How to write a Hypothesis
The second step in the research process is to formulate a hypothesis. The hypothesis is a tentative solution to a problem. The research activities are planned to verify the hypothesis and not to find out the solution to the problem or to seek an answer to a question.
It is very essential to a research worker to understand the meaning and nature of the hypothesis. The researcher always plans or formulate a hypothesis at the beginning of the problem. This Tutorial explores how to write a hypothesis with the best examples.
In outline, an example research program might proceed as follows. A researcher investigating algorithms might speculate as to whether it is possible to make better use of the cache on a CPU to reduce computational costs.
The preliminary investigation might lead to the hypothesis that a tree-based structure with poor memory locality will be slower in practice than an array-based structure with high locality, despite the additional computational cost.
The hypothesis suggests the research question of whether a particular sorting algorithm can be improved by replacing the tree structure with the array structure. The phenomenon that should be observed if the hypothesis is correct is a trend.
For example, as the number of items to be sorted is increased, the tree-based method should increasingly show a high rate of cache misses compared to the array-based method.
The evidence is the number of caches misses for several sets of items to be sorted. Alternatively, external evidence might be used, such as changes in execution time as the volume of data changes.
As this example illustrates the structure of the research program flows from having a definite research question and hypothesis.
A hypothesis or research question should be specific and precise, and should be unambiguous; the more loosely a concept is defined, the more easily it will satisfy many needs simultaneously, even when these needs are contradictory. And it is important to state what is not being proposed—what the limits on the conclusions will be.
The exercise of refining and clarifying a hypothesis may expose that it is not worth pursuing.
For example, if complex restrictions must be imposed to make the hypothesis work, or if it is necessary to assume that problems that are currently insoluble must be addressed before the work can be used, how interesting is the research?
A form of research where poor hypothesis seems particularly common is “black box” work, where the black box is an algorithm whose properties are poorly understood.
For example, some research consists of applying a black-box learning algorithm to new data, with the outcome that the results are an improvement on a baseline method. (Often, the claim is to the effect that “our black box is significantly better than random”.)
The apparent ability of these black boxes to solve problems without creative input from a scientist attracts research of low value. A weakness of such research is that it provides no insights into the data or the black box, and has no implications for other investigations.
In particular, such results rarely tell us whether the same behavior would occur if the same approach were applied to a different situation, or even to a new but similar data set.
That is, the results are not predictive. There may be cases in which it is interesting to observe the behavior of an algorithm on some data, but in general, the point of experimentation is to confirm models or theories, which can then be used to predict future behavior.
That is, we use experiments to learn about more general properties, a characteristic that is missing from black-box research.
However, the hypothesis should not follow the experiments. A hypothesis will often be based on observations, but can only be regarded as confirmed if it is able to make successful predictions.
There is a vast difference between an observation such as “the algorithm worked on our data” and a tested hypothesis such as “the algorithm was predicted to work on any data of this class, and this prediction has been confirmed on our data”.
Another perspective on this issue is that, as far as possible, tests should be blind. If an experiment and hypothesis have been fine-tuned on the data, it cannot be said that the experiment provides confirmation.
At best the experiment has provided observations on which the hypothesis is based. In other words: first hypothesize, and then test.
Where two hypothesis fit the observations equally well and one is clearly simpler than the other, the simpler should be chosen. This principle, known as Occam’s razor, is purely a convenience; but it is well-established and there is no reason to choose a complex explanation when another is available.
How to form or write a Hypothesis
The first stages of a research program involve the choice of interesting topics or problems, and the identification of particular issues to investigate. The research is given direction by the development of specific questions that the program aims to answer.
These questions are based on an understanding—an informal model, perhaps—of how something works, or interacts, or behaves.
They establish a framework for making observations about the object being studied. This framework can be characterized as a statement of belief about how the object behaves—in other words, a hypothesis.
Many hypothesis concern some aspect of the physical world: whether something is occurring, whether it is possible to alter something in a predictable way, or whether a model is able to accurately predict new events.
Astronomers use nuclear physics to predict the brightness of stars from their mass and chemical composition, for example, while a geneticist may seek to know whether substituting one gene for another can improve the health of a cell.
In computer science, some hypothesis is of this kind. We examine the limits of speech recognition, ask whether Web search can be used effectively by children, or predict how well a service will respond to increased load.
Other hypotheses are constructive. For example, we propose new technologies and explore their limitations and feasibility, or propose theorems that imply that there may be new solutions to long-standing algorithmic problems.
Regardless of field, if you wish to achieve robust research outcomes it is essential to have a hypothesis.
MEANING OF HYPOTHESIS
The word hypothesis is made up of two Greek roots which mean that it is some sort of ‘sub-statements’, for it is the presumptive statement of a proposition, which the investigation seeks to prove.
The scientist observes the man of a special class of phenomena and broads over it until by a flash of insight he perceives an order and intelligent harmony in it. This is often referred to as an ‘explanation’ of the facts he has observed. He has a ‘theory’ about the particular mass of fact.
This theory when stated testable proposition formally and clearly subjected to empirical or experimental verification is known as a hypothesis. The hypothesis furnishes the germinal basis of the whole investigation and remains to the end its cornerstone, for the whole research is directed to test it out by facts.
At the start of the investigation, the hypothesis is a stimulus to critical thoughts offers insights into the confusion of phenomena. At the end, it comes to prominence as the proposition to be accepted or rejected in the light of the findings. The word hypothesis consists of two words:
Hypo + thesis = Hypothesis
‘Hypo’ means tentative or subject to the verification and ‘Thesis’ means a statement about the solution to a problem.
The world meaning of the term hypothesis is a tentative statement about the solution to the problem. Hypothesis offers a solution to the problem that is to be verified empirically and based on some rationale.
Another meaning of the word hypothesis which is composed of two words: ‘Hypo’ means the composition of two or more variables which is to be verified. ‘Thesis’ means the position of these variables in the specific frame of reference.
This is the operational meaning of the term hypothesis. The hypothesis is the composition of some variables which have some specific position or role of the variables i.e. to be verified empirically.
It is a proposition about the factual and conceptual’ elements. The hypothesis is called a leap into the dark. It is a brilliant guess about the solution to a problem.
A tentative generalization or theory formulated about the character of phenomena under observation are called a hypothesis. It is a statement temporarily accepted as true in the light of what is known at the time about the phenomena. It is the basis for planning and action- in the research for new truth.
DEFINITIONS OF HYPOTHESIS
The term hypothesis has been defined in several ways. Some important definitions have been given in the following paragraphs:
Hypothesis – James E. Creighton
A tentative supposition or provisional guess “It is a tentative supposition or provisional guess which seems to explain the situation under observation.”
Hypothesis A Lungberg thinks
“A hypothesis is a tentative generalization the validity of which remains to be tested. In its most elementary stage the hypothesis may be any hunch, guess, an imaginative idea which becomes the basis for further investigation.”
Hypothesis: Shrewd Guess
According to John W. Best, “It is a shrewd guess or inference that is formulated and provisionally adopted to explain observed facts or conditions and to guide in further investigation.”
Hypothesis: Guides the Thinking Process
According to A.D. Carmichael, “Science employs hypothesis in guiding the thinking process. When our experience tells us that a given phenomenon follows regularly upon the appearance of certain other phenomena, we conclude that the former is connected with the latter by some sort of relationship and we form a hypothesis concerning this relationship.”
Hypothesis A proposition is to be put to test to determine its validity:
Goode and Han, “A hypothesis states what we are looking for. A hypothesis looks forward. It is a proposition which can be put to a test to determine its validity. It may prove to be correct or incorrect.
Hypothesis An expectation about events based on generalization:
Bruce W. Tuckman, “A hypothesis then could be defined as an expectation about events based on a generalization of the assumed relationship between variables.”
Hypothesis A tentative statement of the relationship between two or more variables:
“A hypothesis is a tentative statement of the relationship between two or more variables. the hypothesis is always in declarative sentence form and they relate, either generally or specifically variable and variables.”
Hypothesis A theory when it is stated as a testable proposition.
M. Verma, “A theory when stated as a testable proposition formally and clearly and subjected to empirical or experimental verification is known as a hypothesis.”
ASSUMPTION, POSTULATE, AND HYPOTHESIS
The terms assumption. Postulate and hypothesis occur most frequently in the research literature but are often confused by research scholars. Hence these terms need clear explanation.
Assumption means taking things for granted so that the situation is simplified for a logical procedure. Assumptions are not the very ground of our activity as the postulates are.
