The Scientific
Method Tutorial
The Scientific Method
Steps in the Scientific Method
There is a great deal of
variation in the specific techniques scientists use explore
the natural world. However, the following steps characterize the
majority of scientific investigations:
Step 1: Make observations
Step 2: Propose a hypothesis to explain observations
Step 3: Test the hypothesis with further observations or experiments
Step 4: Analyze data
Step 5: State conclusions about hypothesis based on data analysis
Each of these steps is explained
briefly below, and in more detail later in this section.
Step 1: Make observations
A scientific inquiry typically
starts with observations. Often, simple observations will trigger a question in
the researcher's mind.
Example: A biologist frequently sees monarch caterpillars feeding on milkweed plants, but rarely sees them feeding on other types of plants. She wonders if it is because the caterpillars prefer milkweed over other food choices.
Step 2: Propose a hypothesis
The researcher develops a
hypothesis (singular) or hypotheses (plural) to explain these observations. A
hypothesis is a tentative explanation of a phenomenon or observation(s) that
can be supported or falsified by further observations or experimentation.
Example: The researcher hypothesizes that monarch caterpillars prefer to feed on milkweed compared to other common plants. (Notice how the hypothesis is a statement, not a question as in step 1.)
Step 3: Test the hypothesis
The researcher makes further
observations and/or may design an experiment to test the hypothesis. An
experiment is a controlled situation created by a researcher to test the
validity of a hypothesis. Whether further observations or an experiment is used
to test the hypothesis will depend on the nature of the question and the practicality
of manipulating the factors involved.
Example: The researcher sets up an experiment in the lab in which a number of monarch caterpillars are given a choice between milkweed and a number of other common plants to feed on.
Step 4: Analyze data
The researcher summarizes
and analyzes the information, or data, generated by these further observations
or experiments.
Example: In her experiment, milkweed was chosen by caterpillars 9 times out of 10 over all other plant selections.
Step 5: State conclusions
The researcher interprets the
results of experiments or observations and forms conclusions about the meaning
of these results. These conclusions are generally expressed as probability
statements about their hypothesis.
Example: She concludes that when given a choice, 90 percent of monarch caterpillars prefer to feed on milkweed over other common plants.
Often, the results of one
scientific study will raise questions that may be addressed in subsequent
research. For example, the above study might lead the researcher to wonder why
monarchs seem to prefer to feed on milkweed, and she may plan additional
experiments to explore this question. For example, perhaps the
milkweed has higher nutritional value than other available plants.
The Scientific Method Flowchart
The steps in the scientific
method are presented visually in the following flow chart. The question raised
or the results obtained at each step directly determine how the next step will
proceed. Following the flow of the arrows, pass the cursor over each blue box.
An explanation and example of each step will appear. As you read the example
given at each step, see if you can predict what the next step will be.
Use the steps of the scientific method described above to solve a problem in
real life. Suppose you come home one evening and flick the light switch only to
find that the light doesn’t turn on. What is your hypothesis? How will you test
that hypothesis? Based on the result of this test, what are your conclusions?
Follow your instructor's directions for submitting your response.
The above flowchart illustrates
the logical sequence of conclusions and decisions in a typical scientific
study. There are some important points to note about this process:
1. The steps are clearly linked.
The steps in this process are
clearly linked. The hypothesis, formed as a potential explanation for the
initial observations, becomes the focus of the study. The hypothesis will
determine what further observations are needed or what type of experiment should
be done to test its validity. The conclusions of the experiment or further
observations will either be in agreement with or will
contradict the hypothesis. If the results are in agreement
with the hypothesis, this does not prove that the hypothesis is true! In
scientific terms, it "lends support" to the hypothesis, which will be
tested again and again under a variety of
circumstances before researchers accept it as a fairly
reliable description of reality.
2. The same steps are not
followed in all types of research.
The steps described above present
a generalized method followed in a many scientific
investigations. These steps are not carved in stone. The question the
researcher wishes to answer will influence the steps in the method and how they
will be carried out. For example, astronomers do not perform many experiments
as defined here. They tend to rely on observations to test theories. Biologists
and chemists have the ability to change conditions in
a test tube and then observe whether the outcome supports or invalidates their
starting hypothesis, while astronomers are not able to
change the path of Jupiter around the Sun and observe the outcome!
