boys you should measure. There are only five boys on the basketball team. If you include the second string, ten. Is this a sufficient number? Another example to understand this point – let’s say I want to know how well students in your school are doing in English. So, I select five students (n=5). Their average score on the final exam was a 75… so, the average English student in your school is making a 75. Right away, with the peer review system, you can see this is wrong. It is right and wrong at the same time. Yes, the average is a 75 but you ONLY tested five students! To teach this point in a course I took they said that in the 1930s 33% of the girls attending John Hopkins University dated their professors. There were three girls at John Hopkins at this time, and one was dating her professor. So, it was correct but…
A study published in recent years shows this concept. The study stated that a significant number of children receiving the booster vaccinations required to enter public school were developing serious symptoms and health issues. This terrified the general public and many families chose NOT to have their kids vaccinated. This caused an uproar because the school systems were not allowing unvaccinated children to attend school, and there were law suites. However, the original study only looked at 12 families who had been vaccinated… (n=12). Was this enough? No, it wasn’t. The number tested was broadcasted to the media, but it was too late, people had made up their minds, and the issue continues to this day. Folks ignored the fact that only 12 families were included in the study. This certainly was not enough to draw any real conclusions.
So, how many students in your school should have been “tested” for their English grades? We agree five is not enough, but how many? Well, if you want the truth you would collect the final exam grades from ALL of the students. This would be a stronger statement and “better” science. But what if I want to know the average length of red snapper in the Gulf of Mexico? Can I measure EVERY snapper in the Gulf? Of course not… so how many do you measure? It has been said as many as time and money will allow. But in your experimental design you have a target – maybe 1000 snapper. Again, if you can measure more – do so.
AND if you are going to take a subsample of a population to test in your experiment, how to do you avoid experimental bias (cheating) as you saw in the example of are boys taller than girls? You would not want to only measure what snapper the fishing captains brought in; they are restricted in the size. Let’s look at the who is taller question. We understand that the BEST method would be to measure the height of every boy and girl in your school – not just the boys basketball team and the girl’s gymnastics team. But maybe you have SO many students in your school this in not possible – probably could be – but let’s say I am looking at ALL of the high school boys and girls in the district. I am not going to measure all of them. So, how do you decide WHO to measure?
There are two basic methods for selecting who – random and systematic. In a random sample you would literally select them randomly. Go to the computer and randomly select 100 students. There are programs on the internet, and in graphing calculators, that will randomly select 100 numbers. Do this, for example the first five numbers of your list of 100 could be 5, 16, 37, 89, 91. I go to the computer and find student number 5, 16, 37, 89, 91. I would continue to do this until I had 100 boys and 100 girls. Now measure them. In a systematic sample it would be something like every third name, or every fifth name until I had my group of 100 each. Now we are ready. There are computer programs that do this for scientists and I have honestly seen one who threw darts at a chart of Pensacola Bay trying to select sample sites for a field experiment (random) – so that is still done 😊
Then there is the question of experimental bias (cheating) due to outside factors. Let’s try and tie all of this together with an example experiment.
Observation – The common pinfish is found everywhere. Fishermen catch them in the Sound, the Gulf, all of the bayous, and even in the upper reaches of the bay near the rivers. However, they are noticeably absent around the Mobile Bay area…. Why?
Logical explanation (hypothesis) – the salinity is lower. Mobile Bay drains 1/5th of the rivers in the eastern United States and there is a lot of freshwater there. It is low enough that pinfish are not found there.
Test (experiment) – What is the salinity tolerance of the common pinfish – how low can they go?
Here we introduce the concept of experimental variables. The independent variable is the one you are testing – in this case, the salinity of the water. We could set up a series of tanks with different salinities. Freshwater is 0‰ and the Gulf is around 35‰. We know they can live in the Gulf; it is how low can they go we want to know. So, we can set up tanks at 35, 30, 25, 20, 15, 10, 5, and 0. How will you know it is too low for them? In decades past they would introduce the fish into these tanks and see if they survived. Maybe we would say if 50% of the fish in this tank survived – then it is not too low – if 51% died it is. This would have to discussed and decided in the experimental planning phase of the project.
This is a good place to stop and discuss experimental ethics.
