Scientific Method

a. Scientific Approach/Scientific method/Research

Science has been defined as a specific method or logic of inquiry. This definition suggests that the method of science is somehow unique and different than other methods of inquiry or that it has specific rules or characteristics that have to be followed. These characteristics, while necessary to distinguish science, are not limited to the realm of science.

Methods of collecting data

Observation

The scientific method begins with the collection of data. Data for scientific and technical purposes are obtained in one of two ways, by observation and by experimentation. The method of observation appears simple enough; we watch carefully and record what we see. However, this is not quite a s easy as it sound. If you have or could listen to a court proceeding arising out of a road accident, you may wonder sometimes, whether the witnesses are all describing the same event.

Accurate and reliable observation, therefore, requires much training and practice.

Experimental Work

Experimenting is the second method of obtaining data mentioned above. In many ways the method of controlled, repeatable experimentation is the very essence of the scientific method.

The aim of a properly conducted experiment is to create an artificial, reproducible situation in which the factor (variable) to be studied can be isolated and observed. This is not quite as easy as it sounds. Therefore, the design of an experiment becomes extremely important. The first care of the experimenter must be to ensure that he/she has in fact set up a truly reproducible situation - its importance cannot be over-stressed. If observations are not repeatable, then our observations and explanations are likewise unreliable and therefore useless.

The second essential is to isolate the factor (variable) under study. If this is not done, it is possible, even likely, that false conclusions will be reached. This can be very complex and difficult but can be explained with the aid of a simple and trivial example. Suppose you drop a lump of sugar in to a cup of tea and stir it. When you taste the tea, it is sweet. What is the cause of the sweetening? Obviously, the sugar. Obviously, until some individual comes along and asks, "How do you know it was the sugar and not the stirring?"

With the experiment as described, you do not know. It is necessary to separate the two factors (variables). It is easy enough to show that stirring a cup of teat without sugar does not cause the tea to become sweet, and this seems to settle the issue until it is noticed that simply adding a lump of sugar to an unstirred cup of tea does not seem to affect sweetness, either. To be certain, it is necessary to take two identical cups of tea, add sugar to one, and stir both. Only the one with the sugar becomes sweet, so it is reasonable to infer that sugar is the cause of sweetness. Again taking identical cups of tea, adding sugar to one, and allowing them to stand can confirm this. After a time the sugar diffused through the tea, and sweetens it, so the final conclusion is that sugar causes the sweetness, and stirring merely hastens its dissolution.

Control

Control is perhaps the single most important element in the scientific methodology. Control is important because it enable the scientist to identify the causes of his or her observations. The above described experiment with the tea is a controlled experiment. The factors which might have affected the outcome have been isolated and studied separately. The cup without the sugar in these experiments is described as the control, and it is always advisable, if it is possible, to arrange for a control or "blank " experiment as a check. In any case, only one thing should be studied at a time, and if there are several factors which can be altered, observations and measurements on one of them must be completed while the others are held constant; only then may another factor (variable) be selected for alteration, the same precautions being observed. There are methods whereby valid results may be obtained when observing several variable factors at once.

Systematization

The next step in the scientific method is the systematization of data. A simple collection of facts may be interesting, but it is not of much use to anyone. A primary concern of science is unification. If we can show that two apparently unrelated facts are actually connected, we have taken a step forward in our effort to understand them. Hence, the importance of systematization is getting data into coherent form and order. This may be merely a matter of careful arrangement and tabulation or data may have to be subjected to mathematical manipulation. This class will deal with both tabulation of data and statistical analysis.

Hypothesis

So far, we have imagined workers in the field and laboratory collecting data and arranging them systematically. The next step is to show how the new facts fit into the existing body of knowledge, whether they are connected and if so, how. This is an intellectual exercise. This explanation of the facts is termed an hypothesis. It must fit the fact, all the facts, or it is useless. If it does fit the facts, it can be tested. The test of an hypothesis and the only test which is accepted as valid for scientific purposes is prediction. Thus, if a series of experiments, conducted with the precautions described above, has yielded a number of measurements for a certain factor under differing, known conditions, it may be possible to set up and hypothesize linking the value of the measurements with the varying conditions. For example, it is noted that the pressure of steam in a boiler varies with the temperature of the water, and it is found that the relationship can be expressed by a mathematical formula. The hypothesis in this case is that pressure and temperature were not included in the original experiment.