3. Collected observations may
lead to the development of theories.
When a large
number of observations and/or experimental results have been compiled,
and all are consistent with a generalized description of how some element of
nature operates, this description is called a theory. Theories are much broader
than hypotheses and are supported by a wide range of evidence. Theories are
important scientific tools. They provide a context for interpretation of new
observations and also suggest experiments to test
their own validity. Theories are discussed in more detail in another section.
The
Scientific Method in Detail
In the sections that follow, each
step in the scientific method is described in more detail.
Step 1: Observations
Observations in Science
An observation is some thing, event, or phenomenon that is noticed or observed.
Observations are listed as the first step in the scientific method because they
often provide a starting point, a source of questions a researcher may ask. For
example, the observation that leaves change color in the fall may lead a
researcher to ask why this is so, and to propose a hypothesis to explain this phenomena. In fact, observations also will provide the
key to answering the research question.
In science, observations form the
foundation of all hypotheses, experiments, and theories. In an experiment, the
researcher carefully plans what observations will be made and how they will be
recorded. To be accepted, scientific conclusions and theories must be supported
by all available observations. If new observations are made which seem to
contradict an established theory, that theory will be re-examined and may be
revised to explain the new facts. Observations are the nuts and bolts of
science that researchers use to piece together a better understanding of
nature.
Observations in science are made
in a way that can be precisely communicated to (and verified by) other
researchers. In many types of studies (especially in chemistry, physics, and
biology), quantitative observations are used. A quantitative
observation is one that is expressed and recorded as a quantity, using some
standard system of measurement. Quantities such as size, volume, weight, time,
distance, or a host of others may be measured in scientific studies.
Some observations that
researchers need to make may be difficult or impossible to quantify. Take the
example of color. Not all individuals perceive color in exactly
the same way. Even apart from limiting conditions such as
colorblindness, the way two people see and describe the color of a particular
flower, for example, will not be the same. Color, as perceived by the human
eye, is an example of a qualitative observation.
Qualitative observations note qualities associated with subjects or
samples that are not readily measured. Other examples of qualitative
observations might be descriptions of mating behaviors, human facial
expressions, or "yes/no" type of data, where some factor is present
or absent. Though the qualities of an object may be more difficult to describe
or measure than any quantities associated with it, every attempt is made to
minimize the effects of the subjective perceptions of the researcher in the
process. Some types of studies, such as those in the social and behavioral
sciences (which deal with highly variable human subjects), may rely heavily on
qualitative observations.
Question: Why are
observations important to science?
Limits of Observations
Because all observations rely to some
degree on the senses (eyes, ears, or steady hand) of the researcher, complete
objectivity is impossible. Our human perceptions are limited by the physical
abilities of our sense organs and are interpreted according to our
understanding of how the world works, which can be influenced by culture,
experience, or education. According to science education specialist, George F.
Kneller, "Surprising as it may seem, there is no fact that is not colored
by our preconceptions" ("A Method of Enquiry," from Science
and Its Ways of Knowing [Upper Saddle River: Prentice-Hall Inc., 1997],
15).
Observations made by a scientist
are also limited by the sensitivity of whatever equipment he is using. Research
findings will be limited at times by the available technology. For example,
Italian physicist and philosopher Galileo Galilei (1564–1642) was reportedly
the first person to observe the heavens with a telescope. Imagine how it must
have felt to him to see the heavens through this amazing new instrument! It
opened a window to the stars and planets and allowed new observations undreamed
of before.
In the centuries since Galileo, increasingly more powerful telescopes have been devised that dwarf the power of that first device. In the past few decades, we have marveled at images from deep space, courtesy of the Hubble Space Telescope, a large telescope that orbits Earth. Because of its view from outside the distorting effects of the atmosphere, the Hubble can look 50 times farther into space than the best earth-bound telescopes, and resolve details a tenth of the size.
