In this article we will discuss about:- 1. Types of Hypotheses 2. Levels of Hypothesis 3. Functions 4. Testing.
Types of Hypotheses:
There are several different kinds of hypotheses used in social and/or geographical analysis, studies and research.
However, the primary types of hypotheses are:
(1) Research Hypotheses,
(2) Null Hypotheses,
(3) Scientific Hypotheses, and
(4) Statistical Hypotheses.
1. Research Hypotheses:
Hypotheses derived from the researcher’s theory about some social and/or geographical phenomena are called research hypotheses or ‘working’ hypotheses.
The social investigator usually believes that his/her research hypotheses are true or they are accurate statements about the condition of things he/she is investigating. The investigator believes that these hypotheses are true to the extent that the theory from which they were derived is adequate.
Theories are, in one sense, suppositions about the true nature of things, and thus regarded as tentative statements about reality. Until they have been verified to the scientist’s satisfaction, the hypotheses derived from theories must also be regarded as tentative suppositions about things until they have been tested. Testing hypothesis means to subject it to confirmation or disconfirmation.
2. Null Hypotheses:
Null hypotheses are, in a sense, the reverse of research hypotheses. They are also statements about the reality of things, except that they serve to refute or deny what is explicitly indicated in a given research hypothesis.
Null hypotheses are hypothetical models used to test research hypotheses. The question that arises as why does the social investigator want to bother with so-called null hypotheses? Why doesn’t the investigator test the hypothesis directly and let it go at that?
These questions have been asked time and again by every researcher confronting null hypotheses for the first time. There are at least four explanations why null hypothesis models are used, none of which, however, may answer this question satisfactorily.
i. Trying to show the truthfulness of research hypotheses would imply to some, at least, a definite bias towards trying to confirm one’s suppositions and possibly ignoring those things that would tend to refute our belief.
ii. There are those who would argue that it is easier to find fault with something, i.e. an idea, belief, or hypothesis than to look for those things that would support it.
iii. It may be summed up in one word convention. It is conventional in social research to use null hypotheses. Null hypotheses, however, also perform specific functions in relation to probability theory and tests of research hypotheses.
iv. Under a probability theoretical model, hypotheses have a likelihood of being either true or false. Null hypotheses are particularly useful in such theoretical models. The null hypothesis is an expression of one alternative outcome of a social/physical observation.
The probability model specifies that the null hypotheses may be either true or false but not both simultaneously. Neither the research hypotheses nor the null hypothesis is absolutely true or absolutely false under any given test of it. Both probabilities (being either true or false) co-exist for each type of hypothesis always.
3. Scientific Hypotheses:
In scientific investigation, however, the term hypothesis is often given a somewhat more restricted meaning. To Braithwaite (1960) – A scientific hypothesis is a general proposition about all the things of a certain sort. It is an empirical proposition in the sense that it is testable by experience; experience is relevant to the question as to whether or not the hypothesis is true, i.e. as to whether or not it is a scientific law.’
A scientific hypothesis, in Braithwaite’s tradition, is a particular kind of proposition which, if true, will be accorded the status of a scientific law. The testability of a hypothesis is crucial, but there are many hypotheses within a theoretical system which cannot be directly tested against sense perception data.
Thus, ‘The empirical testing of the deductive system is effected by testing the lowest level hypotheses in the system. The confirmation or refutation of these is the criterion by which the truth of all the hypotheses in the system is tested’.
Since scientific hypothesis is often regarded as being a proposition where truth or falsity is capable of being asserted, the truth and falsity of it (hypothesis) can be determined only with respect to the domain of some theory.
4. Statistical Hypotheses:
These are statements about statistical population that, on the basis of information obtained from observed data, one seeks to support or refute. The statistical population may refer to either people or things. It is generally the case in the test of statistical hypotheses that observations about people or things are reduced in some way to numerical quantities, and decisions are made about these quantities.
To subject these hypotheses to empirical test, what is required is to reduce the variables used in them to measurable quantities. Research hypothesis and corresponding null hypotheses can be transferred into a statistical hypotheses that may be evaluated by numerical means.
Statistical hypotheses are usually established to delineate:
i. Differences between two or more groups regarding some trait or collection of characteristics that they possess,” association between two or more variables within one group or between several groups, and
ii. Point estimates of sample or population characteristics.
Levels of Hypothesis:
Apart from the aforesaid four types of hypotheses, three broad levels of hypotheses may be distinguished on the basis of the level of abstraction, which are as follows:
1. Some hypotheses state the existence of empirical uniformities. These hypotheses frequently, though not always, represent the scientific examination of common-sense propositions. They usually represent, also, a problem about which some ‘common-sense’ observation already exists. There are many types of such empirical uniformities which are common in social science and/or geographical research.
However, these investigations do not involve the testing of hypothesis at all, but are merely adding up the facts. These are not useful hypotheses for they merely represent what everyone already knows.
2. Some hypotheses are concerned with complex ideal types. These hypotheses aim at testing the existence of logically devised relationships between empirical uniformities. One such hypothesis was Ernest W. Burgess’s statement on the concentric growth circles that characterise the city.
This hypothesis was then tested against a variety of variables in a number of cities. That this ideal type does represent the actual patterns of city growth is not accepted by all ecologists, however, and so this formulation remains a hypothesis until a more crucial test of it is made.
Another hypothesis, concerning an ideal type also, results from these same ecological empirical uniformities. This was the notion that areas tend to represent certain characteristics in a series of predictable patterns. This was called the hypothesis of the ‘natural area’.