They merely facilitate the progress of an agreement a partial simplification by introducing restrictive conditions. For example, the formulas of Statistics and measurement are based on a number of assumptions. Assumption means restrictive conditions before the argument can become valid.
Assumptions are made on the basis of logical insight and their truthfulness can be observed on the basis of data or evidence.
The postulates are the basis and form the original point of an argument whereas assumptions are a matter of choice and less use, we make them more free will and our argument be a general proposition or convention.
Postulates are the working beliefs of most scientific activity. The mathematician begins by postulating a system of numbers which range from 0 to 9 and can permute and combine only thereafter. Similarly ‘Hull’s Theory of Reinforcement’s is based on eight postulates of the behavior of an organism.
With many people, God and Spirit is a postulate of the good life or godly life. Postulates are not proven; they are simply accepted at their face value so that their basic work for the discovery of other facts of nature can begin.
A hypothesis is different from both of these. It is the presumptive statement of a proposition which the investigator seeks to prove. It is a condensed generalization.
This generalization requires knowledge of principles of things or essential characteristics which pertain to an entire class of phenomena. The theory, when stated as a testable proposition formally and clearly and subjected to empirical or experimental verification, is known as a hypothesis.
The hypothesis furnishes the germinal basis of the whole investigation and remains to test it out by facts.
The hypothesis is based on some earlier theory and some rationale whereas postulates are taken as granted true. An assumption is the assumed solution of a major problem. It may be partially true.
The scientific research process is based on some hypothesis. The natural sciences and mathematics are based on postulates. The statistic is based on some assumptions which are considered approximate science. The assumptions are helpful in conducting a research work in behavioral sciences.
OBSERVATION VERSUS SPECIFIC LAND GENERAL HYPOTHESIS
hypothesis is often confused with observation. These terms refer to quite different things. An observation refers to what is….that is to what is seen. From observation, the researcher may infer. For example, a researcher may go into a school and after looking around.
Observe that most of the students are back. From that observation, he may infer that the school is located in a poor neighborhood. Though the researcher does not know that the neighborhood is poor, he expects that the majority of people living there are poor.
Then he has formulated a specific hypothesis setting forth an anticipated relationship between two variables like race and income level. For the test of this hypothesis the researcher could walk around the neighborhood, observes the home and the income levels.
After observation, he provides support for this specific hypothesis for this researcher might make a general hypothesis.
The second hypothesis represents a generalization and must be tested by making an observation as was the specific hypothesis. Since it would be impossible to observe all universe or population, thus, the researchers will take a sample and reach conclusion on a probability basis for the verification of the hypothesis being true or not.
There is some difference between specific and general hypothesis. The specific hypothesis requires fewer observations for testing than the general hypothesis. For testing purpose, a general hypothesis is reformulated to a more specific one.
NATURE OF HYPOTHESIS
The following are the main features of a hypothesis:
1. It is conceptual in nature. Some kind of conceptual elements in the framework is involved in a hypothesis.
2. It is a verbal statement in a declarative form. It is a verbal expression of ideas and concepts, it is not merely an idea but in the verbal form, the idea is ready enough for empirical verification.
3. It has the empirical referent. A hypothesis contains some empirical referent. It indicates the tentative relationship between two or more variables.
4. It has a forward or future reference. A hypothesis is future-oriented. It relates to future verification, not the past facts and informations.
5. It is the pivot of a scientific research. All the research activities are designed for its verification.
The nature of the hypothesis can be well understood by differentiating it with other terms like assumption and postulate.
FUNCTIONS OF HYPOTHESIS
The following are the main functions of hypothesis in the research process suggested by H.H. Mc. Ashan :
1. It is a temporary solution of a problem concerning with some truth which enables an investigator to start his research works.
2. It offers a basis in establishing the specifics of what to study for and may provide possible solutions to the problem.
3. Each hypothesis may lead to formulating another hypothesis.
4. A preliminary hypothesis may take the shape of the final hypothesis.
5. Each hypothesis provides the investigator with a definite statement which may be objectively tested and accepted or rejected and leads for interpreting results and drawing conclusions that are related to the original purpose.
The functions of a hypothesis may be condensed into three. The following are the threefold functions of a hypothesis:
(a) To delimit the field of the investigation.
(b) To sensitize the researcher so that he should work selectively, and have a very realistic approach to the problem.
(c) To offer the simple means for collecting evidence to the verification.
IMPORTANCE OF A HYPOTHESIS
1. The hypothesis as the Investigator’s “Eyes”: Carter V. Good thinks that by guiding the investigator in the further investigation it serves as the investigator’s “Eyes” in seeking answers to tentatively adopted generalization.
2. It Focuses Research: Without it, research is unfocussed research and remains like a random empirical wandering. It serves as a necessary link between theory and the investigation.
3. It Places Clear and Specific Goals: A well thought out set of the hypothesis is that they place clear and specific goals before the research worker and provide him with a basis for selecting sample and research procedure to meet these goals.
4. It Links Together: “It serves the important function of linking together related facts and information and organizing them into wholes.” – Good Barr and Scates
5. It Prevents Blind Research: “The use of hypothesis prevents a blind search and indiscriminate gathering of masses of data which may later prove irrelevant to the problem under study.”– P. V. Young
6. As a Sort of Guiding Light: A hypothesis serves as a powerful beacon those lights the way for the research work.
George J. Mosley thinks that hypothesis serves the following purposes:
1. They provide direction to research and prevent the review of the irrelevant literature and the collection of useful or excess data
2. They sensitize the investigator certain aspects of the situation which are irrelevant from the standpoint of the problem at hand.
3. They enable the investigator to understand with greater clarity his problem and its ramification.
4. They serve as a framework for the conclusive-in short a good hypothesis:
Gives help in deciding the direction in which he has to proceed.
It helps in selecting pertinent fact.
It helps in drawing conclusions.
D.B. Van Dalen advocates the Importance of Hypothesis in the following ways:
1. hypothesis are indispensable research instrument, for they build a bridge between the problem and the location of empirical evidence that may solve the problem.
2. A hypothesis provides the map that guides and expedites the exploration of the phenomena under consideration.
3. A hypothesis pinpoints the problem. The investigator can examine thoroughly the factual and conceptual elements that appear to be related to a problem.
4. Using hypothesis determines the relevancy of facts. A hypothesis directs the researcher’s efforts into productive channels
5. The hypothesis indicates not only what to look for is an investigation but how to obtain data. It helps in deciding research design. It may suggest what subjects, tests, tools, and techniques are needed.
6. The hypothesis provides the investigator with the most efficient instrument for exploring and explaining the unknown facts.
7. A hypothesis provides the framework for drawing conclusions.
8. These hypotheses simulate the investigator for further research studies.
These hypotheses contain variables which must be labeled and then operationally defined to construct predictions. These steps might be considered the logical stages of the research.
These stages are followed by methodological stages, which culminate in the development of research design and the development of measures and finally in the finding themselves.
KINDS OF hypothesis
hypothesis vary in form and some extent, the form is determined by some function. Thus a working hypothesis or a tentative hypothesis is described as the best guess or statement derivable from known or available evidence.
The amount of evidence and the certainty or quality of it determines other forms of the hypothesis. In other cases, the type of statistical treatment generates a need for a particular form of a hypothesis.
The following kinds of hypothesis and their examples represent an attempt to order the more commonly observed varieties as well as to provide some general guidelines for hypothesis, development, and statement.
There are four kinds of hypothesis: (a) Question (b) Declaration Statement (c) Directional Statement and (d) Null form or Non-Directional.
(a) Question form of hypothesis:
Some writers assert that a hypothesis may be stated as a question, however, there is no general consensus on this view. At best, it represents the simplest level of empirical observation.
In fact, it fails to fit most definitions of the hypothesis. It is included here for two reasons: the first of which is simply that it frequently appears in the lists.
The second reason is not so much that question may or may not qualify as a hypothesis. There are cases of simple investigation and search which can be adequately implemented by raising a question, rather than dichotomize hypothesis forms into acceptable/ rejectable categories.
The following example of a question is used to illustrate the various hypothesis forms: Is there a significant interaction effect of schedule of reinforcement and extroversion on learning outcomes?
(b) Declarative Statement:
A hypothesis may be developed as a declarative which provide an anticipated relationship or difference between variables. The anticipation of a difference between variables would imply that the hypothesis developer has examined existing evidence which led him to believe a difference may be anticipated as processes additional evidence.
The following is an example of this form of hypothesis-
H: There is a significant interaction effect of schedule of reinforcement and extroversion on learning outcomes.
It is merely a declaration of the independent variables effect on the criterion variable.
