Hypothesis Testing

In statistics, hypothesis testing is done to draw conclusions about a population based on sample data provided. The data come from a larger population or a data generating process. It analyzes a sample from the population to make meaningful conclusions about the overall probability distribution of the population. 

All analyst use random sample data to test two different hypotheses, they are the null hypothesis and the alternative hypothesis.

Null Hypothesis (H₀):

It states that there is no real difference or change after the test. For example, if you test whether a new medicine works, the null hypothesis would say that the medicine has no effect. 

Alternative Hypothesis (H₁):

On the other hand, the alternative hypothesis is the result that will statistically say if there is an effect or difference. For example, if the medicine is tested, alternative hypothesis would say that the result has negative or positive effects.

Hypothesis Testing P Value (p value):

The P value is used in hypothesis testing to show whether the results of a test are statistically significant.

Important Insights of Hypothesis Testing

• You start with an idea or prediction, called a Hypothesis. In order to find the believability of the hypothesis, we use sample data.
  
   
• The given hypothesis needs an evidence which proves the statement to be true. For that, an evidence must be created.
  
   
• In order to find the evidence, statistical analysts test by examining the hypothesis using sample data. 
  
   
• Hypothesis testing consists of four steps: defining the hypothesis, locating evidence or data samples to support it, evaluating the sample data, and finally obtaining the outcome. 
   

Hypothesis testing is a statistical method to determine whether there is enough evidence to support a particular claim. For this, there are generally two types of hypothesis testing. Let’s understand them in detail.

One-Tailed Test: This test is used when we expect a change in only one direction. That is either an increase or a decrease, but not both. 

For example, you’re testing a new learning app that helps students score higher on a test. We only care about the improvement among the students. There are two types of One-tailored test. 

Left-Tailed Test: Used when we expect a decrease.

Right-Tailed Test: Used when we expect an increase

Two-Tailed Test: The test is used when we want to check for any change, whether it’s increase or decrease. Without knowing the direction beforehand.

For example, if we test whether a new marketing strategy affects sales, we don’t know if sales will go up or down. 
 

Although hypothesis testing is useful in many contexts, it has limitations of its own.  Let's examine a few of hypothesis testing's drawbacks. 

• Limited Focus: It only answers specific questions. May or may not capture the full picture of the problem. 
  
   
• Depends on Data Quality: If the data is bad or incorrect, the results can also be wrong.
  
   
• Might Miss Important Patterns: Since, it only tests certain ideas, it can ignore other useful information on the data.
  
   
• Doesn’t Always Consider the Context: It might oversimplify things by not looking at the bigger picture.
   

Hypothesis testing is a statistical method used in many field to draw conclusions of assumed statements. Let’s look at the different real-life applications of hypothesis testing.

• Medicine and Healthcare: A pharmaceutical company develops a new drug to lower blood pressure. They conduct a clinical trial.

• Economics and Finance: Analyzing economic policy, assessing market trends, and choosing investments are all aided by hypothesis testing. 

• Social Science: They use hypothesis testing to study human behavior, psychology, education and sociology.

• Market Research: Companies use hypothesis testing to understand consumer behavior, improve products and increase sales.