# Statistics Concepts Explained

#### Introduction

Statistics is the study or science that involves collection, analysis, interpretation and presentation of data numerically. It involves collection of data from a community, organization, products and phenomenon’s with intention to analyze it in order to draw conclusions that predict or help to make informed decisions.  Statistics is applied in medicines, industry, environment, government surveys, and market research to predict what is to come to help in preparedness. Statistics has two broad categories. These are Descriptive statistics and inferential statistics.

Descriptive statistics is the type of statistics that involves analysis of data useful in describing, showing or summarizing data meaningfully in a way that patterns result from the data to allow simpler interpretation of the data presented. Descriptive statistics utilizes two measures to describe the data. These are, Measures of Spread which utilizes range, deviation and quartiles to analyze the data, and, Measure of Central Tendency describes the central position of frequency distribution of data using the mode, median, and mean. (Laerd Statistics, 2015). Usually, data is summarized and presented using tables, graphs, charts and statistical comments in Descriptive analysis.

Inferential statistics on the hand will involve taking a sample of data from a population or field of study and using that data to make inferences about the whole population under study. In simple terms, you take a small section of a large population, carry several analyses on it then use that section to describe or explain the entire population. Inferential statistics is used to forecast what should be expected without having gone through to be able to make decisions. It is used to predict market,for example, after examining common trends in the market. Inferential statistics will use threeprocesses that help to reach a conclusion on the usefulness of data results. It uses

Estimation which is the statistics that which allow statisticians to estimate population values based on data samples using parameter estimates and confidence intervals.Modelingwill help to develop mathematical equations that are meant to describe the interrelationships between two or more variables.Hypothesis testingwhich is used test whether a particular hypothesis developed is supported by a systematic analysis of the data (West Virginia Uni. 2015).

#### Hypothesis development and testing

For data to be used after a research or to make sure it is the best preferably for use to make decisions it must undergo hypothesis development and testing. This helps to determine which sample is a true representation of a population under study.A statistical hypothesis is an assumption or conclusion reached from samples of data collectedabout a population or process characteristics (parameters) or else about the form of the population or process distribution. Hypothesis testing basically is the process of determining which among the samples is a true or more accurate presentation of population or process under study.  The steps involved in the development and testing of hypothesis are described here:-

1. One needs to determine and define the type of criteria that is going to be used during the research. This is the research hypothesis definition.
2. The second step is usually to highlight how the whole process is going to be operated. You define what you are carrying out statistics for explaining how you will collect and measure data during the study determining the variables under study
• Next step is to set out the hypothesis. There may be several but what is important is to get the null and alternative hypothesis.
1. Determine and set a significance level.
2. Specify the testing statistic to be put in use together with a probable prediction.
3. Determine rejection rules for hypotheses
• Perform the statistical tests on the data while producing interpreted results or outputs
• Depending on the output result of the test, choose to either reject or fail to reject the null hypothesis. (Leard Statistics, 2015)

#### Selection of appropriate statistical tests

In order to achieve better statistical results it is paramount that one uses reliable statistical tests. It is advisable that one takes time to study and consider the available algorithms and choose one that goes well with the field under research. Selection of appropriate statistical test is very important for analysis of research data. Wrong statistical tests will always result to inaccuracy in decisions and projections outputted.

Wrong statistical tests can be seen in many conditions like use of paired test for unpaired data or use of parametric statistical tests for the data which does not follow the normal distribution or incompatibility of statistical tests with the type of data, etc. Because of the availability of different types of statistical software, performing the statistical tests become easy, but selection of appropriate statistical test is still a problem ( J. Pharm, 2010).

when selecting the best statistical test to use, the statistician should consider the following:

The kind of data being dealt with: The type of data may be nominal, interval, ratio or ordinal type of data. Nominal data cannot be measured or ordered but can be counted, it is categorical. The data can be binomial (e.g. male/female) or multinomial like in medicine where we have table/syrup/capsule.Interval data has meaningful quality and order depending on the quantity measured. However, the data is said to contain to natural zero. Ordinal data is also categorical but can be logically ordered in terms like equal, less or greater than like in scores, tastes, smell and ease of use. Ratio data has a natural zero. Height, length and weight are ratio types of data and can be said to have the natural zero. You can be able to clearly describe a double in length when comparing two things, for example.

Normal data distribution: It is also good to consider if the statistical data follows normal distribution or not. Parametric tests will be used if the data follows a normal distribution but if not then nonparametric tests are used. You can determine whether the data is normally distributed by using a histogram, plotting box and whisker plot etc (J Pharm, 2010)

The aim of the study: Having in mind what the study is geared to achieve is another important thing to consider. It will help much in choosing the appropriate test. You will know what you are supposed to compare with what.

#### Evaluating statistical results

Before submitting and putting results of a statistical research into use it is advisable to evaluate them. This ensure that they are usable and that the decision they bring about will be favorable or will bring an impact that is positive.

1. Determine Sample representation: the sample used in the study must be enough to give a good picture of the entire population or field under study. Determine if the sample used in the study may be biased or compromised in anyway and if it is fit to represent the entire field.
2. Analyze the qualitative aspect of the analysis: Go through the entire research looking at the data presented then in every aspect and see if the conclusion reached at is supported by the data trends presented.

• Hypotheses test: See if the null hypothesis has been disproved. Makes sure that the hypothesis used is disproved. If not at all or if not completely disproved then don’t rely on the statistical results.

1. Charts and Graphs: Have a keen look at all graphs, tables and charts used for data presentation to determine if they correspond to the data.
2. Verify Assumptions of Correlation Analyses: Ensure the assumptions that the research used are reasonable. If possible have a number of people give their assumptions to weigh the results.

#### Conclusion

Carrying out statistics is very vital in any country, organization or business venture. It helps people know how prepare for changes in the political, social, economic, science and medicine fields to be able to counter upcoming issues. Statistical work should be taken seriously and done cautiously. Knowing the right data to use, correct method of colleting it, how to test the data and finallyhow to determine if the conclusions arrived at are appropriate for decision making is a task to be keenly observed. Good statisticsis the ingredient for prosperity.