(c) Directional Hypothesis:
A hypothesis may be directional which connotes an expected direction in the relationship or difference between variables. The above hypothesis has been written in directional statement form as follows:
H: Extrovert learns better through an intermittent schedule of reinforcement whereas introvert learns through a continuous schedule of reinforcement.
The hypothesis developer of this type appears more certain of his anticipated evidence that would be the case if he had used either of the previous examples. If seeking a tenable hypothesis is the general interest of the researcher, this kind of hypothesis is less safe than the others because it reveals two possible conditions.
These conditions are a matter of degree. The first condition is that the problem of seeking a relationship between variables is so obvious that additional evidence is scarcely needed.
The second condition derives because the researcher has examined the variables very thoroughly and the available evidence supports the statement of particular anticipated outcomes.
An example of the obviously safe hypothesis would be ‘hypothesis’ that high intelligence students learn better than low intelligent students. The above hypothesis is in the directional statement form but it requires evidence for the relationship between these two variables reinforcement and personality.
(d) Non-Directional Hypothesis:
A hypothesis may be stated in the null form which is an assertion that no relationship or no difference exists between or among the variables.
This form null hypothesis is a statistical hypothesis which is testable within the framework of probability theory. It is also a non- directional form of a hypothesis. The following are the examples of the null form of hypothesis
H0: There is no significant interaction effect of schedule of reinforcement and extroversion on learning outcomes.
H0: There is no significant relationship between intelligence and achievement of students. A recent trend is to employ or develop a null hypothesis in research work of education and psychology.
A null hypothesis accepted is tentative to stating that on the basis of evidence tested it could be that there is no difference. If the null hypothesis is rejected, there is a difference but we do not know the alternative or the differences.
In this form of a hypothesis, the researcher has not to anticipate or give the rationale for the declaration or directional form. Secondly, it does not make the researcher biased or prejudiced. He can be objective about the expected outcomes of the research or findings.
Actually, this is a form of hypothesis but is a statistical hypothesis which is self explanatory-null hypothesis means zero hypothesis. A researcher has not to do anything in developing such form of a hypothesis.
In the process of reflective thinking research hypothesis is the second step whereas a null hypothesis is the fifth step of the research process.
In order to accommodate the object of the inquiry for extracting this information, a null hypothesis is an appropriate form. A null hypothesis does not necessarily reflect the expectations of the researcher so much as the utility of the full form as the best fitted to the logic of chance in statistical knowledge or science.
A statistical hypothesis must be testable within the framework of probability theory. The theory requires one or the other of two hypothesis forms: the ‘null’ form and the ‘delta’ form.
The full form is the no different form i.e. there is no difference or relationship between or among variables under certain conditions.
The ‘delta’ form for a test hypothesis is simply that A is greater or lesser than B under conditions C, D…..A null form or a delta form which specifies only A and B variables in the relationship permits only a bivariate analysis which is not a very sophisticated research analysis.
The development of computer-assisted data analysis permits the manipulation of a number of variables represented in the C, D..., n conditions of the difference or relationship.
The general hypothesis is the second step and the null hypothesis is the fifth step of the research process. Null hypothesis provides the basis of accepting or rejecting the general hypothesis.
General Hypothesis Programmed instruction is effective than the traditional method in terms of learning outcomes. General Hypothesis: Structural method is more effective than the lecture method of teaching in English.
Null Hypothesis is a statistical hypothesis which is used in analyzing the data. It assumes that the observed difference is attributable by sampling error and true difference is zero.
The statistical tests of significance are used to accept and reject the null hypothesis. If it is rejected. The general hypothesis is accepted.
Occam’s razor has given a principle of economy in the scientific explanation which requires for given set of observations so that generalization can be made.
This class of hypothesis is known as the null hypothesis so-called because of it ‘nullifies’ the positive argument of the findings or non-directional statement of the generalization.
This type of hypothesis is also termed as a statistical hypothesis or non-directional hypothesis or zero hypothesis because it denies the existence of any systematic principles apart from the effect of chance. This hypothesis assumes that no or zero difference exists between the two population means or the treatments.
The symbol He represents the Null-Hypothesis. An alternative formulation of the hypothesis is to assert’s that the two samples drawn from a population having the same mean. The null hypothesis is a trial hypothesis asserting that no difference exists between population parameters. Thus it involves two types of errors.
Two Types of Errors
Type I error (an error): When an alternative hypothesis H1 may be accepted and H0 is rejected. It shows that the obtained difference exists and not due chance or sampling errors.
Type II error (b error): When null hypothesis H1 is accepted and alternative hypothesis H1 is rejected. It indicates that the obtained difference is due to chance or sampling error.
CHARACTERISTICS OF A GOOD HYPOTHESIS
A good hypothesis must possess the following main characteristics:
1. A good hypothesis is in agreement with the observed facts.
2. A good hypothesis does not conflict with any law of nature which is known to be true.
3. A good hypothesis is stated in the simplest possible term.
4. A good hypothesis permits the application of deductive reasoning.
5. A good hypothesis shows very clear verbalization. It is different from what is generally called a hunch.
6. A good hypothesis ensures that the methods of verification are under the control of the investigator.
7.A good hypothesis guarantees that available tools and techniques will be effectively used for the purpose of verification.
8. A good hypothesis takes into account the different types of controls which are to be exercised for the purpose of verification.
9. A good hypothesis ensures that the sample is readily approachable.
10. A good hypothesis indicates clearly the role of different variables involved in the study.
11. A good hypothesis maintains a very apparent distinction with what is called theory law, facts, assumption, and postulate.
VARIABLES IN A HYPOTHESIS
A hypothesis is made testable by providing operational definitions for the terms or variables of the hypothesis. For a testable hypothesis, there are two important things: Variables, and Operational definitions
There are five types of variables. Among students of the same age and intelligence, skill performance is directly related to the number of practice traits particularly among boys but less directly among girls. In such a hypothesis the variables which must be considered are:
(i) Independent variable – the number of practice trails.
(ii) Dependent variable – skill performance.
(iii) Moderator variable – sex.
(iv) Control variable – age, intelligence.
(v) Intervening variable – learning.
(i) The Independent Variable :
The independent variable which is a stimulus variable or input operates either within a person or within the environment to affect his behavior. It is that factor which is measured, manipulated. or selected by the experimenter to determine its relationship to observed phenomena.
If a researcher is studying the relationship between two variables X and Y. If X is the independent variable, then it affects another variable Y: So the characteristics of the independent variables are:
(a) It is the cause for the change in other variables
(b) Independent variables are always interested only it affects another variable, not in what affects it.
(ii) The Dependent Variable :
The dependent variable is the response variable or output. It is an observed aspect of the behavior of an organism that has been stimulated. The dependent variable is that factor which is observed and measured to determine the effect of the independent variables.
It is the variable that will change as a result of variations in the independent variable. It is considered dependent because its value depends upon the value of the independent variable. It represents the consequence of a change in the person or situation studied.
The relationship between Independent and Dependent Variables: Most experiments involve many variables when two continuous variables are compared, as in correlation studies, deciding which variable to call independent and which dependent is sometimes arbitrary.
In such cases, variables are often not labeled as independent or dependent since there is no real distinction. Independent variables may be called factor and their variation may be called levels.
(iii) The Moderator Variable:
The term moderator variable describes a special type of independent variable a second independent variable selected for study to determine if it affects the relationship between the primary independent variable and the dependent variable.
The moderator variable is defined as that factor which is measured, manipulated or selected by the experimenter to discover whether it modifies the relationship of the independent variable to observed phenomena. The sex and rural-urban generally function as moderator variables.
(iv) Control Variable :
All the variables in a situation cannot be studied at the same time, some must be neutralized to guarantee that they will not have a differential or moderating effect on the relationship between the independent and dependent variables. These variables whose effects must be neutralized or controlled are known as control variables.
They are defined as those factors which are controlled by the experimenter to cancel out or neutralize any effect they might otherwise have on the observed phenomena. While the effects of the control variables are neutralized, the effect of moderator variables is studied.
Certain variables appear repeatedly as control variables, although they occasionally serve as moderator variables. For example sex, intelligence and socio-economic status are three subject variables that are commonly controlled, noise, task order and task content are common control variables in the situation.
(v) Intervening Variable:
Each independent, moderator, and control variable can be manipulated by the experimenter and each variation can be observed by him as it affects the dependent variable. Often these variables are not concrete but hypothetical, the relationship between a hypothetical underlying or intervening variable and dependent variable.
