Applied Statistics

Applied Statistics

 

Statistics refers to the scientific discipline of data collection, organization and

interpretation, which includes the data collection planning to be used in surveys and

experiments, and is also considered a distinct mathematical science. The field uses

data and statistical models to enable forecasting in such areas as the natural and social

sciences, government, business and sports. Descriptive statistics refers to the summary

of a collection of data that can be used in research, including the modeling of data in a

way that would indicate randomness in observations. Inferential statistics refers to the

use of such random observations to draw inferences on the studied population.[1]

 

Statisticians are specialists in applying statistical analysis. They design experiments and

conduct survey sampling to improve data quality and can help various organizations

without having access to knowledge relevant to their specific problems.[2]

 

Applied statistics refers both to the work of professional statisticians who produce and

present statistics and to statistical interpretation as performed by statisticians, by

statistical users and by the general public. It often use statistics theory to provide a

ready source of information and has contributed significantly to scientific research.

However, applied statistics has also been a source of misinformation, such as in the

statistical misinterpretation by some in the medical, legal and financial professions that

have led to erroneous diagnoses, imprisonment and financial risk assessments.[3]

 

Statistics are numerical observations of facts and therefore fallible. Their reliability

depend on the representability and accuracy of the observed facts and their correct

interpretation. It is also important to appreciate the methods by which statistics are

generated. A controlled experiment is the easiest method to use because disturbing

influences can be minimized, while the design of experiments range of methodologies,

including its associated software packages, is used when there are lesser degrees of

control. Sources of error in taking samples include the variable results every time a

sample is taken and the difference in the qualities of the sample from the population it

intends to represent. Most published and unpublished statistics contain elements of

subjectivity, which is why their objectivity is usually subject to review by an independent

body. The reliability of published statistics are usually only available when a variable is

is obtained from multiple sources or when revisions are made, as in the differences

between victim-reported and police-reported crime rates.[4]

 

Some statistical findings can be successfully interpreted verbally and by a mixture of

verbal logic and the rules of chance. Their successful use involves the combining of

available tools of inference with their safe application to a particular problem.

Managers of such work have the responsibility of understanding statistical concepts,

including their limitations, before making a decision based on them.[5]

 

The quantification of intuitive significance and confidence that would allow objective

answers on questions like how likely is it or how confident are we in relation to a certain

subject are major contributions of statistics theory. Statistical correlation can also be

used to explore a connection when the possible errors in the available evidence are

attributable only to chance.[6]

 

The human brain’s making of intuitive probability judgments is usually erroneous even

if it is fairly right in intuitive judgments such as those involving speed or distance. The

use of heuristics instead of analysis causes distortion in numerical data interpretation,

such as in the overestimation of bigger events like rail disasters and the

underestimation of smaller events like road accidents. Statistical errors such as

misinterpreting non-significant  findings as significant have occurred in medical research

reports in two respected medical journals. Survivor bias has been used by financial

analysts to statistically mislead potential clients to show the benefits that would have

been obtained if investments had previously been made in their currently recommended

portfolio.[7]

[1] “Statistics”, Wikipedia, 4 May 2011, <http://en.wikipedia.org/wiki/Statistics>  [accessed 9 May 2011]

[2] ibid

[3] “Applied Statistics”, Citizendium, 17 February 2011, <http://en.citizendium.org/wiki/Applied_statistics> 

[accessed 9 May 2011]

[4] ibid

[5] ibid

[6] ibid

[7] ibid

Credit:ivythesis.typepad.com