Causes and Effects Help

Y causes X

Imagine that the horizontal axis in Fig. 7-4 represents 12 different groups of people in a medical research survey.

Each hash mark on the horizontal axis represents one group. Plot X is a point-to-point graph of the relative number of fatal strokes in a given year for the people in each of the 12 groups; plot Y is a point-to-point graph of the relative average blood pressure levels of the people in the 12 groups during the same year. (These are hypothetical graphs, not based on any real historical experiments, but a real-life survey might come up with results something like this. Medical research has shown a correlation between blood pressure and the frequency of fatal strokes.)

Is there a cause-and-effect relationship between the value of X and the value of Y here? Most doctors would answer with a qualified yes: variations in Y cause the observed variations in X (Fig. 7-5B). Simply put: high blood pressure can cause fatal strokes, in the sense that, if all other factors are equal, a person with high blood pressure is more likely to have a fatal stroke than a person identical in every other respect, but with normal blood pressure.

Fig. 7-5B. At B, Y causes X.

What about the reverse argument? Can fatal strokes cause high blood pressure (X causes Y)? No. That's clearly absurd.

Complications

Experts in meteorology and medical science who read this may, at this point, be getting a little nervous. Aren't the above scenarios oversimplified? Yes, they are. The cause-and-effect relationships described aren't ''pure.'' In real life, ''pure'' cause–effect events, where there is one certain cause and one inevitable effect, rarely occur.

The brightness of the sunshine is not, all by itself, the only cause-and-effect factor in the temperature during the course of a day. A nearby lake or ocean, the wind direction and speed, and the passage of a weather front can all have an effect on the temperature at any given location. We've all seen the weather clear and brighten, along with an abrupt drop in temperature, when a strong front passes by. The sun comes out, and it gets cold. That defies the notion that bright sun causes things to heat up, even though the notion, in its ''pure'' form where all other factors are equal, is valid. The problem is that other factors are not always equal!

In regards to the blood-pressure-versus-stroke relationship, there are numerous other factors involved, and scientists aren't sure they know them all yet. New discoveries are constantly being made in this field. Examples of other factors that might play cause-and-effect roles in the occurrence of fatal strokes include nutrition, stress, cholesterol level, body fat index, presence or absence of diabetes, age, and heredity. A cause-and-effect relationship (Y causes X) exists, but it is not ''pure.''

D causes both X and Y

Now suppose that the horizontal axis in Fig. 7-4 represents 12 different groups of people in another medical research survey. Again, each hash mark on the horizontal axis represents one group. Plot X is a point-to-point graph of the relative number of heart attacks in a given year for the people in each of the 12 groups; plot Y is a point-to-point graph of the relative average blood cholesterol levels of the people in the 12 groups during the same year. As in the stroke scenario, these are hypothetical graphs. But they're plausible. Medicine has shown a correlation between blood cholesterol and the frequency of heart attacks.

Before I make enemies in the medical profession, let me say that the purpose of this discussion is not to resolve the cholesterol-versus-heart-disease issue, but to illustrate complex cause–effect relationships. It's easier to understand a discussion about real-life factors than to leave things entirely generic. I do not have the answer to the cholesterol-versus-heart-disease riddle. If I did, I'd be writing a different book.

When scientists first began to examine the hearts of people who died of heart attacks in the early and middle 1900s, they found ''lumps'' called plaques in the arteries. It was theorized that plaques caused the blood flow to slow down, contributing to clots that eventually cut off the blood to part of the heart, causing tissue death. The plaques were found to contain cholesterol. Evidently, cholesterol could accumulate inside the arteries. Consulting data showing a correlation between blood cholesterol levels and heart attacks, scientists got the idea that if the level of cholesterol in the blood could be reduced, the likelihood of a heart attack would also go down. The theory was that fewer or smaller plaques would form, reducing the chances of clot formation that could obstruct an artery. At least, such was the hope.

The obvious first point of attack was to tell heart patients to reduce the amount of cholesterol in their food, hoping that this change in eating behavior would cause blood cholesterol levels to go down. In many cases, a low-cholesterol diet did bring down blood cholesterol levels. (Later, special drugs were developed that had the same effect.) Studies continue along these lines. It is becoming apparent that reducing the amount of cholesterol in the diet, mainly by substituting fruits, vegetables, and whole grains for cholesterol-rich foods, can reduce the levels of cholesterol in the blood. This type of dietary improvement can apparently also reduce the likelihood that a person will have a heart attack in the next year, or two, or three. There's more than mere correlation going on here. There's causation, too. But how much causation is there? And in what directions does it operate?