An intervening variable is that factor which affects the observed phenomenon but cannot be seen and measured or manipulated, Its effect must be inferred from the effects of the Independent and moderator variables on the observed phenomena. The attitude, learning process, habit and interest function as Intervening variables.
Teachers given more positive feedback-experiences will have more positive attitudes towards children than teachers given fewer positive feedback-experiences.
Independent Variable–Number of positive feedback experiences for the teacher.
Intervening Variable–Teacher’s self-esteem or habit pattern.
Dependent Variable–Possessiveness of teacher ’s attitude towards students.
The researcher must operationalize his variables in order to study them and conceptualize his variables in order to generalize from them. Researchers often use the labels independent, dependent, moderator, and control to describe operational statements of their variables.
The intervening variables always refer to a conceptual variable that which is being affected by the independent, moderator control and dependent variables.
The intervening variable can often be discovered by examining a hypothesis. They are usually abstract in nature.
Independent, moderator and control variables are inputs or causes, the first two being these that are studied while the third, control variables are neutralized or ‘eliminated’.
At that other end, dependent variables represent effects or it is also known as criterion variable while intervening are conceptualizations which intervene between operationally stated causes, and operationally stated effects.
The Research Variables Combined
The various research variables interact among themselves. The Independent, moderator, and control variables are under the researcher’s control. They cause an impact on the subject. The impact is referred to as the intervening variable. In addition, the extraneous variables have an impact upon this Intervening variable.
Such extraneous variables are not under the researcher’s control, their presence weakens a study. One of the goals of a researcher is to remove as many significant factors as possible from the extraneous variables category by bringing them into the categories of moderator and control variables. Such a process of removing extraneous variables strengthens a study.
The Variables in the Research Process Causes
The intervening variable is merely hypothesized. It is abstract in nature. It cannot be visually observed. It is defined in conceptual terms. It is produced by some combination of the causal variables in the top row of the figure. It produces the effector dependent variable.
Every experimental study has at least one independent variable and one dependent variable. Both of these variables should be explicitly stated in the hypothesis and in the research predictor.
If either the treatment or the outcome variable is too complex to be stated succinctly, further operational definitions of these variables can be included in the methods section of a report. Every study has also an intervening variable, and often there can be more than one intervening variable.
The intervening variable is, not stated in operational terms, but the rattier conceptual explanation for the observed results. Intervening variables are normally not stated in the hypothesis or research prediction. Sometimes intervening variables are only vaguely described or are not mentioned at all.
Every study does not contain moderator and control variables. When such variables are contained in a study, they should be operationally defined. Moderator and control variables are stated in the research hypothesis and in the research prediction. Often the operational definitions further explanation in the method section of a report.
Some Considerations For Variables Choice
After selecting the independent and dependent variables the researcher must decide which variables are to be included as moderator variables and which are to be excluded or hold constant as Control variables. He must decide how to treat the total part of the other variables (other than the independent).
That might affect the dependent variables. In making these decisions which variables are ‘in’ and which are ‘out’ he should take into account three kinds of considerations:
1. Theoretical Consideration:
In treating as a moderator variable, the researcher learns how it interacts with the independent variable to produce differential effects on the dependent variable. In term of theoretical base researcher is working and in term of what he is trying to find out in a particular experiment, certain variables highly qualify as the moderator variables.
In choosing a moderator variable a researcher should ask: Is the variable related to the theory with which I am working? How helpful would it be to know if an interaction exists? How likely is there to be an interaction?
2. Design Consideration
The questions which relate to the experimental design which has been chosen and its adequacy for controlling for sources of bias, the researcher should ask the following question: Have my decision about moderator and control variables met the requirements of experimental design in terms of dealing with the source of validity?
3. Practical Consideration
A researcher can only study so many variables at one time. There are limits to human and financial resources and the deadlines he can meet.
By their nature, some variables are harder to study than to neutralize, while others are as easily studied as neutralized. In dealing with practical considerations, the researcher must ask a question like the following:
How difficult is it to make a variable a moderator as opposed to a control variable?
What kinds of resources are available and what kinds are required to create moderator variables?
This is a highly significant one. In educational experiment researchers often have less control over the situation than design and the retired considerations might necessitate.
Immediately upon completion of the testable hypothesis, a researcher should examine them and the problem, in general, to determine if there are any terms which may be abstract or misleading.
‘If an is found particularly in the testing hypothesis, they should be defined to make them completely operational for the study being undertaken.
The necessity for operational definitions does not mean that the researcher can define a term to mean whatever he cares to make it mean, but does enable the researcher to limit the meaning of a word. The operational definition should be more specific than those used in ordinary discourse.
In other words, any special term which must be used in the statement of the problem may require an operational definition to ensure clarity. Particular clarification should be given terms which are used in the formulation of a testable hypothesis.
The term selected must be useful and make sense. Even common adjectives may be used if you adequately explain what you mean. A point to remember is that once a researcher makes a definition, he must stick to it.
Words which may need defining are those which appear ambiguous, which have confusing interpretation and which might make a difference to a person attempting to replicate the study.
The Conditions for Making Final Decision
H.H. McAshan suggests that new researcher check the following conditions for operationally defining words before making a final decision:
1. The definition decides upon must withstand subjective analysis to determine if other qualified people could look at the word involved and come to the same conclusion.
2. The reliability of each word should be checked to find out if the subjective judgments are consistent.
3. The meaning of the operationally defined words must be mutually exclusive and not synonymous with other words, terms or, expressions.
4. The definition of each word chosen must fit the material researchers intend to study.
5. The definition decides upon must include all situations of use which will be included in the course of the investigation.
An operational definition is a definition based on the observable characteristics of that which is being defined. The word ‘observable’ in the significant word in describing an operational definition.
There are three approaches to constructing operational definitions:
(i) Type A,
(ii) Type B, and
(iii) Type C.
Type ‘A’ Operational Definition: The ‘Type A’ operational definition can be constructed in terms of the operations that must be performed to cause the phenomenon or state than an object or thing. It tells what manipulation to use to induce a particular state.
They are useful in defining independent variables as prescriptions carried out by the experimenter. The same variable, of course, is operationally defined by more than one type of definition but when what variable is the independent variable. It is often the most useful.
‘Type B’ Operational Definition: The ‘Type B’ operational definition can be constructed in terms of how the particular object or thing being defined operates, that is, what it does or what constitutes its dynamic properties.
‘Type B’ operational definitions see particularly appropriate in an educational context for describing a type of person.
Though they may be used to define other variables, Type B definitions are particularly useful for defining the dependent variable when it is to be operationally based on behavior.
‘Type C’ Operational Definitions: The ‘Type C’ operational definition can be constructed in terms of what the object or phenomenon being defined looks like, that is, what constitutes its static properties.
An intelligent student can be defined as a person who has a good memory, a large vocabulary, good reasoning ability, good arithmetic, skills etc. This type of operational definitions utilizes observable structural properties of the object.
It describes the qualities, traits, or characteristics of people or thing. Thus, they may be used for defining any type of a variable when used for defining a person’s characteristics; they specify the static or internal qualities rather than his behavior as does the ‘Type B’ definition.
‘Type C’ operational definitions often lend themselves to measurement by tests although the ability to be tested is in requisite part of the definition. The testability if any hypothesis depends on whether suitable operational definitions can be constructed for its variables.
ROLE OF HYPOTHESIS
The hypothesis is the basis of a scientific investigation education. It is the pivot of the research process. All the research activities are oriented towards the verification of the hypothesis.
Apart from this role, it has a significant role in the formulation of theory, principles, and laws. It is also known as a tentative theory, after verification, it takes the shape of a final theory.
A theory embers new hypothesis, these are subjected to verification, after the verification it becomes a new theory in field studies. In building up the theories, this cyclic process continues. It has been illustrated with the help of a diagram.
SOURCES OF hypothesis
The hypothesis is originated from essentially the same background that serves to reveal the problem. These sources are namely theoretical background, knowledge, insight, and imagination that come from instructional programmed and wide reading experiences, familiarity with existing practices.
The major sources of the hypothesis are given below:
1. Specialization of an educational field.
2. Programmed of reading: Published studies, abstracts research journals. Handbooks, seminars on the issue, current trends in the research area.
3. Instructional programmers persuaded.
4. Analyze the area studied.
5. Considering existing practices and needs.
6. Extension of the investigation.
7. Offshoots of research studies in the field.
Process for the Formulation of Theories
Researcher employs these sources for formulating a hypothesis of his investigation. He has to use two logical processes to drawn upon in developing a hypothesis. The processes are known as:
(a) Deductive thinking, and (b) Inductive thinking.
