Change One Thing

The two wards

In 1846, a young doctor named Ignaz Semmelweis took a job at the biggest hospital in Vienna. The hospital had two wards where women came to give birth. In the first ward, the doctors and medical students delivered the babies. In the second ward, midwives did. The two wards were in the same building, with the same food, the same air, and the same kinds of patients. A woman went to one ward or the other depending on the day of the week she arrived.

But the two wards were not the same. In the doctors’ ward, about eleven of every hundred mothers died of a fever soon after giving birth. In the midwives’ ward, fewer than three in a hundred did. The women knew it. They begged not to be sent to the doctors’ ward. Some gave birth in the street on purpose rather than go in, and the women who did were safer than the women inside.

Semmelweis could not stop asking why. Doctors of the time blamed bad air, or crowding, or fear. He checked each one. The air was the same in both wards. The midwives’ ward was more crowded, not less. Fear did not explain it either: the women in the street were as frightened as anyone. None of the usual answers fit.

Then, in 1847, a friend of his, a doctor, was cut on the finger by a student’s knife during an autopsy. An autopsy is the examination of a dead body to find the cause of death. The friend died of the same fever that was killing the mothers. Semmelweis saw the pattern. The doctors and students went from the autopsy room to the mothers’ beds, straight from the dead to the living, and they did not wash their hands in between. The midwives never touched the dead.

Wash your hands

He made a guess that could be tested: something from the dead bodies is carried to the mothers on the doctors’ hands. If the doctors wash it off before touching a patient, fewer mothers will die. A guess of that kind is called a hypothesis: a proposed answer to the question, put in a way that a test could prove wrong.

Then he changed one thing. From May 1847, every doctor and student had to wash their hands in a bowl of chlorine water before entering the ward. Nothing else changed: the same building, the same beds, the same doctors. And month after month, he counted the deaths.

The deaths fell at once. Within a few months, about one or two mothers in a hundred were dying in the doctors’ ward, the same as in the midwives’ ward. The hypothesis was supported, and he kept counting through the next year to be sure.

You would think the story ends there. It does not. Semmelweis was slow to write down what he had done in a way that others could follow. For years the only reports were his lectures and the letters his students sent to other hospitals. Many doctors would not believe the claim. Part of the reason was that it accused them: it said their own hands had carried death. Part of the reason was that nobody yet knew what the something on the hands was. His full book did not appear until 1861, fourteen years later, and it was long and angry. He died in 1865, before the world came around.

It did come around. Twenty years after the chlorine bowl, Louis Pasteur showed that living germs cause disease, and the surgeon Joseph Lister used that idea to clean wounds and instruments. Then doctors could see why the washing had worked, and washing the hands became the rule everywhere. The story has two lessons, and this room teaches both. Change one thing, and count. And tell others clearly, so they can check.

The wheel

Here are the steps, drawn as a wheel, because the last step leads back to the first. Tap each station to see what it means and what Semmelweis did there. Real science circles back and skips around; the order is a way to learn it, and it is the order the GED test expects you to know.

Two things to notice. The test is run more than once: the ×n at the fifth station stands for that, n times, as many as it takes. And the outer ring: other people repeat the whole thing. A result that only one person ever got is not yet science.

A question you can test

Now a small experiment, one you could do yourself. Nadia grows basil in pots on her fire escape in Sunset Park, Brooklyn. Her neighbor Mr. Okoro tells her to mix used coffee grounds into the soil. “Plants love it,” he says. “They grow twice as fast.”

That is an observation of a kind: somebody noticed something, or thinks they did. Nadia turns it into a question: do coffee grounds make basil grow taller? A good question for a test names something you could change and something you could measure. She could add grounds or not. She could measure height with a ruler.

“Are coffee grounds good for plants?” is not yet a question a test can answer. Good how? Good for what? Until you say what you will measure, there is nothing to test.

Before she tests anything, Nadia looks up what is already known. The gardening sites disagree. Some say the grounds add nitrogen, which plants need. Others say they pack down and hold too much water. Nobody she can find has measured it. So she will.

A hypothesis

A hypothesis is a proposed answer to the question, put so that a test could prove it wrong. Nadia’s: if I mix used coffee grounds into the soil, then the basil will grow taller than basil without them.

Notice the shape: if I change this, then that will happen. The “if” names the change she will make. The “then” names what she expects to measure. A hypothesis is not a hope, and it does not have to be right. A hypothesis that turns out wrong has taught you something. A guess that no test could ever prove wrong has taught you nothing, and it is not a hypothesis at all.

The two variables

A variable is anything in the experiment that can change. Two of them matter most, and the GED test asks about them again and again.

The independent variable is the one thing you change on purpose. For Nadia, it is the coffee grounds: some plants get them and some do not. Nothing else in the experiment decides that. It is called independent because she chooses it.

The dependent variable is the thing you measure to see what happened. For Nadia, it is the height of the basil. It is called dependent because, if the hypothesis is right, it depends on the coffee grounds.

A way to keep them straight: I change the independent variable; I measure the dependent variable. The change comes first, and the measurement answers it.

Two groups, and everything else the same

To see what the coffee grounds do, Nadia needs something to compare with. So she sets up two groups of plants.

The experimental group gets the change: six pots of basil with a spoonful of used coffee grounds mixed into the soil each week. The control group gets no change: six pots of basil with nothing added. The control group shows what basil does on its own. Without it, Nadia could not tell whether her plants grew because of the grounds or because basil just grows.

