Non-Probability Based Sampling

These techniques are simpler and cheaper than Probability-Based Samplings. These techniques do not follow the concept of probability, meaning the probability of selecting any item is not equally likely, because it depends on the selection criteria and upon your requirements.

Suppose, we want to collect some information over the internet, so we need to target some audience first, and then the survey will be conducted among this selected audience only. But the target audience may differ from other populations. We may target people who are available on Facebook, but there is some population who is not available on Facebook. We may target people who are going to the gym, but others are not going to the gym.

Therefore, we have some selection bias in these sampling techniques, hence the output we get may be biased or maybe not.

Techniques falling under this category are as follows:

1) Convenience Sampling:

We collect the sample at our convenience. The Source of the sample can be anything, be it online or offline, or nearby, but the motive is to get fast access to the sample so that the rest of the time can be spent on analysis. The Analyst is not concerned about the accuracy or biasness of the sample selected.

2) Quota Sampling:

Quotas are decided for sampling. Quota (or reservation or selection criteria) means some sort of rules are decided and based on that, a sample will be collected. Researchers or Business leaders or analysts decide some quotas based on their experience or based on their requirement and according to this quota only, the sample will be chosen.

E.g. Select one sample set of 1000 cars, out of which 30% cars will be Tata’s Sedan, 20% will be Maruti Suzuki’s Sedan, 15% will be Ford’s Sedan, 5% will be Skoda’s Sedan, and the remaining 30% will be Hyundai’s Sedan, and after sampling get their average weight in Kilograms.

Here there is a strict restriction that for your sample, Tata Sedan car's count cannot be lesser or more than 30%, it should be exactly equal to 30% of 1000 which is 300 cars.

3) Purposive Sampling:

The objective of this technique is to target a specific set of items/people/places or things that have some characteristics in common. Items that do not have these characteristics will not be part of a sample.

For example: We want to know whose company’s stock price will go up in near future? Is it Microsoft, Google, Facebook, or Tesla?

Here we will target those people only who have a business mindset and who are educated enough to answer this problem.

So, the right place for conducting this survey will be LinkedIn instead of Facebook, Youtube, or Instagram.

4) Snowball Sampling:

This sampling technique is also known as chain-sampling where one item is chosen and the items which are linked to this item are considered and then this chain goes on. The logical idea behind this technique is that items that are linked to this item will share some kind of common characteristics.

Let us take a look at the below examples to understand some situations where this sampling technique is beneficial:

• Suppose one robbery occurred in the city, so police cops will find out the person on which they have misdoubt. After catching this person, police cops will do their investigation and ask for other suspicious persons who might have done this robbery.
• Banks call the borrower who has taken a loan recently and they ask for other potential customers who need money and can take out a loan from the bank.
• Suppose I want to know who all are the Maths teachers in my city? Find out one Maths teacher in your city and ask the details from this teacher about the other Maths faculties.
  

This is how we can find out linked entities that are hard to find without this sampling technique. It is useful for situations where entities are linked to one another.