(a) The deduction is a process which goes from the general to the specific:
When general expectations about problems or events based on presumed relationships between variables are used to arrive at more specific expectations, that process is called deduction.
(b) Induction is a process which goes from the specific to the general:
In the induction process, the researcher starts with specific observations and combines them to produce a more general statement of relationship namely a hypothesis.
Many researchers begin by searching the literature for relevant specific findings in order to induce a hypothesis, and other often run a series of exploratory studies before attempting to induce a hypothesis.
Induction begins with data’ and observations or empirical events and proceeds toward hypothesis and theories, while deduction begins with theories and general hypothesis and proceeds towards a specific hypothesis.
From any problem statement, it is generally possible to derive more than one hypothesis. There is three simple hypothesis generated from this problem to determine, “the effect of” massive, positive, verbal rewards on the reading achievement of children.”
At first glance this three hypothesis might be offered:
(A) Reward Increases reading achievement.
(B) Reward decreases reading achievement.
(C) The reward has no effect on reading achievement.
Evidence has already been obtained in the laboratory to support the hypothesis (A) rewards increase performance. However, upon closer examination, the primary purpose of this study is to determine whether the enhancing effect of rewards can be incorporated into a classroom setting to facilitate children’s learning to read.
This theory is based on the assumption that the ‘law’ of learning should apply in the classroom. If perhaps more subtly than in a laboratory and on the laboratory findings that support the assumed relationship between reward and performance.
The logical conclusion would be that rewards would have a demonstrable enhancing effect on classroom performance. This conclusion is based on the first assumption arrived at deductively and the second arrived at inductively.
Both induction and deduction. are needed to choose among the possibilities. Many theories, both psychological and educational deal with stabilization (and rigidifying) of behavior patterns as a function of their use.
Researchers formulate hypothesis using induction and deduction, one of the goals of the researcher is to produce that piece for generalizable bodies of theory which will provide answers to practical problems. Hypothesis construction and testing enable researchers to generalize their findings beyond the specific conditions which they were obtained.
Since a hypothesis is a formulation of anticipated findings, students are advised to develop a hypothesis as a means of demonstrating the basis for their study to themselves and their reader. The task of introducing a study and discussing the findings are facilitated by the existence of a hypothesis.
FORMULATION OF TESTABLE HYPOTHESIS
A hypothesis is a tentative assumption drawn from knowledge and theory which is used as a guide in the investigation of other facts and theory that are as yet unknown.
The hypothesis formulation is one of the most difficult and most difficult steps in the entire scientific process. A poorly chosen or poorly worded hypothesis can prevent:
(a) The obtaining of enough pertinent data,
(b) The drawing of conclusions and generalizations, and
(c) The application of certain statistical measures in the analysis of the result.
It is impossible to over-emphasize the role of the hypothesis in research. It is the central core of study that directs the selection of the data to be gathered, the experimental design, the statistical analysis, and the conclusions drawn from the study.
A study may be devoted to the testing of one major hypothesis, a number of subsidiary hypothesis, or both major and subsidiary hypothesis. When several hypotheses are used, each should be stated separately in order to anticipate the type of analysis required and in order to definitely accept or reject each hypothesis on its own merit.
Regardless of the number or type of hypothesis used, it is extremely important that each be specifically testable, and based upon a logical foundation. Hildreth Hoke McAshan says only one possible exception to the above statements, which is that when fact-finding alone is the primary aim of the study, it may not always be necessary to formulate an explicit hypothesis. However, this need not be a concern of most scientific researchers.
FUNDAMENTAL BASES OF HYPOTHESIS
The researcher deals with reality on two levels,
(a) The operational level, and
(b) Conceptual level.
On the operational level researcher must define events in observable terms in order to operate with the reality necessary to do researches.
On the conceptual level, he must define events in terms of underlying commonality (usually causal) with other events. Defining at a conceptual level, the researcher can abstract from a single specific instance to general dues and thus, begin to understand how phenomena operate and variables interrelate.
The formulation of a hypothesis very frequently requires going from the operational or concrete level to the conceptual or abstract level. It is this movement to the conceptual level which enables that the result to be generalized beyond the specific conditions of a particular study and thus to be of wider applicability.
Research requires the ability to move from the operational to the conceptual level and vice-versa. This ability is required not only in constructing experiments but in applying their findings as well.
Consider a hypothetical study in which programmed instruction is being compared to traditional instruction. The term ‘Programmed Instruction” and “Traditional Instruction” are operational terms.
These operational terms should be examined for underlying conceptual similarities and differences. This process of making conceptual contrasts between operational programmed is called conceptualization or dimensionalization.
Dimensions useful for contrasting programmed and traditional instruction might be a degree of feedback, the rate of positive reinforcement, the uniqueness of presentation format, control of pacing size of instructional units and degree of incorporation of student performance feedback in instructional design.
These six dimensions or concepts could be used for classifying any instructional model as a basis for understanding its relation to other models.
Such classification at this abstract level would help one not only hypothesize whether instructional ‘model A’ will be more effective than ‘model B’ on certain specific criteria but to begin to understand why ‘model B’ is better and thus to be able to build ‘model A’ into other instructional procedures.
Moving from the operational to the conceptual level and vice-versa is a critical ingredient of the research to demonstration process.
Difficulties in the Formation of Useful Hypothesis: The following are the difficulties in the formation of the hypothesis:
1. The absence of knowledge of a clear theoretical framework.
2. Lack of ability to make use of the theoretical framework logically.
3. Lack of acquaintance with available research technique resulting in failure to be able to phrase the hypothesis properly.
Testing the Hypothesis
The evidence of the work of hypothesis lies in its abilities to meet the test of its validity. The purpose of testing a hypothesis is to determine the probability that it is supported by fact.
Because a hypothesis is a general expectation about the relationship between variables there is an extremely large number of instances under which it can be tested, and it would be impractical to attempt to gain support in all of these instances. The validity of a hypothesis is established in two stages:
1. The statement of hypothesis allows the investigator to develop deduction and certain implications which when stated in operational terms can lead to rejection of the hypothesis that is in conflict with accepted knowledge at the logical level.
For example, a hypothesis which says, for instance, that nondirective teachers are more effective than directive teachers would have to be tested for many groups of teachers.
In many subjects and many settings, and with many criteria before it could be accepted. If on the basis of limited testing the hypothesis fails to yield confirming results, then it would be fair to reject it.
2. If a hypothesis passes the test of logic, it then must be subjected to an empirical test, perhaps through an experiment or a series of measurement. The hypothesis that boys are stronger or taller than girls, for example, can be verified through measurements.
A hypothesis is never proved it is merely sustained or rejected. If it fails to meet the test of its validity, it must be modified or rejected.
The confirmation of a hypothesis, on the other hand, is always, tentative and relative, subject to later revision and even rejection as further evidence appears or as more adequate hypothesis are introduced. The form of the hypothesis to be tested can be very controversial. The null form’ is probably preferred by most experienced research personnel.
The null hypothesis states that there is no difference between the two groups or treatments. It is generally used to spell out what would be the case if the null hypothesis were true. The no difference statement assumes that the two groups will be tested and found to be equal.
FORMAL CONDITIONS FOR TESTING hypothesis
There are two types of hypothesis statements:
(a) The null hypothesis, and
(b) Hypothesis prediction form
Whether the experimenter chooses the hypothesis prediction or the null form, there are certain formal conditions which must be met in order for the hypothesis to be considered testable. These are listed below:
1. It must be stated so that deductions can be made from it and so that decisions can be reached as to whether or not it explains the facts being considered.
2. It should be worded clearly and unequivocally in operational terms. This should leave no doubt as to what action, what prediction, what quality or quantity, or who is involved?
3. It must be capable of being refuted. There must be some comparisons possible which will allow the researcher to give either a ‘yes’ or ‘no’ answer to the hypothesis stated.
4. It should be specific and testable, with all predictions and operations to be tested spelled out.
5. It should have simplicity. If it is too complex, consideration should be given to dividing it into sub-hypothesis.
6. It should be directly related to the empirical phenomena.
7. It must be stated in final form early in the experiment before any attempt at verification is made.
8. It should be so designed that its test will provide an answer to the original problem which forms the primary purpose of the investigation.
9. It must be related to available techniques of design procedure, and statistical analysis.
10. It should be related to available knowledge or theory concerning the original problem area.
The statement of the problem, review of the literature, and other planning of early stages of a project are largely performed so as to enable the researcher to arrive at a good, clearly stated, testable hypothesis.