Everything else must be the same for both groups. The same seeds from the same packet. The same pots, the same soil, the same window, the same water on the same days. The things kept the same are called constants, or controlled variables. If the two groups differ in only one way, the coffee grounds, then any difference in height can only have come from the grounds. That is what makes the test fair.

Here it is in one line: change one thing, keep everything else the same, and compare.

More than one plant, more than one time

Why six plants in each group, and not one? Because one plant can be odd. It might come from a weak seed, or sit in a draft, or have a worm in the pot. With six, one odd plant cannot fool you, because you look at the average. The number of things you test is called the sample size, and bigger is better.

Nadia measures every plant once a week for six weeks and writes the numbers down. Watch the six weeks. The plants are drawn to scale, with a ruler beside them.

Most single picks agree with the averages, but not all of them do. That is why one plant is never enough, and why a scientist runs the whole experiment more than once. Each complete run is called a trial. Nadia will run hers again in the spring.

Data, then conclusion

Data is what you measured, written down. Here are Nadia’s heights at the end of week six, in centimeters, with the average of each group. The graph shows the same numbers as a picture, drawn to scale.

The plants with coffee grounds grew less, not more. Nadia’s hypothesis was not supported. That is not a failure. It is an answer, and a more useful one than Mr. Okoro’s.

Keep two kinds of sentence apart. The data says what was measured: the plants without grounds averaged 18 centimeters, and the plants with grounds averaged 15. The conclusion says what the data means for the hypothesis: in this test, coffee grounds did not make basil grow taller; the plants with grounds grew less. The conclusion goes only as far as the data. It does not say “coffee grounds kill plants.” None died. It does not say “never use coffee grounds.” This was one test, one kind of plant, one fire escape.

Tell others, so they can do it again

The last step is the one the charts make sound easy: communicate results. It means more than telling. Nadia writes a report with six parts: the question, the hypothesis, the materials, the procedure step by step, the data, and the conclusion. The test of the report is this: could a stranger follow it and do exactly what she did?

That matters because of what happens next. Her cousin Dev in Jackson Heights reads the report and runs the experiment himself. When other people, in another place, follow the report and get the same result, the result is reproducible. When Nadia runs it again herself in the spring and gets the same result, the result is repeatable. A result that holds up when it is repeated and reproduced is one you can trust. A result that nobody else can get is one that nobody should trust yet, and that includes Semmelweis’s, until other hospitals tried it.

When the results come in, they are compared. Here are three runs: Nadia’s first, Nadia’s second, and Dev’s.

The numbers are not identical. They never are. But they agree: in every run, the plants with grounds grew a few centimeters less. That agreement is what turns one person’s result into something known.

In a laboratory, the report goes further. It is sent to a science journal, where other scientists read it, look for mistakes, and ask for more before it is printed. Other labs then try to reproduce it. When an explanation has been tested this way over and over, by many people, and has held up every time, it is called a theory. The germ theory of disease, the idea Semmelweis did not yet have, is one. In science, a theory is not a guess. It is the opposite: an explanation that has passed every test so far.

And the wheel turns. Nadia’s answer raises a new question: why did the grounds slow the basil? She reads that used coffee grounds still hold caffeine, and that caffeine slows the growth of many young plants. So her next hypothesis is: if the grounds are composted for three months first, then they will not slow the basil. A new test begins.

The words in one place

Everything this room has named, in the order the wheel turns.

Observation

Something you notice, with your own senses or with an instrument.

Question

What you want to find out, put so that measuring could answer it.

Hypothesis

A proposed answer that a test could prove wrong. Often in the form: if I change this, then that will happen.

Variable

Anything in an experiment that can change.

Independent variable

The one thing you change on purpose.

Dependent variable

The thing you measure to see the effect. It depends on the change.

Constants (controlled variables)

Everything kept the same for both groups, so that only one thing differs.

Control group

The group that gets no change. It shows what happens on its own, for comparison.

Experimental group

The group that gets the change.

Sample size

How many things or people you test. More is better.

Trial

One complete run of the experiment.

Data

What you measured, written down: numbers in a table or a graph.

Conclusion

What the data means for the hypothesis: supported, or not supported. It goes no further than the data.

Repeatable

You can do it again and get a similar result.

Reproducible

Other people, in another place, can follow your report and get a similar result.

Placebo

A treatment with nothing in it, given to the control group so that the only difference between the groups is the real treatment.

Blind test

The people measuring, or the people taking part, do not know who is in which group. It keeps hope from leaning on the numbers.

Observational study

Science done by watching and recording, without changing anything. It can find a pattern and lead to a hypothesis. It cannot by itself show what causes what.

Theory

An explanation tested many times by many people that has held up every time. In science, the opposite of a guess.

Name the parts

Eight experiments, each with a twist. Read the story, then name the four parts: the control group, the experimental group, the independent variable, and the dependent variable. Wrong picks cost nothing; the page tells you what the trap was. After the questions, the twist is explained.

Find the mistake

Now the mistakes. Each of these experiments has something wrong with it, the kind of mistake the GED test asks you to spot. Say what is wrong, then say how to fix it.

Where the road leads

You can now follow an experiment from the question to the report and name its parts: the hypothesis, the two variables, the two groups, the constants, the data, and the conclusion. You can tell a fair test from an unfair one. And you know the step that Semmelweis missed: tell others clearly, so they can do it again.

On the GED science test, these words come up in reading passages about experiments. The questions ask which is the independent variable, which group is the control, what the data supports, and what would make the experiment better. The Look Again quizzes on this site practice exactly that.