A prediction of this kind can, of course, only be tested by another experimenter, a fact that accounts for the central importance of the experiment in the scientific method. If the new experiment does not confirm the prediction, and assuming that no fault can be found with experimental method, then the hypothesis cannot stand. It must be modified to take account of the new result, if it is possible, discarded. If, however, repeated experiments show that the predictions are consistently verified, the hypothesis is elevated to the dignity of a Law, and can be used as a foundation for further work.

It will be evident that this is a self-perpetuation process. The experiments required to test a hypothesis might themselves reveal fresh facts, which provide the starting point for further work. Thus the progress of science is continuous. When a series of hypotheses has been built up, and these are found to be consistent with one another and with observed phenomena, the result is a scientific theory. Because of its wider scope, a theory is more important than a simple hypothesis; it is also more difficult to test, but is susceptible of exactly the same test or prediction. Predictions from theory are apt to be more spectacular than those from simple hypothesis.

Experiments are conduct in an attempt to answer certain questions. They represent attempts to identify why something is happening, what causes some event, and under what conditions does the even occur.

b. What is Research?

To understand what research is, let's begin by considering what research is not. It seems that in this day and age that we have been conditioned with the term in so many ways that you might not be sure what the term really means.

Many students have labored under the false impression that looking up a few facts and writing them down in a documented paper is research. Such activity is nothing more than fact-finding and fact transcribing. No amount of transfer of information from one place to another, even though the transformation is acknowledged by footnote, is it research. Yet the misconception that fact transferal is research still exists.

Research is essentially a way of thinking and a way of looking at the world. There is no particular reason why this method should be the prerogative of scientists, but the fact is that most people do not organize their thought processes in a way most likely to lead to results.

Research has discrete characteristics, which appear sequentially. These have been described in many different yet similar ways. Paul Leedy (1974) describes the research process in a series of steps which; taken together, comprise the particular approach to the discovery of the truth, which is called research. Leedy steps are represented in the diagram below:

I'm sure that you will recognize Leedy's steps in research as the scientific method. They are one and the same thing. The scientific method or research method is a series of logical and orderly steps that lead to new knowledge and new concepts. The steps are really a system of common sense.

Define the problem

You can't solve a problem unless you see that one exists. Science calls for the kind of mind that recognizes problems and asks questions. For example, how does a root absorb water from the soil? Why does a plant stem bend toward the light? What controls your heart beat? Well-planned experiments can answer each of these questions. But in science, every answer raises new questions. Successful research leads to new research and new knowledge.

Collect information on the problem

Scientists must build on the work of other scientists. Otherwise science could not advance beyond what on e person could learn in a lifetime. Before beginning an experiment, the scientist studies all-important information that has to do with the problem. Often it turns out that someone has largely answered many of the questions involved. For instance, a library of scientific papers, journals, and books is an important part of a research center.

Form a hypothesis

The available information may not fully explain a problem. The researcher must then begin to experiment. At this point, a hypothesis is needed. The hypothesis is sort of a working explanation or trial answer.

The hypothesis gives the experimenters a point to aim at. But no matter how reasonable the hypothesis seems, it cannot be accepted until supported by a large number of tests. The research worker must be open-minded enough to change or drop a hypothesis if the evidence does not support it.

Experiment to test the hypothesis

The scientist must set up an experiment that will either support or disprove the hypothesis. This means that the experiment must test only the condition involved in the hypothesis. All other factors must be removed or otherwise accounted for. The one factor to be tested is called the single variable or the experimental factor. In this way, the control shows the importance of the missing experimental factor.

Observe and record data from the experiment

Everything about the experiment should be recorded accurately. How was it planned and set up? Under what conditions was it carried out? What happened during the experiment? And finally, what were the results? The record may include notes, drawings, tables, graphs, or other forms of information. This information is called data. In modern research, data are often processes by computer.

Draw conclusions

Data have value only when valid conclusions are drawn from them. Such conclusions must be based entirely on facts observed in the experiment. If other experiments continue to support the hypothesis, it may come to be called a theory. A food theory explains and predicts new facts.

Report the data

c. An Example of the Scientific Approach/Method/Research

The following example is reprinted, with permission from Wadsworth Publishing Company, Belmont, California (Statistics and Research Design in the Behavioral Sciences by Richard S. Lehman, 1991) and represents and example of the scientific approach to a problem:

Suppose that you become violently ill immediately after drinking a can of your favorite diet soda. After you recovered, how would you feel about your old favorite soda? Would you still like it? Would you switch brands? Or might you avoid most sweet drinks in the future? Could you induce the same kind of experience on animals? If you could, might you be able to devise a way for animals to avoid consuming toxic chemicals?

