What is sampling?
Sampling is a method which helps researchers save time and money through carefully selecting representative of a specific population. Sampling is common in surveys or quantitative research. With sampling, a marketing or a social researcher conveniently gets information about a population by only interviewing a small subset of the population. This way they can tell what percentage of deodorant users are loyal to brand X.
Furthermore in polling it is possible for a researcher to deduct that a certain percentage of the population with vote mostly to candidate A. Imagine how impractical and costly it would be if the researcher were to ask all the voters who they will vote for in the forthcoming election. Apart from inconvenience of interviewing the entire population, such a task would impossible to accomplish. It is expensive to reach the entire population and even governments carry out national wide census every ten years.
Sampling is therefore a scientific process. There rigorous mathematical process of drawing a sample. The process used must ensure that each selected subject is representative of the entire population.
There are two types of sampling procedures. These are probability sampling and non-probability sampling.
How AfriQSense does probability sampling
The probability sampling is also known as random sampling. In this process, a complete sampling frame from the target population is created. Assuming is a general population survey to determine religious practice in country X, the sampling frame will the country’s latest population census as long as the data has religious information. The sample is then stratified using administrative divisions. For example in country like Democratic Republic of Congo, the population will stratified based on the country’s administrative divisions as shown in the table below.
The sample will first be shared by 25 provinces and the city-province of Kinshasa. Each of the provinces will get a share of the sample according to its population. For example, according to macrotrends, metro or city-province Kinshasa or has an estimated population of 14,342,000 as of 2020. According to worldometers, the Democratic Republic of Congo has a 89,561,403 as of 2020. Therefore, the Kinshasa’s population accounts for 16% of the population. Consequently if our national sample is 1000, then 160 of the respondents will be drawn from Kinshasa. The 160 will then be distributed into the city’s four districts, namely Funa, Lukunga, Mont Amba and Tshangu districts based on their population share.
Each of the districts is further divided into municipalities or communes and so will be our sample, this too will be based on each of the commune share of the population. However, to ensure that we do not spread the sample thin, not all the municipalities or communes will have a sample. For example, we may have 5 interviews for each Primary Sampling Units (PSU) which means that the 160 interviews once shared by the districts, the municipalities or communes will be selected randomly and the sample shared by the selected units. This helps in saving cost. Each PSU is work for a day for each interviewer. Similar process will be repeated in each of the other 25 provinces. Such is an example of a Stratified sampling, a type of random or probability sampling. There are several types of probability sampling. These are:
- Simple random sampling
- Systematic sampling
- Stratified sampling
- Clustered sampling
How AfriQSense does non-probability sampling
The non-probability sampling is preferred by some since probability sampling is time consuming which means it’s more costly. Non-probability or non-random sampling does not require a sampling frame. This reduces representation since all the members of a population do not have equal chance of being selected. The risk in this type of sampling is that it is not representative and hence cannot help in generalization. Advantages of non-random sampling are in its convenience and lower cost. This type of sampling is used in exploratory studies or when clients want to generate hypothesis. Examples of non-probability sampling are:
- Convenience sampling
- Quota sampling
- Purposive or Judgement Sampling mainly used in qualitative research
- Snowball sampling
We work with clients to ensure that sampling for each research meets the objective of the study. Contact us for your research fieldwork in Africa