CRITERIA FOR EVALUATING HYPOTHESIS
Some hypothesis is considered more satisfactory than others. The following are the serious considerations of a satisfactory hypothesis and these criteria may be helpful to make this judgment
1. Plausibility of Explanation
Several criteria are involved in establishing the plausibility of explanations. A satisfactory hypothesis should have relevant and logical possibility about the relationship of variables included in them.
2. Testability of Explanation
The variables should be defined operationally and the predicted relations among them can be tested empirically. The variables of the hypothesis should be measurable or quantifiable. The suitable measuring instrument is available or it can be considered easy.
3. Adequacy of Scope
The most useful hypothesis explain all the facts that are relevant to the phenomena being explained and contradict none of them. The broader the scope of a theory, the more valuable it is.
The more consequences that a hypothesis yields, the greater is its fruitfulness. A hypothesis is of greater value if it establishes a generalization that can be applied in many areas of education or in many fields.
The most satisfactory hypothesis not only explain all the known facts that gave rise to the original problems but also enable scientists to make predictions about as yet unobserved events and relationships.
4. Usefulness of False hypothesis
hypothesis need not be the correct answers to problems to be useful. In almost every inquiry a scholar formulates several hypotheses and hopes that one will provide a satisfactory solution to the problem. By eliminating the false hypothesis one by one the investigator keeps narrowing the field in which the answer must lie.
The testing of a false hypothesis is also of value if it directs the attention of scientists to unsuspected facts or relations they eventually help in solving the problem.
5. Roots in Existing Theories
A useful educational hypothesis, therefore, adds something to previously established knowledge by supporting, qualifying, refuting or enlarging upon existing theories. A hypothesis that is compatible with well-attested theories is in a favorable position to advance knowledge.
If progress is to be made new hypothesis must fit into the framework of existing theories and transform them into more perfect explanatory schemes. Thus, even the more revolutionary theories are not completely different from the existing edifice of knowledge.
6. Suitability for Intended Purpose
Each hypothesis that offers a satisfactory explanation of what it intends to explain is useful for that purpose. Every hypothesis serves a specific purpose and must be adequate for the purpose it claims to serve. Thus, suitability is also the important criterion for an effective hypothesis.
7. Simplicity of Explanation
If two hypothesis is capable to explain the same facts, the simpler one is the better hypothesis. Simplicity means that the hypothesis explains the phenomena with the least complexes theoretical structure. The hypothesis that accounts for all facts with the fewest independent or special assumptions and complexities is always preferable.
8.Levels of Explanation
The value of a hypothesis can best be comprehended by tracing their relationship to facts theories and laws. The scientists build gradually a hierarchy of knowledge consisting of (1) hypothesis (2) theories and (3) laws. The following discussion will distinguish among these levels of knowledge.
(a) hypothesis and Facts
A hypothesis is the first step in the direction of scientific truth. In the hierarchy of scientific knowledge, it is the lowest on the scale. If empirical evidence can be found to verify the hypothesis, it gains the status of a fact. Thus, a fact is the verified hypothesis.
(b) hypothesis and Theories
A theory may contain several logically interrelated hypothesis and postulates may be used as a synonym for the hypothesis. hypothesis and theories are both conceptual in nature.
A theory usually provides a higher level explanation than a hypothesis. A theory presents a comprehensive conceptual scheme that may involve several related hypotheses and explain diverse phenomena; considerable empirical evidence is needed to support it.
(c) hypothesis and Laws
Some hypothesis receives sufficient confirmation to lead to the formulation of theories; some lead to the establishment of laws. Laws utilize highly abstract concepts, for they provide the most comprehensive type of explanations.
Laws may explain phenomena that have been explained previously by two or three theories. A law retains its lofty scientific status which it claims to explain.
THE ROLE OF hypothesis
The hypothesis plays a significant role in scientific studies. The following are some of the important roles of a hypothesis
The purpose of stating hypothesis, like the purpose of theories that may be involved, is to provide a framework for the research procedure and methodology. It directs the research activities.
A research project needs to proceed from a statement of the hypothesis. Such a hypothesis are not ended in themselves but rather aids to the research process.
A hypothesis takes on some characteristics of a theory which is usually considered as a larger set of generalization about a certain phenomenon.
The verification. of a hypothesis does not prove or disprove it; it merely sustains or refutes the hypothesis.
The hypothesis may imply research procedures to be used and the necessary data to be organized.
Such a hypothesis are not ended in themselves but rather aids to the research process.
The conclusions of the research problem may also be stated in the context of the initial hypothesis.
The stating hypothesis in experimental research provides the basis for designing the experiment and collecting evidence empirically for its verification so as to formulate a new theory in the field of education.
The hypothesis orients the research process for its verification rather than finding out the solution to the problem.
OBJECTIONS AGAINST STATING Hypothesis
The following objections are raised against stating hypothesis which is directional in nature
One is that hypothesis bias the researcher in favor of certain conclusions or retain the hypothesis.
Another is that in his pursuit of the stating hypothesis the researcher may overlook another possibly worthwhile hypothesis
The statement of hypothesis in some situations also may appear premature
A directional hypothesis needs some theoretical rationale but in some situations, there is very little background information about them
The researcher may decide to defer any hypothesis or theories until he has some empirical evidence upon which is to base them
The hypothesis is stated in a vacuum. These should be concerned with a situation in which it can be experienced.
The directional hypothesis should be so stated as to reveal the role of variables involved in the investigation.
The overall consensus is in favor of stating a hypothesis whenever they are feasible. In view of the above objections. Researchers prefer to formulate the non-directional hypothesis these days.
Hypothesis IN HISTORICAL RESEARCH
The historical researcher uses his information to describe and interpret conditions, events, and phenomena that existed during the period under study. Some of the scholars of research methodology are of this view that the historical researcher also can formulate a hypothesis to direct the research activities.
These hypotheses are attempts at explaining and interpreting the phenomena of the period under the study.
There is the difference between scientific hypothesis and the historical hypothesis. the hypothesis in historical research are not formulated in a statistical sense or null hypothesis. Historical hypothesis takes on a broader meaning as a conjecture of the situation.
An example, a researcher is pursuing historical research on the development of teacher-education of the secondary stage in India. There would be several hypotheses. One hypothesis may be- ‘The development of teacher-education as an outgrowth of secondary schools and inadequate supply of teachers produced by the colleges’.
This hypothesis is based on the assumption that there has been the development of teacher education. If this assumption was not correct, the hypothesis would have no basis.
The matter of basing a hypothesis on accurate assumptions may seem obvious, but failure to do so is not unknown. The position of the hypothesis is based on the assumption.
USES OF hypothesis IN EDUCATIONAL RESEARCHES
The educational researches may be classified into four types:
1. Experimental research
2. Normative survey research
3. Historical Research, and
4. Complex casual research
1. hypothesis is indispensable for experimental researches. The experiments are conducted to collect empirical data to verify the hypothesis. The experimental method or experimental designs are based on a hypothesis. hypothesis are the crucial aspects of such researches.
2. In normative survey research, the investigator may or may not employ hypothetical type thinking, depending upon the purpose of the research study. the hypothesis is essential for analytical studies and there is little scope in descriptive type studies.
3. In historical research, the purpose may be either to produce a faithful record of the past events irrespective of the present-day problem or to extend the experience with phenomena in the present to past in order to make the view of the phenomena.
There is a little scope of hypothesis in historical research because the hypothesis has the future reference and its verification on empirical data. Case study method has no scope for constructing a hypothesis because it is a developmental type study.
4. In complex casual research, the hypothesis has an important role in such investigations. These types of studies are conceptual in nature whereas historical are more factual in nature. Therefore formulation of a hypothesis is a crucial step in this type of studies.
One component of a strong paper is a precise, interesting hypothesis. Another component is the testing of the hypothesis and the presentation of the supporting evidence.
As part of the research process you need to test your hypothesis and if it is correct—or, at least, not falsified—assemble supporting evidence. In presenting the hypothesis, you need to construct an argument relating your hypothesis to the evidence.
This may or may not be good evidence, but it is not convincing because there is no argument connecting the evidence to the hypothesis. What is missing is information such as “results for previous methods indicated an asymptotic cost of 8(log n)”.
It is the role of the connecting argument to show that the evidence does indeed support the hypothesis and to show that conclusions have been drawn correctly.
In constructing an argument, it can be helpful to imagine yourself defending your hypothesis to a colleague, so that you play the role of inquisitor. That is, raising objections and defending yourself against them is a way of gathering the material that is needed to convince the reader that your argument is correct.
Starting from the hypothesis that “the new string hashing algorithm is fast because it doesn’t use multiplication or division” you might debate as follows:
I don’t see why multiplication and division are a problem.