This problem and the questions that arise illustrate the broad range of concerns raised by scientists. The following will illustrate not only the scientific method, but also the theoretical and practical side to scientific investigation.

Conditioned Taste Aversion

If you let a rat consume some substance (saccharin, for example) and then administer a small dose of lithium chloride, the animal will become violently ill. After the animal recovers from the brief sickness, it will avoid the substance that it tasted immediately before getting sick. This phenomenon is known as a conditioned taste aversion and is usually learned very quickly - normally in one experience. Conditioned taste aversion is uncommon among humans despite our earlier suggestion about diet sodas, but can easily be induced in animals. Conditioned aversion provides a valuable tool for studying animal's perceptual abilities.

While humans certainly are more complex in most ways than other animals, our taste system is very similar to that of a laboratory rat. As a result, the rat often makes an ideal organism for researchers studying the chemical senses of taste and smell. Not only are animals more dependable than humans (rats almost never forget to show up for an experiment, for example), but also we can ask them to taste and smell substances that might be distasteful or injurious to humans. Using animals, taste researchers have discovered some facts about chemical senses that, while they may not apply to humans, are interesting and possibly useful because they illuminate the basic processes of taste that seem to be shared by humans and other animals.

Lest you think that such an unusual phenomenon, as conditioned taste aversion has no interest to you, consider the fate of the California condor. Their numbers reduced to the point of near extinction, condors now exist only in breeding programs in zoos. What killed them? It was the pesticide DDT that was in wide use until banned a few years ago. The DDT entered the condor's food chain from agricultural spraying and weakened the shells of the condor eggs badly so that the parents' weight crushed them and few hatched. The result? The condors are extinct in the wild. If only they could have been given a conditioned taste aversion, they would have avoided anything containing DDT. A number of researchers are currently investigating the taste preferences and aversions of a large number of species. One of their goals is to develop pesticide flavors that are specific to the target species and that will be avoided by animals that are targeted. Dr. Steward's research in taste is closely related to this pesticide problem.

Saccharin is a common sweetener in many dietetic beverages. It is used because it tastes sweet and has many fewer calories than does sugar. Rats and humans both seem to like its taste, at least in moderation. Given a choice between plain water and water containing saccharin, rats will prefer saccharin-water.

Chemically, saccharin and sugar are quite different, yet both taste sweet. An important question for a taste researcher to ask is whether sugar and saccharin seem behaviorally the same. One way to find out would be to present a choice between sugar-water and saccharin-water and observe how much of each solution the animals drank in a test period. If the rats choose them in roughly the same proportion, then we could conclude that the two sweet substances were the same behaviorally. (And in fact they are, at least for reasonable concentrations.)

Another identification of behaviorally similarity might make use of conditioned taste aversion. If sugar and saccharin are behaviorally the same, then a conditioned taste aversion to one of them should lead animals to avoid the other. In other words, suppose we were to introduce a conditioned taste aversion to saccharin. We know that animals with such an aversion will avoid saccharin in the future. But will they also avoid sugar-water? If they do, then the aversion to saccharin has generalized to sugar, another sweet taste, indicating that the two are behaviorally the same for the animals. To verify that only the sweet taste has generalized, we might present some other taste substances as well. If the assumption of, say, a salty tasting solution doesn't change after the aversion training, then we know the aversion to saccharin hasn't generalized to salt. One of Dr. Steward's experiments investigated just those questions.

Twenty laboratory rats were used in the experiment. Dr. Steward randomly divided animals into two groups of 10 rats each. In one group, which we can call the experimental group, the animals were given saccharin-water and then administered lithium chloride to induce a conditioned taste aversion. The other group, the control group, was given plain water, followed by lithium chloride. After recovery, all animals were offered four taste solutions; sugar, salt, quinine (bitter), and hydrochloric acid (sour). The overall plan of the experiment is illustrated in Figure 1.1. If the rats treat sugar and saccharin behaviorally the same, the conditioned taste aversion to a sweet taste in the experimental group should lead to less concentration of sugar and to no change in the consumption of the other solutions. Did it?

FIGURE 1.1

In a 15-mintue period, Dr. Steward measured the amount of each taste solution consumed (in millimeters). On the average, the experimental group drank 4.05 milliliters of sugar water, while the control animals drank 20.35 milliliters. On the basis of these values, and ignoring for the moment the other three taste solutions, it certainly looks like the taste aversion generalized to sugar. We can probably conclude that, by this test, saccharin and sugar are behaviorally the same. We can't be absolutely certain, since the results could be due just to chance and a second experiment might confirm this finding; but the difference between 4 and 20 milliliters suggests that animals have avoided the sweet taste of sugar.

