Often in our lives, we need to prove or disprove a hypothesis, to find out if something really matters or is it just an individual’s thinking, to find whether our idea really solves the problem of the masses or not. For this purpose, we use surveys for which a group of people are required. Now say suppose, you pick a group of people who have a very similar type of thinking, and do the survey. It is obvious that it wouldn’t give the actual results. That group didn’t represent the society as a whole. This is why you need an appropriate sampling strategy. To get information from a small group of people which represents a larger section of society.
I see a lot of people posting google forms on Current Campus Junta (Our College Facebook Group) for their research and projects. But is it the right way? You should know what is your target population and then ask relevant questions to them. How does one expect to get answers/validation from people who are not even your customers/target. Surveys like these will only give you biased and skewed answers which you definitely don’t want ! Also 10-20 responses doesn’t quite give a true picture so make sure that you select the appropriate sample size too.
Strategy (रऩनीति)- a plan of action designed to achieve a long-term or overall aim (Source: Google).
First of all, what is census and what is a sample? Census is everyone in the population and sample is just a piece of it which represent everyone (ideally).
When do we use census and when we use a sample?
Census is when we need actual data for each and every person and we have enough resources and time to conduct it. But if we want to conduct a study or draw some conclusions, we can run our tests on a smaller portion of the population which represents the society as a whole. And when you don’t have the time and money. By time I mean a lot of time. Just to give you an idea, population census of India takes around 10 years whereas thousands of study are done on a sample every day.
As the name suggests it includes some form of random sampling and thus give more accurate results. Probability sampling can be divided into-
- Simple Random Strategies: Consider you have a group of 200 people. And you want to run a survey on 40 people? How to choose them? In simple random strategy you start numbering those 200 people from 1. Then randomly choose numbers. You can think of it like taking out number from a hat which has chits of 1 to 200 in it. Or if you are working on computer you can generate a random number by typing a command “RAND”. Best used when whole population is available.
- Systematic Sampling : As the name suggests, in this sampling strategy, you create a proper system for sampling. Suppose you have 200 people lined up in front of you. And you need to interview 40 people. So you could create a system where you interview every fifth person in the group. This may look like the best way to do sampling but this might leave you with some biased opinions. Because suppose while interviewing people, you found out that there is no girl at the positions to be interviewed. Best used when a stream of representative people are present.
- Stratified Sampling: Here basically you create different groups( strata) of people based on their attributes. So in the previous example, you create different groups on basis of education: Secondary(60), Senior Secondary(30), Undergraduates(45) and Graduates(65). Now you want to select 10 from each strata. To select 10 people from each strata, you use Simple Random Sampling. Just give numbers to each person of the strata and pick 10 numbers from the hat. Best when there are specific sub-groups to investigate.
- Cluster Sampling: Okay this is confused a lot of times with stratified sampling. In this type too, you would be grouping people but that would be natural grouping. Suppose those 200 people live in different streets. Let’s say: 70 people live at Street A, 40 people live at Street B, 90 people live at Street C. Now these groups are naturally made. You didn’t make them, they were already made. Suppose you have to do a survey in these 3 streets. Now obviously it’s tedious to select people from every street. Here cluster sampling eases your job. You randomly pick a cluster using simple random sampling and go interview people from that cluster only. Suppose you get Street B. Now you interview all people from this street for your research. Best used when population groups are separated and access to all is difficult.
The elements that make up the sample are selected through non-random methods and are often selected on the basis of convenience. This gives less accurate information about the question in hand but is easy and time-saving.
Which one to choose?
A sampling strategy should be chosen to maximize precision within the budget and time constraints. So it’s totally your call to choose the type of sampling strategy !
However I found this website which may help you to choose the right strategy.
P.S. Those questions on facebook like “Which How I met your Mother character are you? ” and “Which GOT House do you belong to?” are actually surveys which are recorded for various market studies. So next time be sure while filling those random cool surveys that your data is being recorded somewhere. Anything you say can and will be used against you. 😛
Thanks ! Follow Baniya Buddhii for more updates.