Since
2021, we have received images from the James Webb telescope, which is 100 times
more powerful than the Hubble Space Telescope, allowing us to observe distant
galaxies with unprecedented clarity.
Although the amount of detail
observed by Galileo and today's astronomers is vastly different, the stars and
their relationships have not changed very much. Yet with each technological
advance, the level of detail of observation has been increased, and with it,
the power to answer more and more challenging
questions with greater precision.
Question: What are
some of the differences between a casual observation
and a 'scientific observation'?
Step 2: The Hypothesis
A hypothesis is a
statement created by the researcher as a potential explanation for an
observation or phenomena. The hypothesis converts the researcher's original
question into a statement that can be used to make predictions about what
should be observed if the hypothesis is true. For example, given the
hypothesis, "exposure to ultraviolet (UV) radiation increases the risk of
skin cancer," one would predict higher rates of skin cancer among people
with greater UV exposure. These predictions could be tested by comparing skin
cancer rates among individuals with varying amounts of UV exposure. Note how
the hypothesis itself determines what experiments or further observations
should be made to test its validity. Results of tests are then compared to
predictions from the hypothesis, and conclusions are stated in terms of whether or not the data supports the hypothesis. So the hypothesis serves a guide to
the full process of scientific inquiry.
The Qualities of a Good Hypothesis
- A
hypothesis must be testable or provide predictions that are testable. It
can potentially be shown to be false by further observations or
experimentation.
- A
hypothesis should be specific. If it is too general it cannot be tested,
or tests will have so many variables that the results will be complicated
and difficult to interpret. A well-written hypothesis is so specific it actually determines
how the experiment should be set up.
- A
hypothesis should not include any untested assumptions if they can be
avoided. The hypothesis itself may be an assumption that is being tested,
but it should be phrased in a way that does not include assumptions that
are not tested in the experiment.
- It
is okay (and sometimes a good idea) to develop more than one hypothesis to
explain a set of observations. Competing hypotheses can often be tested
side-by-side in the same experiment.
Question: Why is the hypothesis
important to the scientific method?
Step 3: Testing the Hypothesis
A hypothesis may be tested in one
of two ways: by making additional observations of a natural situation, or by
setting up an experiment. In either case, the hypothesis is used to make
predictions, and the observations or experimental data collected are examined
to determine if they are consistent or inconsistent with those predictions.
Hypothesis testing, especially through experimentation, is at the core of the
scientific process. It is how scientists gain a better understanding of
how things work.
Testing a Hypothesis by Observation
Some hypotheses may be tested
through simple observation. For example, a researcher may formulate the
hypothesis that the sun always rises in the east. What might an alternative hypothesis be? If
his hypothesis is correct, he would predict that the sun will rise in the east
tomorrow. He can easily test such a prediction by rising before dawn and going
out to observe the sunrise. If the sun rises in the west, he will have
disproved the hypothesis. He will have shown that it does not hold true in
every situation. However, if he observes on that morning that the sun does in
fact rise in the east, he has not proven the hypothesis. He has made a
single observation that is consistent with, or supports, the hypothesis. As a
scientist, to confidently state that the sun will always rise in the
east, he will want to make many observations, under a variety of circumstances.
Note that in this instance no manipulation of circumstance is required to test
the hypothesis (i.e., you aren't altering the sun in any way).
Testing a Hypothesis by Experimentation
An experiment is a
controlled series of observations designed to test a specific hypothesis. In an
experiment, the researcher manipulates factors related to the hypothesis in
such a way that the effect of these factors on the observations (data) can be
readily measured and compared. Most experiments are an attempt to define a
cause-and-effect relationship between two factors or events—to explain why
something happens. For example, with the hypothesis "roses planted in
sunny areas bloom earlier than those grown in shady areas," the experiment
would be testing a cause-and-effect relationship between sunlight and time of blooming.
A major advantage of setting up
an experiment versus making observations of what is already available is that
it allows the researcher to control all the factors or events related to the
hypothesis, so that the true cause of an event can be more easily isolated. In
all cases, the hypothesis itself will determine the way the experiment will be
set up. For example, suppose my hypothesis is "the weight of an object is
proportional to the amount of time it takes to fall a certain distance." How would you test this hypothesis?