Let's call the amount of dietary cholesterol ''factor D.'' According to current popular medical theory, there is a cause-and-effect relation between this factor and both X and Y. Some studies have indicated that, all other things being equal, people who eat lots of cholesterol-rich foods have more heart attacks than people whose diets are cholesterol-lean. The scenario is shown in Fig. 7-5C. There is a cause-and-effect relation between factor D (the amount of cholesterol in the diet) and factor X (the number of heart attacks); there is also a cause-and-effect relation between factor D and factor Y (the average blood cholesterol level). But most scientists would agree that it's an oversimplification to say this represents the whole picture. If you become a strict vegetarian and avoid cholesterol-containing foods altogether, there's no guarantee that you'll never have a heart attack. If you eat steak and eggs every day for breakfast, it doesn't mean that you are doomed to have a heart attack. The cause-and-effect relationship exists, but it's not ''pure,'' and it's not absolute.

Fig. 7-5C. At C, D causes both X and Y.

Multiple factors cause both X and Y

If you watch television shows where the advertising is aimed at middle-aged and older folks, you'll hear all about cholesterol and heart disease – probably a good deal more than you want to hear. High cholesterol, low cholesterol, HDL, LDL, big particles, small particles. You might start wondering whether you should go to a chemistry lab rather than a kitchen to prepare your food. The cause-and-effect relationship between cholesterol and heart disease is complicated. The more we learn, it appears, the less we know.

Let's introduce and identify three new friends: factors S, H, and E. Factor S is named ''Stress'' (in the sense of anxiety and frustration), factor H is named ''Heredity'' (in the sense of genetic background), and factor E is named ''Exercise'' (in the sense of physical activity). Over the past several decades, cause-and-effect relationships have been suggested between each of these factors and blood cholesterol, and between each of these factors and the frequency of heart attacks. Figure 7-6 illustrates this sort of ''cause-and-effect web.'' Proving the validity of each link – for example, whether or not stress, all by itself, can influence cholesterol in the blood – is a task for future researchers. But all of the links shown in the diagram have been suggested by somebody.

Coincidence

The existence of correlation between two phenomena doesn't necessarily imply any particular cause-and-effect scenario. Two phenomena can be correlated because of a sheer coincidence. This is most likely to happen when the amount of data obtained is not adequate. In the case of blood cholesterol versus the frequency of heart attacks, test populations have traditionally contained thousands of elements (people). The researchers are justified in their conclusions that such correlation is not the product of coincidence. It is reasonable to suppose that some cause-and-effect interaction exists. Researchers are still busy figuring out exactly how it all works, and if they ever get it completely unraveled, it's a good bet that an illustration of the ''cause-and-effect web'' will look a lot more complicated than Fig. 7-6.

Correlation indicates that things take place more or less in concert with one another. This allows us to predict certain future events with varying degrees of accuracy. But there is another form of order that can be found in nature. This form of order, defined by a new science called chaos theory, illustrates that some phenomena, totally unpredictable and which defy statistical analysis in the short term or on a small scale, can nevertheless be highly ordered and predictable on a large scale. In a moment, we'll look at this.

Causes and Effects Practice Problems

Practice 1

What are some reasonable cause-and-effect relationships that might exist in Fig. 7-6, other than those shown or those directly between X and Y? Use arrows to show cause and effect, and use the abbreviations shown.

Solution 1

Practice 2

What are some cause-and-effect relationships in the diagram of Fig. 7-6 that are questionable or absurd?

Solution 2

Consider the following. Think about how you might conduct statistical experiments to find out whether or not the first three of these might be moved into the preceding category.

• H → E (Question: Do people of certain genetic backgrounds naturally get more physical exercise than people of other genetic backgrounds?)
• S → E (Question: Can stress motivate some types of people to exercise more, yet motivate others to exercise less?)
• S → D (Question: Do certain types of people eat more under stress, while others eat less?)
• S → H (Obviously absurd. Stress can't affect a person's heredity!)
• E → H (Obviously absurd. Exercise can't affect a person's heredity!)
• D → H (Obviously absurd. Dietary habits can't affect a person's heredity!)

Practice problems for these concepts can be found at:

Correlation, Causation, Order, and Chaos Practice Test