Hypothesis Development in Business Research
From previous: The Research Process
Once it has been decided that a research be conducted to solve the problem, the researcher would develop a plan or a framework. A conceptual or theoretical model would be drawn taking into account various factors which have identified links with the problem. Suppose profitability has gone down for no apparent reasons. There are various reasons for fall in profit such as lower prices, higher costs or lower sales in term of quantity. These factors have a direct bearing on profitability and would be good point to start with.
The theoretical framework helps develop testable hypotheses which in simple words mean probable causes. Afterwards, some hypotheses can be rejected and the research can be narrowed down to a most plausible causes. For example, we may find from actual records that sales, both in quantity and value, have improved over the previous year. If so, we can safely say that price and quantity is not reason for fall in the profit. This leaves cost which we can analyse further in term of raw materials consumed, labor charges and overheads. Our focus would be that under which area the cost increase was significant. The research process would continue till a valid reason for fall in profit is found out.
A framework is necessary to keep the research within necessary bounds else the researcher may wander to irrelevant areas wasting time and money.
In fact, a theoretical framework established boundaries of research: what to look for and what not.
Dictionary meaning of variable is "changing" or "changeable". A variable is anything that can change in its quantum or value. It is opposite is invariable, consistent, constant or fixed like the Rock of Gibraltar. In research, it is an item like sale and profit. Normally, the profit increases with the increase in sales. So both sales and profit could be variables in any study. On the other hand, names, identity numbers, telephone numbers and codes are not variables but non-variable or fixed.
Variables used in business research are classified into four categories:
- The dependant variables
- The independent variables
- The moderating variables
- The intervening variables
The first two represent cause and effect relationship. The last two are equally important as they can exert their influences on the effect or results or outcome. Of this, moderating variables can be identified and controlled. Sometimes, these are called secondary independent variables. The last one, intervening variable, is a sort of confounding variable and arises during the course of the research. The examples are motivation, boredom and tiredness. The researchers feel their influence after the experiment. It is like a ‘black box’ which conveys results after the event.
Deductive Method - Sherlock Holmes way
Dependent variable and Independent Variable
As the names imply, a dependent variable is a one which is dependent on another variable. With the increase in temperature, the sales of ice cream may go up. If so, one can say that the sale of ice cream depends upon the temperature. In such case, the sale of ice cream is a dependent variable whereas the temperature is an independent variable.
The dependent variable is of primary interest to the researcher. It is observed as to how it responds to the changes made to the independent variable. If a researcher wants to prove that increase in overhead lights would increase productivity, the researcher would note down changes in the productivity (dependent variable) with every increase of overhead lights by 5% (Independent variable). If per chance, the productivity does not increase with increase in wattage, the productivity would not be dependent upon lights or would not be considered as dependent variable.
In normal circumstances, increase in returns (independent variable) would attract more bank deposits (dependent variable). If an area is known for bank failures or bank frauds, no one would divert his or her savings to a bank simply on the basis of higher returns. In such a case, the depositor would look for a solid bank even if the returns are low.
Take another example that a medical student is interested to find out that how stress affects heart rate in humans. If so, the heart rate would be dependent variable and stress would be independent variable. By increase stress levels in our human subjects by giving false alarms, we can measure how much the heart rate has increased.
An independent variable influences the dependent variable in either a positive or negative way. The independent variable is changed deliberately to establish a correlation or linkage. In any experiment, one independent variable is changed at one time. If there are two are more independent variables, a way is found to neutralize their effect by some technique.
Suppose a manager believes that (i) good supervision and (ii) good training would increase the production level of workers. For finding out effect these two factors, we would have to carry out two experiments. In one, same workforce would be supervised by different supervisors known to be good or bad. In second experiment, the supervisor would remain the same but the workers would be changed from untrained to trained. If there is a difference in productivity, it would be attributable to good supervision or good training as the case maybe.
Moderating Variables modify or affect the cause and effect relationship. It could be qualitative (gender, race, class) or quantitative (level of rewards or comfort). It changes the direction or strength of correlation between Independent and Dependent Variables.
If we start with a normal cause and effect relationship, it would be linear or perfectly correlative. Let us say that a 10% increase in advertisement would increase sale by 10%. There may be another variable which may alter the strength of causal relationship say incentives or rewards for achieving certain sale target. This may boost the sales to 25%. If incentives are withdrawn, the growth in sales may come down to original level of 10%.
Psychotherapy may work better for reducing depression experienced by men than by women. So we can conclude that gender moderates the causal effect of psychotherapy (X) on depression (Y).
Intervening Variables (IVV)Click thumbnail to view full-size
A factor which theoretically affects the observation or results but cannot be seen, measured, or manipulated. It is inferred after watching the effects of the independent and moderating variables on dependent variables. It surfaces between start and stop of the experiment. It helps to conceptualize and explain the result of the experiment.
A good speaker can increase understanding of students on a particular topic. Here good speaker is independent variable while understanding is dependent variable. There is a cause and effect relation, the better the speaker the better the understanding. Good environments like air-conditioned room, sound-proof lecture hall and good lights can act as moderating variables. If the speaker was good, environments were excellent yet the students could not recall basics of the lecture when asked the very next day. There must be some intervening variable which blocked minds of the students. Maybe students were worried about their examination and did not think the aforesaid lecture is anything to do with the syllabus. So they paid no attention. This could be inferred or can be stated as possible reason for the poor show.
Theoretical Framework & hypothesis development
After ascertaining the relevant variables and their inter-relationships, the researcher can draw a theoretical framework or conceptual model to start with. To be short, a theoretical framework would require (i) collection of inter-related factors, (ii) picking up the right factors and (iii) determining suitable tests.
Suppose, we want to khow what determines job satisfaction in a certain organization. Is it salary or job freedom? Once we start studying influence of these factors, so many other relevant factors would exert their influences such as physical environments, feedback system, job enrichment and enlargement. In order to measures the contribution of each variable, we would have to go through a process like (i) labeling the variables, (ii) survey of job-satisfaction through in-depth interviews of some random selected employees and (iii) a general survey. Finally we can calculate correlation between various factors and percentages etc. and draw our conclusion.
In this hub, more emphasis has been given on identifying the relevant factors concerning a research problem keeping in view the old wisdom - well began is half done. Once we have drawn a framework, we can develop hypotheses based on the factors included in our conceptual model.
Developing hypotheses requires a technique as discussed in the previous chapter i.e. Null and Alternate. Subsequently, each and every hypothesis would tested in isolation and its behavior recorded. Later, through use of statistical analysis, it would ascertained whether the said factor is of any significance or not. This would be taken up later while discussing research report.