The
Qualities of a Good Experiment
Experiments can vary considerably
depending upon the hypothesis that is being tested. However, most experiments
have the following elements in common.
- The
experiment must be conducted on a group of
subjects that are narrowly defined and have certain aspects in common.
This is the group to which any conclusions must later be confined.
(Examples of possible subjects: female cancer patients over age 40, E.
coli bacteria, red giant stars, the nicotine molecule and its
derivatives.)
- All
subjects of the experiment should be (ideally) completely alike in all
ways except for the factor or factors that are
being tested. Factors that are compared in scientific experiments are
called variables. A variable is some aspect of a subject or event
that may differ over time or from one group of subjects to another. For
example, if a biologist wanted to test the effect of nitrogen on grass
growth, he would apply different amounts of nitrogen fertilizer to several
plots of grass. The grass in each of the plots should be as alike as
possible so that any difference in growth could be attributed to the
effect of the nitrogen. For example, all the grass should be of the same
species, planted at the same time and at the same density, receive the
same amount of water and sunlight, and so on. The variable in this case
would be the amount of nitrogen applied to the plants. The researcher
would not compare differing amounts of nitrogen across different grass
species to determine the effect of nitrogen on grass growth. What is the problem with using different
species of plants to compare the effect of nitrogen on plant growth?
There are different kinds of variables in an experiment. A factor that the experimenter controls, and changes intentionally to determine if it has an effect, is called an independent variable. A factor that is recorded as data in the experiment, and which is compared across different groups of subjects, is called a dependent variable. In many cases, the value of the dependent variable will be influenced by the value of an independent variable. The goal of the experiment is to determine a cause-and-effect relationship between independent and dependent variables—in this case, an effect of nitrogen on plant growth. In the nitrogen/grass experiment, (1) which factor was the independent variable? (2) Which factor was the dependent variable? - Nearly
all types of experiments require a control group and an experimental
group. The control group generally is not changed in any way, but
remains in a "natural state," while the experimental group is
modified in some way to examine the effect of the variable which of interest to the researcher. The control group
provides a standard of comparison for the experimental
groups. For example, in new drug trials, some patients are given a placebo
while others are given doses of the drug being tested. The placebo serves
as a control by showing the effect of no drug treatment on the patients.
In research terminology, the experimental groups
are often referred to as treatments, since each group is treated
differently. In the experimental test of
the effect of nitrogen on grass growth, what is the control group?
In the example of the nitrogen experiment, what is the purpose of a control group?
- In
research studies a great deal of emphasis is placed on repetition. It is
essential that an experiment or study include
enough subjects or enough observations for the researcher to make valid
conclusions. The two main reasons why repetition is important in
scientific studies are (1) variation among subjects or samples and (2)
measurement error.
Variation among Subjects
There is a great deal of
variation in nature. In a group of experimental subjects, much of this
variation may have little to do with the variables being studied,
but could still affect the outcome of the experiment in unpredicted
ways. For example, in an experiment designed to test the effects of alcohol
dose levels on reflex time in 18- to 22-year-old males, there would be
significant variation among individual responses to various doses of alcohol.
Some of this variation might be due to differences in
genetic make-up, to varying levels of previous alcohol use, or any number of
factors unknown to the researcher.
Because what the researcher wants
to discover is average dose level effects for this group, he must run
the test on a number of different subjects. Suppose he
performed the test on only 10 individuals. Do you think the average response
calculated would be the same as the average response of all 18- to 22-year-old
males? What if he tests 100 individuals, or 1,000? Do you think the average he
comes up with would be the same in each case? Chances are it would not be. So
which average would you predict would be most representative of all 18- to
22-year-old males?
A basic rule of statistics is,
the more observations you make, the closer the average of those observations
will be to the average for the whole population you are interested in. This is
because factors that vary among a population tend to occur most commonly in the
middle range, and least commonly at the two extremes. Take human height for example. Although you may find a man who is 7
feet tall, or one who is 4 feet tall, most men will fall somewhere between 5
and 6 feet in height. The more men we measure to determine average male height,
the less effect those uncommon extreme (tall or short) individuals will tend to
impact the average. Thus, one reason why repetition is so important in
experiments is that it helps to assure that the
conclusions made will be valid not only for the individuals tested, but also
for the greater population those individuals represent.