On most machines, they use several cycles, or may not be implemented in hardware at all. The new algorithm instead uses two exclusive-or operations per character and a module in the final step. I agree that for pipelined machines with floating-point accelerators the difference may not be great.
Modulo isn’t always in hardware either. True, but it is only required once.
So there is also an array lookup? That can be slow. Not if the array is cache-resident.
In an argument, you need to rebut likely objections while conceding points that can’t be rebutted, while also admitting when you are uncertain.
If in the process of developing your hypothesis, you raised an objection but reasoned it away, it can be valuable to include the reasoning in the paper. Doing so allows the reader to follow your train of thought, and greatly helps the reader who independently raises the same objection.
A hypothesis can be tested in a preliminary way by considering its effect, that is, by examining whether there is a simple argument for keeping or discarding it. For example, are there any improbable consequences if the hypothesis is true? If so, there is a good chance that the hypothesis is wrong.
For a hypothesis that displaces or contradicts some currently held belief, is the contradiction such that the belief can only have been held out of stupidity? Again, the hypothesis is probably wrong. Does the hypothesis cover all of the observations explained by the current belief? If not, the hypothesis is probably uninteresting.
Always consider the possibility that your hypothesis is wrong. It is often the case that a correct hypothesis at times seems dubious—perhaps in the early stages, before it is fully developed, or when it appears to be contradicted by initial experimental Evidence.
But the hypothesis survives and may even be strengthened by test and refinement in the face of doubt.
But equally often a hypothesis is false; in which case clinging to it is a waste of time. Persist for long enough to establish whether or not it is likely to be true, but to persist longer is foolish.
A corollary is that the stronger your intuitive liking for a hypothesis, the more rigorously you should test it—that is, attempt to confirm it or disprove it—rather than twist results, and yourself, defending it.
Be persuasive. Using research into the properties of an algorithm as an example, issues such as the following need to be addressed.
Will the reader believe that the algorithm is new?
Only if the researcher does a careful literature review, and fully explores and explains previous relevant work. Doing so includes giving credit to significant advances, and not overrating work where the contribution is small.
Will the reader believe that the algorithm is sensible?
It had better be explained carefully. Potential problems should be identified, and either conceded—with an explanation, for example, of why the algorithm is not universally applicable—or dismissed through some cogent argument.
Are the experiments convincing?
If the code isn’t good enough to be made publicly available, is it because there is something wrong with it? Has the right data been used? Has enough data been used?
Every research program suggests its own skeptical questions. Such questioning is also appropriate later in a research program, where it gives the author an opportunity to make a critical assessment of the work.
Forms of Evidence
A paper can be viewed as an assembly of evidence and support explanations; that is, as an attempt to persuade others to share your conclusions. Good science uses objective evidence to achieve aims such as to persuade readers to make more informed decisions and to deepen their understanding of problems and solutions.
In a write-up, you pose a question or hypothesis, then present evidence to support your case. The evidence needs to be convincing because the processes of science rely on readers being critical and skeptical; there is no reason for a reader to be interested in work that is inconclusive.
There are, broadly speaking, four kinds of evidence that can be used to support a hypothesis: proof, modeling, simulation, and experiment.
A proof is a formal argument that a hypothesis is correct (or wrong). It is a mistake to suppose that the correctness of a proof is absolute—confidence in a proof may be high, but that does not guarantee that it is free from error;
It is common for a researcher to feel certain that a theorem is correct but have doubts about the mechanics of the proof.
Some hypothesis is not amenable to formal analysis, particularly the hypothesis that involves the real world in some way. For example, human behavior is intrinsic to questions about interface design, and system properties can be intractably complex.
Consider an exploration to determine whether a new method is better than a previous one at lossless compression of images—is it likely that material that is as diverse as images can be modeled well enough to predict the performance of a compression algorithm?
It is also a mistake to suppose that an asymptotic analysis is always sufficient. Nonetheless, the possibility of formal proof should never be overlooked.
A model is a mathematical description of the hypothesis (or some component of the hypothesis, such as an algorithm whose properties are being considered) and there will usually be a demonstration that the hypothesis and model do indeed correspond.
In choosing to use a model, consider how realistic it will be, or conversely how many simplifying assumptions need to be made for analysis to be feasible. Take the example of modeling the cost of a Boolean query on a text collection, in which the task is to find the documents that contain each of a set of words.
We need to estimate the frequency of each word (because words that are frequent in queries may be rare in documents); the likelihood of query terms occurring in the same document (in practice, query terms are thematically related, and do not model well as random co-occurrences);
The fact that longer documents contain more words, but are more expensive to fetch; and, in a practical system, the probability that the same query had been issued recently and the answers are cached in memory.
It is possible to define a model based on these factors, but, with so many estimates to make and parameters to tune, it is unlikely that the model would be realistic.
A simulation is usually an implementation or partial implementation of a simplified form of the hypothesis, in which the difficulties of a full implementation are sidestepped by omission or approximation.
At one extreme a simulation might be little more than an outline; for example, a parallel algorithm could be tested on a sequential machine by use of an interpreter that counts machine cycles and communication costs between simulated processors;
At the other extreme, a simulation could be an implementation of the hypothesis but tested on artificial data. A simulation is a “white coats” test: artificial, isolated, and conducted in a tightly controlled environment.
A great advantage of a simulation is that it provides parameters that can be smoothly adjusted, allowing the researcher to observe behavior across a wide spectrum of inputs or characteristics.
For example, if you are comparing algorithms for removal of errors in genetic data, use of simulated data might allow you to control the error rate, and observe when the different algorithms begin to fail.
Real data may have unknown numbers of errors or only a couple of different error rates, so in some sense can be less informative. However, with a simulation, there is always the risk that it is unrealistic or simplistic, with properties that mean that the observed results would not occur in practice.
Thus simulations are powerful tools, but, ultimately, need to be verified against reality.
An experiment is a full test of the hypothesis, based on an implementation of the proposal and on real—or highly realistic—data. In an experiment, there is a sense of really doing it, while in a simulation there is a sense of only pretending.
For example, artificial data provides a mechanism for exploring behavior, but corresponding behavior needs to be observed on real data if the outcomes are to be persuasive.
In some cases, though, the distinction between simulation and experiment can be blurry, and, in principle, an experiment only demonstrates that the hypothesis holds for the particular data that was used; modeling and simulation can generalize the conclusion (however imperfectly) to other contexts.
Ideally, an experiment should be conducted in the light of predictions made by a model so that it confirms some expected behavior.
An experiment should be severe; seek out tests that seem likely to fail if the hypothesis is false, and explore extremes. The traditional sciences and physics, in particular, proceed in this way. Theoreticians develop models of phenomena that fit known observations; experimentalists seek confirmation through fresh experiments.
Use of Evidence
Different forms of evidence can be used to confirm one another, with say a simulation used to provide further evidence that a proof is correct. But the different forms should not be confused with one another. For example, suppose that for some algorithm there is a mathematical model of expected performance.
Encoding this model in a program and computing predicted performance for certain values of the model parameters is not an experimental test of the algorithm and should never be called an experiment; it does not even confirm that the model is a description of the algorithm. At best it confirms claimed properties of the model.
When choosing whether to use a proof, model, simulation, or experiment as evidence, consider how convincing each is likely to be to the reader. If your evidence is questionable—say a simplistic and assumption-laden model.
An involved algebraic analysis and application of advanced statistics, or an experiment on limited data the reader may well be skeptical of the result. Select a form of evidence, not so as to keep your own effort to a minimum, but to be as persuasive as possible.
Having identified the elements a research plan should cover, end-to-start reasoning suggests how these elements should be prioritized. The write-up is the most important thing, so perhaps it should be started first.
Completing the report is certainly more important than hastily running some last-minute experiments, or quickly browsing the literature to make it appear as if past work has been fully evaluated.
Some novice researchers feel that the standards expected of evidence are too high, but readers—including referees and examiners—tend to trust work that is already published in preference to a new, unrefereed paper, and have no reason to trust work where the evidence is thin.
Moreover, experienced researchers are well aware that skepticism is justified. It has been said, with considerable truth, that most published research findings are false; and unpublished findings are worse.
This means that a paper must be persuasive. Your written work is the one chance to persuade readers to accept the ideas, and they will only do so if the evidence and arguments are complete and convincing.
Approaches to Measurement
A perspective on the history of science is that it is also a history of the development of tools of measurement. Our understanding of the laws of physics followed by the development of telescopes, voltmeters, thermometers, and so on. Each improvement in the measurement technology has refined our understanding of the underlying properties of the universe.
