1 The Scientific Method

The Scientific Method: Thinking like a scientist

It was once commonly believed that flies are generated spontaneously from rotting meat & garbage. If you try and argue with someone that flies hatch from eggs, not garbage, who is right? How do you know for sure? Have you ever actually watched flies emerge from eggs?

Scientific Reasoning

“Science” may be thought of as a body of knowledge that attempts to explain the natural world, AND the process by which we gain that knowledge. The body of knowledge consists of facts, observations, and well-supported hypotheses (more on these in a moment) that explain a set of observations. The process by which a scientist arrives at these bits of knowledge is the scientific method. This process is what separates scientific reasoning and conclusions, from other attempts at explanation or understanding. Scientific reasoning requires conjecture (attempts to explain observations) and confirmation (repeatable, consistently observed causes or relationships). In order to evaluate the validity of conjecture, scientists devise tests to confirm whether or not their generalizations are supported. In other words, scientists start with the “what if”, “how come”, and “what happens when” questions and attempt to reach solid answers through experimentation. Then they repeat these experiments to confirm the outcomes. Requiring confirmation is what separates scientific reasoning from assumption, expectation, or just a willingness to believe or disbelieve.

Deductive and Inductive Reasoning

Science involves both deductive reasoning and inductive reasoning. Deductive reasoning starts with a broad generalization (the “big picture”) and attempts to arrive at a specific conclusion. For example, “All humans are mortal” is a broad statement. “I am a human” is a specific observation. “Therefore, I am mortal”, is a specific conclusion. This is deductive reasoning.

Inductive reasoning uses specific observations to support a generalized conclusion. If you followed 16 sets of humans (each set is called a cohort) throughout their lives and found that they all died at the end, you could make the following observation. “Every human in my 16 sample groups died at the end of their lives.” This is a specific set of observations. From this set of observations, you could infer that “All of my humans were mortal.” This generalized conclusion can inductively be applied to all humans as, “All humans are mortal”. This is inductive reasoning.

By combining these two types of reasoning, scientists formulate well-supported conclusions about how things work (including life itself), what causes different observed phenomena, and what circumstances or environmental conditions change these phenomena.

The importance of observation – Step 1

Note that the whole process described above starts with an observation. There is the moment when you see something and the light bulb comes on in your brain, and you say, “Aha!” There are several qualities of a good observation. A good observation allows you to notice something you had not seen before, like a different result than you expected or a change in previously observed behavior, and makes you wonder why?, how?, etc. Second, the best observations form patterns. If you observe only a single event it could just be a fluke. If that same observation happens over and over, it is a pattern and it warrants an explanation. Finally, the observation shouldn’t have a simple, obvious answer. If you already know the answer, then who cares? Once a person goes from “Aha!” and wondering why to seeking an answer, this is the beginning of scientific reasoning.

• Observation: There sure are a lot of flies around this garbage dump. In fact, there are a lot of flies around all of these garbage dumps!

Stating a research question – Step 2

It is not science to simply notice something, to merely make an observation. Scientists ask questions. You observed a pattern, it’s interesting, and you don’t know what causes it. You have to ask how a certain event happened, where it happens the most often, why it happens in certain places, times or conditions, but not others, or when it is likely to happen again. In other words you must formulate a research question. The purpose of science is to answer these questions.

• Question: Why are there so many flies at the dump?

Formulating hypotheses – Step 3

You made an observation and stated a question that you don’t have an answer to. A hypothesis is a tentative answer to this question, and can be thought of as a working explanation (It is more than just an ‘educated guess’!). A good hypothesis answers the question by explaining one natural phenomenon in terms of another.

• Hypothesis 1 (H1): There are lots of flies at the dump because flies spontaneously generate from the garbage.

This hypothesis is a working explanation, a tentative answer to your research question. It also explains one phenomenon (lots of flies) based on another (spontaneous generation from garbage). BUT you may be thinking that another explanation is possible. You can propose an alternative hypothesis, and test them both to see if either is supported.