"The use of a sample (or
subset) of a population, an event, or some other aspect of nature for an
experimental group that is not large enough to be representative of the
whole" is called sampling error (Starr, Cecie, Biology: Concepts and
Applications, 4th ed. [Pacific Cove: Brooks/Cole, 2000],
glossary). If too few samples or subjects are used in an experiment, the
researcher may draw incorrect conclusions about the population those samples or
subjects represent.
Let’s
use a jellybean jar to learn about sampling error:
There
are 400 jellybeans in a jar. You cannot see what is inside the jar. You reach
in to take 1 jellybean, and it is green. You might assume the jar only contains
green jellybeans. The jar actually contains both green and black jellybeans. If
you pick 5 jellybeans you may by chance still get only green jellybeans, or you
may get 1 green and 4 black jellybeans. The latter may result in you assuming
that there are about 20% green and 80% black jellybeans in the jar. You then
choose 20 jellybeans and get 10 green and 10 black jellybeans, which may change
your estimate to 50% of each color. Lastly, you pick 100 jellybeans and find
that the ratio has changed to 31% green and 69% jellybeans, which turns out to
be close to the actual ratio between green and black jellybeans in the jar.
Reflection:
What do you think may be the impact of a small sample size in a controlled
experiment?
Measurement Error
The second reason why repetition
is necessary in research studies has to do with measurement error. Measurement
error may be the fault of the researcher, a slight difference in measuring
techniques among one or more technicians, or the result of limitations or
glitches in measuring equipment. Even the most careful researcher or the best
state-of-the-art equipment will make some mistakes in measuring or recording
data. Another way of looking at this is to say that, in any study, some
measurements will be more accurate than others will. If the researcher is
conscientious and the equipment is good, the majority of
measurements will be highly accurate, some will be somewhat inaccurate, and a
few may be considerably inaccurate. In this case, the same reasoning used above
also applies here: the more measurements taken, the less effect a few
inaccurate measurements will have on the overall average.
Step 4: Data Analysis
In any experiment, observations
are made, and often, measurements are taken. Measurements and observations
recorded in an experiment are referred to as data. The data collected
must relate to the hypothesis being tested. Any differences between
experimental and control groups must be expressed in some way (often
quantitatively) so that the groups may be compared. Graphs and charts are often
used to visualize the data and to identify patterns and relationships among the
variables.
Statistics is the branch of mathematics that deals
with interpretation of data. Data analysis refers to statistical methods
of determining whether any differences between the control group and
experimental groups are too great to be attributed to chance alone. Although a
discussion of statistical methods is beyond the scope of this tutorial, the
data analysis step is crucial because it provides a somewhat standardized means
for interpreting data. The statistical methods of data analysis used, and the
results of those analyses, are always included in the publication of scientific
research. This convention limits the subjective aspects of data interpretation
and allows scientists to scrutinize the working methods of their peers.
Why
is data analysis an important step in the scientific method?
Step 5: Stating Conclusions
The conclusions made in a
scientific experiment are particularly important. Often, the conclusion is the
only part of a study that gets communicated to the general
public. As such, it must be a statement of reality, based upon the
results of the experiment. To assure that this is the
case, the conclusions made in an experiment must (1) relate back to the
hypothesis being tested, (2) be limited to the population under study, and (3)
be stated as probabilities.
The hypothesis that is being
tested will be compared to the data collected in the experiment. If the
experimental results contradict the hypothesis, it is rejected and further
testing of that hypothesis under those conditions is not necessary. However, if
the hypothesis is not shown to be wrong, that does not conclusively prove that
it is right! In scientific terms, the hypothesis is said to be "supported
by the data." Further testing will be done to see if the hypothesis is
supported under a number of trials and under different
conditions.