From this perspective, the purpose of experimentation is to take measurements that can be used as evidence. A good choice of measure is essential to practical system improvement and too persuasive and insightful writing. The measurements are intended to be a consequence of some underlying phenomenon that is described by a theory or hypothesis.
In this approach to research, phenomena—the eternal truths studied by science—cannot change, but the measurements can because they depend on the context of the specific experiment.
Measurements can be quantitative, such as number or duration or volume—the speed of a system, say, or an algorithm’s efficiency relative to a baseline. They can also be qualitative, such as an occurrence or difference—whether an outcome was achieved, or whether particular features were observed.
As you develop your research questions, then, you should ask what is to be measured? And what measures will be used? For example, when examining an algorithm, will it be measured by execution time? And if so, what mechanism will be used to measure it? This question can be tricky to answer for a single-threaded process running on a single machine.
For a distributed process using diverse resources across a network, there probably is no perfect answer, only a range of choices with a variety of flaws and shortcomings, each of which needs to be understood by you and by your readers.
There is then a critical, but more subtle, question: you need to be satisfied that the properties being measured are logically connected to the aims of the research. Typically, research aims are qualitative. We seek to improve an interface, accelerate an algorithm, extract information from an image, and generate better timetables for lectures, and so on.
Measurement is quantitative; we find a property that can be represented as a quantity or value. For example, the effectiveness of machine translation systems is sometimes assessed by counting the textual overlap (words or substrings) of a computer translation with that made by a human.
However, such a measure is obviously imperfect: not only are there many possible human translations, but a highly overlapping text can still be incoherent, that is, not a good translation.
As another example, we might say that the evidence for the claim that a network is qualitatively improved is that average times to transmit a packet are reduced—a quantity that can be measured. But if the aim of network improvement is simplified to the goal of reducing wait times, then other aspects of the qualitative aim (smoothness of transmission of video, say, or effectiveness of service for remote locations) may be neglected.
In other words, once a qualitative aim is replaced by a single quantitative measure, the goal of research in the field can shift away from the achievement of a practical outcome, and instead consist entirely of optimization to the measure, regardless of how representative the measure is of the broader problem.
A strong research program will rest, in part, on recognition of the distinction between qualitative goals and different quantitative approximations to that goal.
The problem of optimization-to-a-measure is particularly acute for fields that make use of shared reference data sets, where this data is used for the evaluation of new methods.
It is all too easy for researchers to begin to regard the standard data as being representative of the problem as a whole and to tune their methods to perform well on just these data sets. Any field in which the measures and the data are static is at risk of becoming stagnant.
Good and Bad Science
Questions about the quality of evidence can be used to evaluate other people’s research, and provide an opportunity to reflect on whether the outcomes of your work are worthwhile. There isn’t a simple division of research into “good” and “bad”, but it is not difficult to distinguish valuable research from work that is weak or pointless.
The merits of formal studies are easy to appreciate. They provide the kind of mathematical link between the possible and the practical that physics provides between the universe and engineering.
The merits of well-designed experimental work are also clear. Work that experimentally confirms or contradicts the correctness of formal studies has historically been undervalued in computer science: perhaps because standards for experimentation have not been high;
Perhaps because the great diversity of computer systems, languages, and data has made truly general experiments difficult to devise; or perhaps because theoretical work with advanced mathematics is more intellectually imposing than work that some people regard as mere code-cutting.
However, many questions cannot be readily answered through analysis, and a theory without practical confirmation is of no more interest in computing than in the rest of science.
Research that consists of proposals and speculation, entirely without a serious attempt at evaluation, can be more difficult to respect. Why should a reader regard such work as valid?
If the author cannot offer anything to measure, arguably it isn’t science. And research isn’t “theoretical” just because it isn’t experimental. Theoretical work describes testable theories.
Reflections on Research
Philosophers and historians of science have reflected at length on the meaning, elements, and methods of research, from both practical and abstract points of view.
While philosophy can seem remote from the practical challenges of research, these reflections can be of great benefit to working scientists, who can learn from an overall perspective on their work. Being able to describe what we do helps us to understand whether we are doing it well.
Such philosophies and definitions of science help to establish guidelines for the practical work that scientists do, and set boundaries on what we can know. However, there are limits to how precise (or interesting) such definitions can be. For example, the question “is computer science a science?” has a low information content.
It is true that considered as a science, computing is difficult to categorize. The underlying theories—in particular, information theory and computability—appear to describe properties as eternal as those of physics. Yet much research in computer science is many steps removed from foundational theory and more closely resembles engineering or psychology.
A widely agreed description of science is that it is a method for accumulating reliable knowledge. In this viewpoint, scientists adopt the belief that rationality and skepticism are how we learn about the universe and shape new principles while recognizing that this belief limits the application of science to those ideas that can be examined in a logical way.
If the arguments and experiments are sound, if the theory can withstand skeptical scrutiny, if the work was undertaken within a framework of past research and provides a basis for further discovery, then it is science. Much computer science has this form.
Many writers and philosophers have debated the nature of science, and aspects of science such as the validity of different approaches to reasoning. The direct impact of this debate on the day-to-day activity of scientists is small, but it has helped to shape how scientists approach their work. It also provides elements of the ethical framework within which scientists work.
One of the core concepts is falsification: experimental evidence, no matter how substantial or voluminous, cannot prove a theory true, while a single counter-example can prove a theory false.
A practical consequence of the principle of falsification is that a reasonable scientific method is to search for counter-examples to the hypothesis. In this line of reasoning, to search for supporting evidence is pointless, as such evidence cannot tell us that the theory is true.
A drawback of this line of reasoning is that using falsification alone, we cannot learn any new theories; we can only learn that some theories are wrong. Another issue is that, in practice, experiments are often unsuccessful.
But the explanation is not that the hypothesis is wrong, but rather that some other assumption was wrong—the response of a scientist to a failed experiment may well be to redesign it.
For example, in the decades-long search for gravity waves, there have been many unsuccessful experiments, but a general interpretation of these experiments has been that they show that the equipment is insufficiently sensitive.
Thus falsification can be a valuable guide to the conduct of research, but other guides are also required if the research is to be productive. One such guide is the concept of confirmation.
In science, confirmation has a weaker meaning than in general usage; when a theory is confirmed, the intended meaning is not that the theory is proved, but that the weight of belief in the theory has been strengthened. Seeking experiments that confirm theories is an alternative reasonable view of scientific method.
A consequence is that a hypothesis should allow some possibility of being disproved—there should be some experiment whose outcomes could show that the hypothesis is wrong. If not, the hypothesis is simply uninteresting.
Consider, for example, the hypothesis “a search engine can find interesting Web pages in response to queries”. It is difficult to see how this supposition might be contradicted.
In the light of these descriptions, science can be characterized as an iterative process in which theory and hypothesis dictate a search for evidence—or “facts”—while we learn from facts and use them to develop theories. But we need initial theories to help us search for facts.
Thus confirmation, falsification, and other descriptions of method help to shape research questions as well as research processes, and contribute to the practice of science.
We need to be willing to abandon theories in the face of contradictions, but flexible in response to the failure; contradictions may be due to an incorrect hypothesis, faulty experimental apparatus, or poor measurement of the experimental outcomes.
We need to be ready to seek plausible alternative explanations of facts or observations and to find experiments that yield observations that provide insight into theories. That is, theories and evidence are deeply intertwined.
A scientific method that gives one primacy over the other is unlikely to be productive, and, to have a high impact, our research programs should be designed so that theory and evidence reinforce each other.
A “hypothesis, Questions, and Evidence” Checklist
What phenomena or properties are being investigated? Why are they of interest?
Has the aim of the research been articulated? What are the specific hypothesis and research questions? Are these elements convincingly connected to each other?
To what extent is the work innovative? Is this reflected in the claims?
What would disprove the hypothesis?
Does it have any improbable consequences?
What are the underlying assumptions? Are they sensible?
Has the work been critically questioned? Have you satisfied yourself that it is sound science?
Regarding evidence and measurement, What forms of evidence are to be used? If it is a model or a simulation, what demonstrates that the results have practical validity?
How is the evidence to be measured? Are the chosen methods of measurement objective, appropriate, and reasonable?
What are the qualitative aims, and what makes the quantitative measures you have chosen appropriately to those aims?
What compromises or simplifications are inherent in your choice of measure? Will the outcomes be predictive?
What is the argument that will link the evidence to the hypothesis?
To what extent will positive results persuasively confirm the hypothesis? Will negative results disprove it?
What are the likely weaknesses of or limitations to your approach?