How about the effect of the aversion training on the other three substances? The average amounts consumed for all four-taste solutions follow:

Condition Experimental Control

Sugar 4.05 20.35

Salt 19.70 19.80

Sour 10.20 8.75

Bitter 9.00 5.55

Graphically, the results are striking (see Figure 1.2). There are practically no differences between the two groups of animals, except for their consumption of sugar. Slight variation sexist between the groups in their consumptions of the sour and bitter solutions, but these are nothing like the magnitude of the variation for sugar. Our conclusion, without further analysis at least, is that the sweet taste of saccharin generalizes to sugar. The aversion conditioning may have slightly increased the consumption of sour and bitter, but a conclusion concerning that question can wait for a later analysis.

FIGURE 1.2

This example illustrates several features of the scientific method. We began with a question - are the tastes of saccharin and sugar behaviorally the same for rats? Dr. Steward devised a method of answering the question, collected data, looked at the data, and then drew conclusions.

More involved in the research. The question about the similarity of the tastes of sugar and saccharin didn't arise in a vacuum. It fits into a broader context of the research are that Dr. Steward studies. Establishing the equivalence of two taste substances may appear earthshaking, and perhaps it's not all by itself. But suppose that Dr. Steward were searching for a way to teach California condors to avoid DDT. He might proceed the same way, this time looking for a harmless substance that tasted like DDT. He could use the same taste aversion procedure we've described; but in the larger context, the methodology seems important and useful.

In addition to illustrating a context for research, our example shows the empirical nature of science. Scientists seek evidence in data collected in experimental situations to provide support for theories.

Finally, our example illustrated a well-conducted experiment and the handling of data. We didn't describe the research in detail, but a great deal of care went into it. The situation where the data were collected was carefully controlled to avoid the disturbing influences of outside factors. The amounts and administration of saccharin and the lithium chloride were controlled and followed standard laboratory practices. the concentrations of the taste solutions were appropriate followed standard laboratory practices. The concentrations of the taste solutions were appropriate for the animals and for the question being asked (This is an important point; too high a concentration of saccharin tastes bitter.) The measurements of the amount consumed were made properly and the data were analyzed appropriately, allowing an inference to be drawn about the behavior of the rats. In our study of research design and statistics, all of these matters are of concern.

d. Research Design

Research design is the plan, structure, and the strategy of investigation so as to obtain answers to research questions and to control variance. Designs are worked out to enable the researcher to answer the research questions as validly, objectively, accurately, and economically as possible. Research designs provide answers to research questions. Research problems are stated in the form of a hypothesis. Research designs include the methods to control experimental, extraneous, and error variances.

e. What are Variables?

Variables are properties that can take on different values. There are several types of variables that students need to familiarize themselves with such as:

• an independent variable(X) is the variable manipulated by the researcher; an independent variable is the presumed cause.
• the dependent variable(Y) is the variable predicted to, whereas the independent variable is predicted form; the dependent variable is the presumed effect.

f. What is Variance?

Variance is the measure of the spread of data. There are several types of variance that must be considered to design a controlled experiment. These include:

• Error variance refers to unexplained variance; it does not mean that a mistake has been made; error variance is the fluctuation or varying of measures due to chance; error variance is unpredictable. Sources of error variance include: variance due to individual differences and variance associated with what is called errors of measurement - variation or responses from trial to trial, momentary inattention. Since these are not likely to be eliminated, the researcher tries to keep error variance to a minimum by reducing errors of measurement through controlled conditions and to increase the reliability of measurements.
• Experimental error or between-group error is the variance that reflects the differences between groups of individuals. It is variance introduced into the dependent variable by the independent variable of variables being manipulated. Design, plan, and conduct research so that the experimental conditions are as different as possible.
• Extraneous error are unwanted variables that the researcher can control. The control of extraneous variables means the influences of independent variables extraneous to the purposes of the study are minimized, nullified, or isolated. There are three ways to control extraneous variables: eliminate them by choosing subjects that are as homogeneous as possible; the second way to control extraneous variables is through randomization; whenever possible do so - randomly assign subjects to experimental conditions. The third way to control extraneous variables is to build the variable into the design as the independent variable.

g. Hypothesis Formation and Evaluation

General

The most common kind of hypothesis currently used in science is one we shall call the "glass-box" hypothesis. We use the name to emphasize the point that what is hypothesized is something that, at least in good part, can be seen. That is, it is open to direct examination. Suppose, for example, that we make the hypothesis that nerves are necessary for the control of some process. This is a typical glass-box hypothesis, since nerves are things accessible to observation and so is their absence. We may not be able to observe the act of control, but we can observe the presence of nerves, remove them, and note the consequent existence or disappearance of control.