If the hypothesis holds up to
extensive testing then the temptation is to claim that
it is correct. However, keep in mind that the number of experiments and
observations made will only represent a subset of all the situations in which
the hypothesis may potentially be tested. In other words, experimental data
will only show part of the picture. There is always the possibility that a
further experiment may show the hypothesis to be wrong in some situations.
Also, note that the limits of current knowledge and available technologies may
prevent a researcher from devising an experiment that would disprove a
particular hypothesis.
The researcher must be sure to
limit his or her conclusions to apply only to the subjects tested in the study.
If a particular species of fish is shown to consume their young 90 percent of
the time when raised in captivity, that doesn't necessarily mean that all fish
will do so, or that this fish's behavior would be the same in its native
habitat.
Finally, the conclusions of the
experiment are generally stated as probabilities. A careful scientist would
never say, "drug x kills cancer cells;" she would more likely
say, "drug x was shown to destroy 85 percent of cancerous skin
cells in rats in lab trials." Notice how very different these two
statements are. There is a tendency in the media and in the general
public to gravitate toward the first statement. This makes a terrific
headline and is also easy to interpret; it is absolute. Remember though, in
science conclusions must be confined to the population under study; broad
generalizations should be avoided. The second statement is sound science. There
is data to back it up. Later studies may reveal a more universal effect of the
drug on cancerous cells, or they may not. Most researchers would be unwilling
to stake their reputations on the first statement.
As a student, you should read and
interpret popular press articles about research studies very carefully. From
the text, can you determine how the experiment was set up and what variables
were measured? Are the observations and data collected appropriate to the
hypothesis being tested? Are the conclusions supported by the data? Are the
conclusions worded in a scientific context (as probability statements) or are
they generalized for dramatic effect? In any researched-based assignment, it is
a good idea to refer to the original publication of a study (usually found in
professional journals) and to interpret the facts for yourself.
|
Activity: Interpretation of a Science Study
as Presented in the Popular Media Read the article below, and then answer the questions that follow. Follow your instructor's directions for submitting your responses. E. Coli Kills Cancer Cancer is often fought with chemotherapy, and the effects of these toxic drugs can be excruciating. But Canadian researchers have discovered that a familiar, yet potent toxin can actually shrink brain tumors in less than 48 hours with no apparent ill effects. The cancer-fighting chemical is verotoxin, which is produced by the ubiquitous E.coli bacteria. This toxin, which causes diarrhea, was injected into human brain tumors implanted in mice. It not only shrank the tumors, but none of the tumors reappeared. How can a substance dangerous in the stomach not be dangerous to the brain cells? "What is important is the amount of the toxin," says Dr. Clifford Lingwood of the Hospital for Sick Children in Toronto. Just a little bit of it won't hurt you, but the more you're exposed to the sicker you'll get. The idea is to find a level that is harmless to the animal as a whole, but deadly to the cancer cells. A study of baboons measured how much verotoxin it would take to make an ape sick. Animals given small doses showed no side effects, nor did the mice in Lingwood's study. Lingwood says that the verotoxin stops the growth of new blood vessels. "Tumor cells are particularly susceptible," he explains, because the tumors are marked by a specific glycolipid, a receptor that acts as a gateway into the cell. The verotoxin finds the glycolipids on tumors and the blood vessels that surround the tumor cells. It attaches itself to the receptor and causes the cells to commit suicide. Verotoxins ignore normal, non-cancerous brain cells, which don't contain the receptor. With the toxin attacking both its outer membrane and its food supply, the brain tumor shrivels almost immediately after treatment begins. In cancer cells in Petri dishes, "You can see a significant difference in 90 minutes," says Lingwood. Lingwood's results were reported in the June issue of the journal, Oncology Research. —Martha
Heil Answer the following questions:
|
Qualities of a Good Experiment
The following are qualities of a
good experiment:
- narrowly
defined subjects
- all
subjects treated alike except for the factor or variable being studied
- a control group is used for comparison
- measurements
related to the factors being studied are carefully recorded
- enough
samples or subjects are used so that conclusions are valid for the
population of interest
- conclusions
made relate back to the hypothesis, are limited to the population being
studied, and are stated in terms of probabilities
