Correlation

Correlation is used to measure the degree of association that is there between two variables. Whenever the researcher is dealing with two variables, the talk is about simple correlation and when the involvement is of more than two variables. On the other hand, regression is used to explain the variations that happen in one variable. This is usually referred to as the changes made in the dependant variable by the independent variable. It helps to identify the nature of the relationship. It is not necessary that the independent variable in regression would be only one. They can be more than one also. When there is only one independent variable, we call it simple regression and when   there is more than one variable, we call it multiple regression analysis.

Correlation helps in measuring the degree of association that is there between two or more variables. When the research is dealing with two variables, then the correlation applicable is simple correlation. When the involvement is of more than two variables then the subject matter moves towards multiple correlation. Correlation is largely of three types:

• Positive Correlation:  When the two variables X and Y move in the same direction, it is said that the correlation between the two is positive. If one variable increases the other variable also increases and likewise in the situation of one variable decreasing, the other one also decreases.
• Negative Correlation:  In the situation when the two variables X and Y move in the opposite direction, it is called as negative correlation. In the case when one variable increases and the other variable decreases or vice versa.
• Zero Correlation:  When the correlation between two variables X and Y is completely null or zero. The increase or decrease in Y is not dependent upon an increase or decrease in X.

If the correlation coefficient is equal to 1, the two variables are said to be positively correlated. If the coefficient of correlation is -1 the variables are said to be lying on a  negatively sloped straight line

Factor Analysis has proven to be a very useful technique in the field of market research and analysis. Its various uses are:

• It helps to make sense of a big data which has interlinked relationships
• It helps to decipher relationships that have been hidden
• It helps to surface up the underlying relationship that is there between tastes preferences of consumers where factor analysis is largely used.
• It helps in the condensing of the data
• It helps in correlating the data and draw conclusions from the gathered data.
• Helps in formation of the empirical clusters.

Types of Factor Analysis:
The larger use of factor analysis is for understanding the interpretation of data and analysing the underlying relationship that exists between variables and the other underlying factors. Factor analysis works beyond grouping responses and their types; on the other hand it segregates the variables and then groups them according to their co relevance.

Factor Analysis can largely be segregated into three categories, depending upon its varied use in the market.

• Exploratory Factor Analysis
• Confirmatory Factor Analysis
• Structural Equation Modelling

The exploratory factor analysis is used for the measurement of the underlying factors that have an effect on the variables in the data structure. This is done without any biased perspective and setting a pre-defined structure to the outcome. The second kind of factor analysis, which is the confirmatory factor analysis is used to confirm the correlation in the existing set of the factor that have been predefined and the different variables that affect these factors. The third type of factor analysis which is called the structural modelling hypothesises the relationship between a set of variables. It can be used for both exploratory as well as confirmatory modelling.

Factor analysis would yield accurate and beneficial results only when the expertise of the researcher is there in selecting the data and assigning it the attributes. Choosing of the correct factors so as to avoid a lot of overlapping in characteristics is also very important. If done in the right manner, factor analysis would assist in very critical decision making. It is particularly useful in consumer behaviour studies and it woyld help in product development, pricing segmentation, penetration, distribution, pricing and other important decisions.

The Big Data Bandwagon has picked up momentum and all the consultants, professors, organisers, writers pundits, crooks, cheats, equity firms are queuing up to get aboard. A bandwagon has rarely before called for so much attention and passengers.

The basic premises for big data are:
There is common perception that more data is always better than less data.
Greater volume, variety and the velocity of the data creates further avenues of knowledge that can be called potential.

It is possible to answer “ALL” the question through big data and it is all the more easy to predict the future.
The questions that still create ripples are that, “Can we create an accurate picture of the future through Big Data or is it just a glittery mirage that shimmers far away in the distance in the heat of a desert? Is it the final truth or a bandwagon of overstated commitments and mirage dreams?

The truth to all this is that the solution to the business problems and the determining of strategic opportunities often rests in the boundary of little data and not Big Data. It is not required to boil the ocean to find out the salt content in it and nor is it required to eat the full steer to understand it is tough.
Corporate decision makers would be served better if they could trust on tools from the world of little data that were tried and tested and not like the illusionary Big Data. Sampling theory does state that in the case of a random sample it is possible to measure the behaviour or mood of the entire universe of the population even by taking a very few people.

A sample of 2000 suffices to predict the winner of the Lok Sabha Elections. A random sample of 200-300 would sufficiently predict the response of the whole population towards a new product.

With those examples of little data it becomes evident that survey research is comparatively less costly yet quite accurate. However it is dependent upon the knowledge of the source, stimulus, context and history by the researcher. It is also important that the measuring instruments are tried and tested and the researcher has normative data, quality assurance and controls.