Much research has been done on this hypothesis, and the results, although they have modified the original statement somewhat, have generally supported it. With the growth of supporting evidence, notions about natural area have become a part of geographical theory rather than remaining hypotheses.
It is important to see that this level of hypothesising moves beyond the expectation of simple empirical uniformity, by creating a complex referent in society. The function of such hypothesis is to create tools and problems for further research in otherwise very complex area of investigation.
3. Some hypotheses are concerned with the relation of analytic variable. These hypotheses occur at a level of abstraction beyond that of ideal types. The hypotheses of empirical uniformities lead to the observation of simple differences, and those dealing with ideal types lead to specific coincidences of observations. The study of analytical variables requires the formulation of a relationship between changes in one property and changes in another.
On the basis of the above discussion, three major points can be identified:
(1) That a hypothesis is a necessary condition for successful research;
(2) That formulation of the hypothesis must be given considerable attention, to clarify its relation to theory, remove vague or value judgemental terms, and specify the test to be applied, and
(3) That hypotheses may be formulated on different levels of abstraction.
Functions of Hypotheses:
Theories are relatively elaborate tools used to explain and predict events. The social scientist develops a theory to account for some social phenomena, and then he devises a means whereby the theory can be tested or subjected to verification or refutation. Seldom does the researcher test theory directly. Most of the time he/she conducts tests of hypotheses that been generated and derived from that theory.
If the hypotheses ‘test out’ as the researcher has specified, or if his empirical observations are in accordance with what has been stated in the hypotheses, we say that his/her theory is supported in part. It usually takes many tests of different hypotheses from the same theory to demonstrate its predictive value and its adequacy as a tool of explanation for some event or sequence of events.
A major function of hypotheses is to make it possible to test theories. In this regard, an alternative definition of a hypothesis is that it is a statement of theory in testable forms. All statements of theory in testable form are called hypotheses.
Some hypotheses are not associated with any particular theory. It could be that as a result of some hypothesis, a theory will be eventually constructed. Consequently, another function of hypotheses is to suggest theories that may account far some event.
Although it is more often the case that research proceeds from theories to hypotheses, occasionally the reverse is true. The social investigator may have some idea about why a given phenomenon occurs and he/she hypothesises a number of things that relate to it.
He/she judges that some hypotheses have greater potential than others for explaining the event, and as a result, he/she may construct a logical system of propositions, assumptions and definitions linking his explanation to the events. In other words, the researcher devises a theory.
Working from the hypothesis back to the theory is not necessarily poor methodology. Eventually, the investigator is going to have to subject the resulting theory to empirical test to determine its adequacy. The predictive value of the theory can be assessed at that time.
Hypotheses also perform a descriptive function. Each time a hypothesis is tested empirically, that tells something about the phenomenon it is associated with. If the hypothesis is supported, then the information about the phenomenon increases.
Even if the hypothesis is refuted, the test tells something about the phenomenon that is not known before. The accumulation of information as a result of hypothesis testing reduces the amount of ignorance we may have about why a social event occurs in a given way.
Hypotheses also have some important secondary functions. As a result of testing certain hypotheses, social policy may be formulated in communities, penal institutions may be redesigned and revamped, teaching methods may be altered or improved solutions to various kinds of social problems may be suggested and implemented, and supervisory practices may be changed in factories and business.
Testing hypotheses refute certain ‘common sense’ notions about human behaviour, raises questions about explanations we presently use to account for things, and most generally alters our orientation towards our environment to one degree or another. All hypotheses have to do with our knowledge of things, and as this knowledge changes, we change also.
Testing hypotheses means ‘subjecting them to some sort of empirical scrutiny to determine if they are supported or refuted by what the researcher observes’. Testing hypotheses means that the researcher will need to do a number of things.
Following are the two prerequisites to hypotheses testing:
1. A real social situation is needed that will suffice as a reasonable testing ground for the hypothesis. If the hypothesis concerns managerial behaviour, it will be necessary for the investigator to study some real organisation or organisations where managerial behaviour can be taken into empirically.
This particular prerequisite is frequently spoken of as ‘getting access to data that will enable the investigator to verify or refute his/her hypotheses’. Once a given social setting is selected, the relevant data in that situation must be obtained to make the hypothesis test a valid one.
2. The investigator should make sure that his hypotheses are testable. This means that he/she should limit his/her investigations to empirical phenomena or events that can be taken into through the senses. The variables used in the hypotheses tested should be amenable to measurement of some kind.
If they are not subject to measurement, the resulting test of the hypothesis will be relatively meaningless. Testing hypotheses must be a part of the empirical world. This is a fundamental requirement wherever the scientific method is employed in studying what is and why.
Terms that cannot be taken into empirically, render the hypothesis irrefutable and untestable. How can a scientist reject a hypothesis containing variables that he cannot experience in some empirical form? For example, if a researcher were to hypothesise that ‘evil spirit causes delinquency’, he/she can neither support nor refute this statement by using conventional scientific methods.
He/ she obviously has empirical tools to determine the incidence of delinquent or non-delinquent behaviour, but by what empirical means is he/she able to assess meaningfully the influence or impact of ‘evil spirits’ on delinquent behaviour?
Unless there are empirical means of evaluating the impact of non- empirical phenomena on particular variables, the researcher cannot validly subject the hypothesis to true scientific test. However, it is possible that terms that are presently indefinable empirically, might at some later date become amenable to the senses through the discovery of new means of measuring such phenomena. This always exists as a possibility.