Inferential Statistics

Descriptive Statistics merely provides a summary of a sample drawn from the population. It describes characteristics of the sample in terms of location, spread, and frequency etc.

Inferential Statistics goes one step further to draw inferences from the sample, largely using the central and spread measures, about the population from which it has been drawn.

This extentison of our knowledge from a particular random sample to the whole population is called inductive or statistical inference. The main function of business statistics is to provide techniques for making statistical inferences, and measuring the degree of uncertainty associated with such inferences.

Necessary Conditions for Statistical Inference: The following conditions are necessary for almost all the statistical inference tools used to draw inferences about the population.

1. Outliers in the sample data may have major impact and will influence the results of almost all statistical estimation and hypothesis testing procedures. It is important to treat outliers before carrying out the tests.

2. The population we are trying to make inferences about is expected to be Homogeneous. That is, there is not more than one mode.

3. The sample must be collected in such a way that it is representative of the population. That is, there are no processes (or errors) that systematicly influence the sample. For example, a biased sampling procedure will lead to incorrect inferences.

4. In addition to being homogeneous, the population is also expected to follow the normal distribution. This is one of the imporatant necessary conditions.

5. When inferences are needed to be drawn about two or more populations, for example to test the equality of two population means, their variances are expected to be almost homogeneous.

Inferential Statistics involve two branches of statistics: Statistical Estimation and Hypothesis Testing. Inferential statistics is an integral part of the statistical Information Techniques.

This site is designed to be informative to cross-disciplinary university students and research scholars, and anyone interested in dealing with uncertainty problems. We plan to include more numerical examples to make it interesting and useful for the the high school students too.

Back: Sample Variance - Biased?

Next: Statistical Estimation