 Hello guys, welcome back to my classes. So today, we are seeing about sampling designs. So in last class, we had covered probability sampling design. So in probability we had seen simple random systematic stratified cluster and other sampling techniques. So today we are seeing about known probability sampling, this convenient judgment and water and snow sampling. So tagline is all are not equal. The tagline of probability was all are equal. The chance of being into the study was equal for everyone, but in nonprobability the chance is not there. So it is nonprobability, all are not equal. So let's go into detail about nonprobability sampling. So basically we have four methods that is convenient sampling, judgmental snowball and quota sampling. So the basic difference between probability and nonprobability, there is no random selection here, nonprobability sampling. So the result will not be very accurate as obtained from sampling design. We just cannot accurately extrapolate the result of the study back to the population. So every study is meant to extrapolate the result obtained from the sample to the population because it's very impractical to do a study in a big population. So we conducted study in a small population which is a representative of a big population, a small sample which is representative of a big population and we extrapolate the result obtained from the sample to the population. But if we are doing a study with a nonprobability method sample selection, the extrapolation is out of question. It will not be as accurate as probability. So we'll come back to the methods. The first one is convenience. It is based on the accessibility. What is accessible for the investigator? He takes a sample from that area or that group of people. It is totally based on his accessibility. Second one, the judgmental. It is the investigator notion about the hypothesis or his idea about the research. We deliberately select sample to make sure that that follows some criteria. Quota sampling is nothing but a corresponding part of stratified sample. In the probability there we had seen homogenous group making from a heterogeneous group, boys, girls, such strata we make here. With the problem is the simple random technique will be missing in quota sampling. Snowball sampling is a very unique technique where sample will be selected by the reference of one person and it goes to the next and then next to the next. It goes like this until we reach the total sample size. So we'll move on to the first one that is purpose sampling or it is also known as judgmental sampling. So research select a particular group which might represent the population. So it is totally based on his judgment, the investigator judgment. So we just collect a group of people. There is no evidence base. It is an option about the hypothesis or his idea, the investigator's idea that he thinks it might be the representative of this population. Suppose we go to a dental college we take a group of people. He sees the first 100 students and he thinks that it will be the representative of the entire students. That is his idea about the college. It might not be the actual scenario. So non-probability always lacks quality. So that is the purposeive or judgmental. His judgment is working on the sampling. Second one is a convenience. This is the most common used sampling design in a big population because when we are doing a study we need to do the study in a very short span of time. So what we do is we take sample from the accessible areas, the nearby areas. So most conveniently we get units from nearby areas because the cases are readily available. Suppose we are doing a study among the school children. So we don't go very far away from our place. So we go to the nearby schools and we take sample. So that is based on our convenience. It doesn't mean that the nearby schools are representative of the actual school children. It might not be the case. But we are going and taking this course based on our convenience. So that is convenient sampling. It is totally based on the convenience of investigators. So the researcher stays here. So he takes the nearby population. That is convenient sampling. Judgement lays about his understanding about the study. He thinks that this will be the representative of the population and he selected sample. So the quota sampling is nothing but the counterpart of stratified sampling, which we have seen in the probability sampling. Here what we do is we needed a sample interviewer, needed 40 adults and 20 adolescents in order to study the students' television viewing habits. So what he does is he takes 20 men and 20 women and he takes 10 adolescent girls and 10 adolescent boys. So the best part of quota sampling, it maintains the representative units because we had seen if we are not done quota sampling, all people might be come from one subgroup. The other subgroup might lose its representative units. So quota sampling is almost like stratified sampling because a heterogeneous group will be divided into homogeneous and a particular quota will be selected from each. So suppose if we are going back to our dental college scenario, we needed 100 people. So we have five batches, one first year, second year, third year, final year and interns. So what we do is we suppose we need 100 people. So we take 20 people from each category. So that itself represents the whole college. But the point we are not doing is we are not taking random selection from each batch. So that was done in probability sampling. So quota sampling, a fixed quota of 20 will be taken from each subgroup. So that is a counterpart of stratified sampling. And the last one is snowball sampling. Snowball sampling, it's a very interesting sampling technique. I'll just show you one picture. This snowball, okay, this snowball, this idea of snowball is when the snowball, they say when snowball is starting from one point as it grows, it increases its size because it collects snow over the track. So that's how this snowball is working. Snowball, it is done to select population, very people from populations such as homosexuals, such as any peculiar diseases, which are very obscure. So people might not be knowing actual cases, but one person can point out the next person. Because if you are doing a study like smoking habit in a college, nobody will turn up for the study. They must get first find out one person. And we asked that person to find out two person keeping that the person's details will be confidential. So we can get the next two people from that two people again, we can get next four people. So it goes like chain by chain. So every time the person who is selected first will be getting the next two sample or next three sample, the investigator has no role after the first point, the sample itself selecting the further sample until it reaches the ultimate sample size. So the researcher asks for a referral from other individuals. So it is also known as chain sampling, chain referral for sampling referral. So just like a snowball, as it grows down the hill, it increases size. So once when we starting the study, it is just one or two samples. So as it grows, they select other samples from the hidden group. So it grows day by day and we get the sample we need it. So that is snowball sampling, just a diagrammatic representation. The first person select three person each one person select the three person. So likewise, we get the total sample size. So probability and non probability sample, the main difference is we cannot generalize the result of the sample back to the population. That is a possible only in probability because probability sampling maintains the equality. Every participant had a chance in probability does not have any equality. So we cannot just put back the result to the population. So it is usually generate hypothesis, probability it is to test hypothesis. So probability sampling usually eliminates bias, but non probability so many biases will come but non probability as most of studies will be non probability because it is cheaper, easier and quicker to carry out probability sampling techniques. It's very difficult to conduct usually when we have a very small sample size. If you're doing a randomized control trial or a clinical trial with very small population like 1020 3040, we can follow the simple random technique or any other such techniques. If you have a very bigger sample size, this probability is very difficult to keep up. So most commonly used is non probability even though it has flaws. So that's all about non probability sampling. There are various techniques. I'll just recap. So first one was convenience. It's based on the patients convenience. It is the nearby locality people or participants will be selected judgmental or perceived sampling. It is based on the judgment of investigator and the quota sampling just like the stratified sampling. Each quota will be selected from each homogeneous group but there is no random sampling snowball sampling from referral chain sampling. The first person will select the next few people and those few people will select the further until we reach a sample size. So non probability it's very easy to conduct but the quality is less compared to probability one. So that's all about non probability sampling. So I'll come up with a new class. So thank you.