• Hypothesis 2 (H2): There are lots of flies at the dump because flies lay their eggs in the garbage there, and the eggs hatch new flies. [No way! Garbage turns into flies!]

Developing predictions – Step 4

If your hypothesis is plausible, then you can extrapolate and make predictions based on it. Predictions are statements about what may happen, based on your hypothesis, and allow you to evaluate your hypothesis. Predictions are stated in an “if, then, result” format. IF my hypothesis is correct, THEN a manipulation to these conditions will RESULT in a specific, measurable outcome. An alternate prediction would be, “IF my hypothesis is incorrect, THEN manipulating the conditions will NOT RESULT in any difference in measurable outcome. For example, let’s use hypothesis 1, that flies are spontaneously generated from garbage.

• Prediction 1 (P1): If flies spontaneously form from garbage, then garbage placed in a closed container (without flies) will produce flies.
• Prediction 2 (P2): If flies do not spontaneously form from garbage, then garbage placed in a closed container (without flies) will not produce any flies.

Notice that each statement predicts a certain outcome, and that the outcome can be directly measured.

Independent Variables (IV) and Dependent Variables (DV)

Generally speaking, when you are conducting a scientific experiment you are examining the effect of one ‘thing’ on another ‘thing.’ In the fly example, you can manipulate the conditions of the garbage and count how many flies are produced. The Independent Variable (IV) is the condition that you can manipulate (closed or open container). The Dependent Variable (DV) is the measured outcome in your experiment (number of flies produced). You can think of it this way; “The outcome (DV) depends on the manipulated condition (IV)”.

Testing your predictions & evaluating your hypotheses – Step 5

Without a test of your predictions, it doesn’t matter if flies form from garbage, hatch from eggs, or are all just hallucinations from sniffing garbage fumes. You can believe whatever you want, but you will have no basis of support for your belief. To answer your research question, you need to devise a test that will produce results that support one hypothesis, but refute another. You accomplish this by testing your predictions through experiments. Ultimately, you are attempting to refute all of your hypotheses.

Devise a test – Step 6.

Who is right? Where do the flies at the dump come from? The steps that you follow to test your predictions is your experiment, and a good experiment has some specific parts.

Treatments

In the design of an experiment, treatments are applied to the units you are testing. Treatments are the manipulation of the Independent Variable. In our fly example you may have two treatments for garbage conditions (covered or exposed to flies). You test how many flies are produced (DV) in each treatment, and you can conclude which prediction (if any) was correct.

Controls

In order to properly test your predictions, you need to control for additional causes of change you might see, other explanatory variables in your proposed cause/effect relationship. When scientists speak of experimental controls, we are referring to limiting the influence of non-manipulated variables during our experiment. This can be accomplished by keeping all conditions the same between treatments except the single IV we are evaluating. In the case of our garbage experiment, we would want to make sure that the amount of garbage and its contents are the same between treatments, that the temperature and light exposure are the same, and that all extraneous variables that we can think of are the same between our treatments. In other words, that we have controlled our variables. In this way, we can be reasonably sure that whatever difference in flies (DV) that we measure between treatments is actually a result of the IV we are manipulating (covered or exposed).

The term control treatment may also refer to one particular set of conditions within an experiment. For example, if we wanted to determine the effect of fertilizer on plant growth, we could set up a series of treatments with one being the natural conditions in which our experimental plant normally grows. Other treatments would be identical in every way, except that we would add fertilizer to them. The “natural conditions” treatment would be called our control treatment. Once again, by limiting the differences between our treatment conditions, we can be reasonable sure that whatever outcome differences we find (plant growth) are due to our Independent Variable (fertilizer application).