The experiment described above was common for many years but then there was discussion about cruelty to animals. Is it right to kill animals during experiments in order to answer some question? Again, for many years, the answer was yes. Today the answer is no. It is understood that some animals are going to die, and there is nothing you can do about that. You could set up an aquarium at home right now, with NO intention of killing the fish – but some fish die anyway. This is expected, and tolerated, in scientific experiments as well. But they shy away from experiments with the intent to kill. All modern experiments are designed and must be approved by a review board for good design and for experimental ethics. There is a review board for your science fair projects as well!
In an experiment like this, many will say the end point of the experiment is Loss of Equilibrium (LOE) or something like this. So, we would lower the salinity in the tanks until the fish begin to behave differently – at THAT is “too low”.
And this – the LOE point – would be our dependent variable.
The dependent variable depends on the independent variable. So, the LOE of pinfish DEPENDS on the salinity… get it?
But are there other variables that could cause LOE in pinfish other than salinity? Sure there are. The water temperature, the amount of dissolved oxygen in the water, the pH, are they eating enough? Are they eating the right thing? Are the lights too bright? Could the results vary due to whether the fish was a boy or a girl? Could they vary with size or age of the fish? Any and all of these COULD affect the results of the experiment and give you false information. Yes, the pinfish reached LOE in tank #4 – 20‰ – but it was NOT the salinity that caused the LOE, it was something else. So, you could not publish that pinfish cannot live in water less than 20‰, it would not be true.
So, how do you address this issue?
All variables that could affect the results OTHER than your independent variable are called constants. They are called this because you want to KEEP THEM CONSTANT throughout the experiment. We would hit the library and find out what temperature, dissolved oxygen, and pH pinfish like it – and KEEP THOSE CONSTANT in our experiment. EVERYTHING would be constant except the salinity. We would use the same size tanks, same aerators, same type and amount of sand, we would feed each tank the same food at the same amount at the same time each day. We would use only pinfish of the same size and age class and have a 50/50 representation of male and females. You get the idea?
The idea of a control in an experiment is no treatment. If I am increasing, or decreasing, the independent variable in the experiment, in one tank I am not – this is the control tank. So, we could set up a series of tanks, all at 35‰, and begin to lower the salinity in each tank by adding freshwater (a specific amount at the same time each day to each tank – constant) – EXCEPT ONE… There would be one tank I would not change the salinity in. If those fish also show LOE, it was not the salinity but something else in the experimental design that caused it.
So, I think we’ve got it. NOW back to another question we brought up a while back. How many pinfish do we want to test? Let’s say we set up an array of tanks – all the same size, etc. We will do five control tanks – all remaining at 35‰, five that we will lower to 30, five we will lower to 25, etc. That would be seven sets of five tanks – 35 tanks. We will place 10 pinfish in each, all between 2-3”, so that is 350 pinfish. What do you think? Are we good to go?
So, we run the experiment and find that most reached LOE at 10‰ (making this up by the way). Now we have (n=350) LOE – 10‰. Feel good about this? Honestly, many scientists would run the series again increasing (n=) to (n=700) and do this until (n=1000 or more – however much money and time they have). Then publish a paper. The paper would be reviewed by scientists BEFORE publication and once published it would be reviewed again. Maybe another scientist in another lab would repeat the experiment until they reached (n=1000) and you see how it works – you are getting closer and closer to the truth. Over time we would have a better idea of how low pinfish can survive and then it is time to begin monitoring salinity in, and around, Mobile Bay.
THIS WAS A LONG LESSON – and honestly, there is MUCH more to add to the process of science. But I hope you get the idea of how it works. Let’s do an activity.
The activity for this lesson will be a mental one. We are going to design an experiment. We are going to try and incorporate everything we discussed in this lesson.
How many are you going to test?
What are your independent and dependent variables? How are you going to measure these?
What constants should you consider and keep constant? How are you going to measure these?
How many trials are you going to do?
Here is your experimental question…
Which of these vaccines (A, B, or C) are effective against the COVID-19 virus…?
Maybe you see better understand what scientists are up against right now.
To experiment and run multiple trials there are a few things young kids can do…
1) How many blocks can you stack before they fall over? Stack them but repeat the experiment several times and see if your answer changes
2) How many glasses of water does it take to fill a pitcher?
3) What is the temperature inside your refrigerator over the course of a day?
That is it for the Nature of Science. As you can see, you could spend the entire semester on this one topic alone – but there is more science to learn. If you are conducting experiments at home, I encourage to use this process as much as possible. It will, eventually, become second nature to you. I would always run ideas past others to see potential flaws in your design – the more eyes the better. And remember, ALWAYS, remember – experimental ethics.
Next up – October – THE WATER PLANET