What is Science?

Descriptive Science and Experimental Science

Scientists may be concerned with describing natural phenomena and they may limit their investigation to describing nature. However, they often wish to explain natural phenomena by a logical process that involves creating and testing hypotheses.

Experimental Science

The "Scientific Method" is a systematic approach to advancing knowledge. There are differences among scientists and among disciplines concerning the details of the method.  Many of the steps involved in a typical scientific investigation are discussed below using research published on the peppered moth (Biston betularia).

There are two forms of the peppered moth (Biston betularia) in England- a dark-colored form and a light form. In the early 1800's, the population was almost entirely the dark form.  During the 1800’s the dark form increased in frequency in urban areas. The change occurred at a time when increased pollution from coal fired factories darkened the trees. By 1895, 98% of the moths in Manchester were the dark form.

A hypothesis is an explanation of an observation. 

In 1896, J. W. Tutt hypothesized that bird predation was responsible for the increase in abundance of the dark form of the moth.  He reasoned that birds had difficulty seeing the dark form on the dark trees; the moths were camouflaged and survived better.  In clean areas, the trees were covered with lichens, making the pale form more difficult for birds to see.  In the early 1800s, the trees were light and the light form were more difficult to see. 

A hypothesis must be testable, and if false, it must be falsifiable.

An experiment is a test of the hypothesis.

Experiments are useful in disproving hypotheses. Hypotheses are not proved. This idea will be discussed later.

To test the bird predation hypothesis, H.B.D. Kettlewell  released moths of each type and then measured the number that were later recaptured. The experiment was performed in a polluted area in Birmingham, England and in a clean area in Dorset. The moths were marked with a dot of paint so that he could identify them after they were released and then recaptured.

The bird predation hypothesis (above) predicts that the light form should survive better in the clean area and the dark form should survive better in the polluted area. An estimate of survival can be obtained by using the number of moths recaptured. Obviously, moths that were taken by predators were not recaptured; only survivors were recaptured.

Results (Data)

The data are often numbers (measurements, counts, etc.).

The data below agree with the hypothesis. More light colored moths were recaptured in the clean area and more dark moths were recaptured in the polluted area.

	Light moths recaptured	Dark moths recaptured
Clean	13.7%	4.7%
Polluted	13%	27.5%

Statistics

Is it possible that the data above could have been due to chance and not really to the effect of bird predation? Statistical analysis allows you to calculate the probability that the data could have come out as they did if there really was no effect of predation.

If the probability that the data could be due to random chance is less than 0.05, we conclude that the difference is real; it is not due to chance. If the probability is greater than 0.05, we conclude that the difference is due to chance.

In the example above, a statistical analysis revealed that it is very unlikely that data could have produced the observed results due to chance.

Conclusion

You must accept or reject your hypothesis. In this case, the data suggest that the hypothesis should be accepted. Otherwise, it needs to be modified to account for the data.

Scientists generally do not regard information unless it has been published in a peer-reviewed or refereed journal. Typically, the scientist writes a report on the research and submits it to a journal editor. The editor distributes copies of the article to several peer reviewers that are recognized experts in the field. After reviewing the research, the peer reviewers recommend whether the paper should be published or not.

Junk science has not undergone this peer review process. Politicians, journalists, and others interested in swaying public opinion may mislead the public by presenting junk science as sound science or by presenting it along with sound science.

Suppose that a pet store owner observes that the mice in her colony grow slowly. She hypothesizes that they grow slowly because she does not give them enough food. To test this hypothesis, she uses two groups of mice. One group is given more food than they can eat while the other group receives the normal amount.

Suppose that all of the mice in her experiment grow slowly. Her hypothesis that she does not give them enough food must be incorrect. She rejects her hypothesis and decides to test an alternative hypothesis.

An alternative hypothesis is that the mice grow slowly because she does not give them enough water. She tests this hypothesis by giving the mice in one group more water than they can drink and giving the mice in another group the normal amount.

Suppose that mice in the group that received the extra water grow faster. She accepts her hypothesis that lack of water caused the poor growth. She has not proved her hypothesis. Perhaps it is not the water but instead, a mineral in the water that causes faster growth when given in higher amounts.

Science advances by rejecting false hypotheses. In the above experiment, she has not proved that it was the lack of water that caused poor growth. There may be other factors involved that she does not know about.

When a hypothesis is disproved it is rejected and an alternative hypothesis is accepted. However, you cannot be 100% sure that the hypothesis is false. You may reject a hypothesis that is true. Statistical techniques are used to reduce this type of error.

Suppose that a scientist wanted to test the effectiveness of a newly-developed growth hormone. She injected one group of mice with the hormone while a second group was not treated. After two weeks she weighed the mice and recorded the data below.

Group 1 - Hormone Weight (grams)	Group 2 – No Hormone Weight (grams)
12.5	8
13	8.5
12	8
12	8
13	7.5
14	10.5
13	7
10.5	8.5
9.5	6.5
11	7

The average weight of mice in group 1 is 12.05g; the average of group 2 was 7.95g. Is 7.95g significantly different than 12.05g? Could the results have come out this way by chance? The investigator needs to perform a statistical test to determine if the hormone is effective or if the mice in group 1 are heavier due to chance.

If there is really no difference between the two groups, what is the chance that the two groups above could weight as differently as they do? The difference between the two groups is 4.1 grams. If the probability is less than 0.05 (1 in 20), we conclude that the two groups really do differ. If the probability is greater than 0.05, we conclude that The two groups do not differ; any differences are due to random fluctuations.

Hypotheses that have been tested by different investigators numerous times and have not been disproved become theory.

Scientists often perform experiments by doing some sort of a manipulation. In the example above, one group of mice was given a hormone and the other group was not. The group of mice that was not given the hormone is referred to as a control group. A control serves for comparison. If the average weight of the control mice had been the same as that of the group given the hormone, it would be necessary to conclude that the hormone did not cause a change in growth.

Many experiments are controlled as described above. A controlled experiment is not one in which variables are controlled, it is usually not possible to control all of the factors that may affect the outcome of an experiment. In a controlled experiment, the control group experiences the same environment as the experimental (manipulated) group. If an environmental variable affects the experimental group, it also affects the control group equally. In the example above, the researcher wanted to determine if a hormone caused mice to grow larger. It is possible that the temperature of the room also affected mouse growth. The control mice were kept in the same room as the experimental (hormone) mice. If the mice receiving the hormone were heavier, the weight difference was not due to temperature because the control mice were kept at the same temperature.

The data in the graph below was obtained by measuring the heart rate of a runner while running at different speeds. As can be seen, when the runner ran faster, heart rate increased.

Heart rate and speed are both referred to as variables. Heart rate depends on speed, so it is called a dependent variable. Speed does not depend on heart rate; it is an independent variable.

Dependent variables are always plotted on the Y-axis (vertical axis). Independent variables are plotted on the X-axis (horizontal axis).

Modeling

Models are often used to describe natural phenomena. Models may be as simple as a diagram or a verbal explanation. Others may involve sophisticated mathematical equations and require computer analysis.

Mathematical models are often developed to explain complex phenomena. For example, a model could be developed to show nitrogen cycling in a montane (mountain) watershed. The model may have many variables and involve mathematical calculations. Complex models are often computerized (computer modeling). Example – climate models are often very sophisticated and require fast computers to analyze. 

Mathematical models enable researchers to make predictions. For example, we could ask what will happen to Earth's temperature if carbon dioxide in the atmosphere increases by 100 ppm? The solution can be found by entering this number into the model and performing the calculations. The model must reflect what really happens in nature. If the model is not realistic, the predictions will not be accurate.