Replication

If you do your experiment one time with each treatment, how certain are you of your results? What if something happened in that one trial that changed the “normal” outcome. Let’s say you do the experiment 100 times and get the same result 95 times. How certain are you now? Replication simply means repeating the experiment so you can assess your level of certainty and rule out random events (known as “experimental error”) that may lead to a false conclusion. In the fly example, you may decide to have 50 replicates of each treatment. Think of each replicate as a “mini-experiment”. In other words, you carry out your procedure 100 times (50 replicates x 2 treatments) in the course of your experiment.

Now, what do you do with all that data?

Analyzing results – Step 7

If you carry out your experiment once, your data analysis is pretty simple. Compare the outcome (DV) between treatments (IV) and find out if there is a difference. However, one replicate is not a valid experiment. If you carry out your experiment 50 times, you will probably get 50 slightly (or not so slightly) different measured outcomes (DV). How do you reconcile all these different measurements? Which value is “correct”? Answer: All of them! You use all of the data and find the mean (average) of all replicates for each treatment. For example, in your experiment you counted the number of flies produced in garbage that was in a covered container. You take each of these counts (all 50 of them) and calculate the mathematical average. This average becomes the result the “covered” treatment. You do the same for the “exposed” treatment, and compare the mean values of flies produced in each treatment to determine which of your predictions was confirmed.

Forming Conclusions – Step 8

Now that you have tested your predictions, you need to form a conclusion. The conclusion answers your original research question (or attempts to, at least). In the fly example, if after 1 week your experiment resulted in a mean of 27 flies in the “covered” treatment and a mean of 457 flies in the “exposed” treatment, you might conclude that your data (the difference in mean fly numbers) supports Hypothesis 2 (flies→garbage→eggs→flies). But where did those 27 flies come from in the “covered” treatment? You might conclude that those 27 flies are evidence of support for Hypothesis 1 (garbage→flies)! Now what?

Looks like you’ve got more questions to answer. Formulate new hypotheses (or modify the old ones), make new predictions, devise new experiments with new variables (or variable combinations) and try again!

Termites and the Scientific Method

In lab this week, you will follow the scientific method to study termites. During lab, you will record your work in your printed workbook as you complete the steps below.

Procedure

1. Draw a circle on a piece of paper by tracing the base of a large petri dish. Cut the circle out.
2. Use the pens you have been provided to draw 2-3 large, closed shapes on the paper and place the paper inside the petri dish. Each shape should occupy ~25% of the total area of the paper.
3. Place two worker termites within each shape. Watch them mill about for at least two minutes.
4. Briefly list your observations. Do you notice any particular behavior? What do you observe happening? Is the same behavior happening for all termites? All drawings? All shapes? Is there a pattern of behavior? Think about these observations, talk with your group, and come up with a research question regarding your observations. State the research question you want to answer.
5. What is a possible and reasonable answer to your research question? For example, why is this observed pattern of behavior happening? What particular factor is causing the behavior? Write a hypothesis for the termite’s behavior that meets these three requirements.
6. State another possible answer to your research question; an alternative hypothesis to explain your observations. Why else might this be happening?
7. State your predictions. If your hypothesis is correct, what will happen if the environment changes in some way? Choose one of your hypotheses and state a prediction that would support that hypothesis. Make sure you are predicting an outcome that can be measured and analyzed.
8. It’s not enough just to look for support for your hypotheses; you must also try to falsify them. What results would refute your hypothesis? State a prediction that would refute your hypothesis. Give some real thought to this, not just, “If it doesn’t happen, it’s not true.”
9. What are your independent variable (IV) and your dependent variable (DV) in these predictions? Remember, your IV is the variable that you are manipulating (e.g., color, shape, ink, etc.) and the DV is the outcome you are measuring (e.g., time spent on a line, speed, number of turns, etc.).
10. Devise an experiment. What can you do to test which of your predictions (if any) is correct? Discuss with your team and come up with a procedure (before you try it!) to test your predictions. Describe your experiment in simple steps, making sure to describe your treatments (IV manipulations) and what you will be measuring (DV).
11. Carry out your experiment with items available in the lab. Then, fill in the results table to show your results. Include the independent variables and the dependent variables you measured.