The other kind of hypothesis we may call the "black-box" type. In black-box investigations we conceive of an underlying mechanism inaccessible to direct observation. Mendel's genetic determination and the mechanism for their segregation and recombination is an example of a black-box hypothesis. Sutton, of course, by pointing out the parallel between the behavior of Mendel's hypothetical units (genes)and the visible behavior of chromosomes in meiosis and fertilization, took it out of the class of black boxes, but this does not alter the fact that it as a black-box hypothesis when Mendel created it.

Unlike the first kind of hypothesis, the black-box type cannot, by definition, be verified "directly". Instead we seek necessary implication of the black-box mechanism, which can be looked for. From Mendel's mechanism, for example, we should get a 1:2:1 ratio when some F1 organisms are bred to one another. Again, if the hypothesis is a good one, we expect that two independent characteristics, both with a dominant alternate, will show in the ratio of 9:3:3:1, and so on.

Note that we said if the hypothesis is a good one some of the consequences necessarily implicated in it will be found. We did not say that discovery of implilcated consequences proves the hypothesis to be true. As many logicians have pointed out, there is no complete "verification" of a black-box hypothesis - only a demonstration of how much it will account for and how will it do. These - the how much and how well - are the measures of the goodness or usefulness of a hypothesis. It is usefulness, not truth, which we demand of the black-box hypothesis. Thus in physics we have, for example, the Bohr model of the atom, which accounts for a great many phenomena. We also have a more recent and a more sophisticated model that accounts for still more.

For both black box and glass-box kinds of hypotheses, the same logical form precedes and guides the experimental test of it same value. This is the "if...then..." logical linkage. For example: "If nerve connections are necessary for the normal function of tissue A, then severing the nerves to it should result in failure of that function." In either the glass-box hypothesis or the black-box case, then the "if...then..." logic tells the reader what data to look for.

Characteristics of a good hypothesis

1. It should be written as an "if...then..." statement; an interrogative statement about the relationship between variables.
2. It accounts for the know facts relating to the particular problem.
3. The variable you are studying are observable (operational).
4. The hypothesis makes specific and precise predictions.
5. It is testable.

Developing and Evaluating a Hypothesis

In the process of develop9ing a research proposal, don't panic! At the beginning, lack of knowledge and lack or originality is normal. You do not need to come up with a completely original, highly sophisticated and theoretically important research hypothesis. A reasonable expectation might be to come up with a research proposal that asks an interesting question; one that is derived from past literature.

In the research process, becoming familiar in an area that holds your interest is a fundamental step in the research process. As you investigate your interests:

1. Keep track of interesting variables and the relationships that may exist between them.
2. Can you locate studies that explore those relationships and find conflicting results? Are there relationships that were not explored? Maybe the investigator failed to consider a relationship that you feel is important? Or perhaps the research hasn't been done? These may be areas to pursue.
3. Start to formulate a research hypothesis involving an independent variable, a dependent variable, and a logical connection between them. Keep it simple! A common student problem is to have too many variables that seem interesting. Focus on one or a few. Remember hypotheses need to be stated because they represent the predicted relationship, which exists for the variables under study.

As you evaluate your hypothesis:

1. Have you defined your independent variable clearly, you dependent variable?
2. Is the logical link between the independent and dependent variable sound and clear? Is it in the for of "if...then..." logic? Does the relationship ask the question(s) that you are interested in? Does it predict? The investigator must formulate an interrogative sentence or statement asking about the relationship between two or more variables. This allows a hypothesis to be stated in such a way that it is capable of being refuted or confirmed.
3. Is the hypothesis testable? Remember in an experiment it is the hypothesis that is being tested and not the problem. Being practical helps here: do you have the skill and knowledge to test the hypothesis and secondly, do you have access to the physical and financial resources to carry out the study?
4. Last but not least! If you are using animals or human subjects, does you hypothesis cause any physical or psychological harm to your subjects? It would be unethical, for example, to study the effects of prenatal malnutrition on IG by deliberately providing pregnant women with inadequate diets.