Before getting familiar with sampling techniques, it is necessary to name the basic terms. So, let’s define them.
3 basic definitions:
- “population” (the total amount of a specific element with a predefined set of characteristics);
- “sample” (subgroup of elements in the population)
- “Sampling” (the process of classifying samples)
this three link words A further distinction must be made:
population= global data array
sample = specific subset
Let’s take an example. Imagine having a field. There are many flowers there.
The percentage of dandelion is 40%, 40% = sample, and “all flowers in the field” =population.
You can sample in different ways, like 10% of each, or just select tulips.
sampling
This process was created to subdivide the entire group (population) into several valuable segments. These sections are usually differentiated either according to the goal principle or arbitrarily. The sampling technique defines the segmentation method.
If you want to learn more about advanced Artificial intelligence and software development techniques, check out this blogI got a lot of insights from their posts.
Two selection methods:
Probabilistic choice (sampling) (‘Simple (random) selection’; ‘Hierarchical selection’; ‘System selection’; ‘Cluster selection’);
- Simple (random) selectionG
The probability principle of this method is similar to the bottle game, when you need to spin the bottle to choose who to kiss. The “play a lot” principle. We use simple (random) selection when there is no specified information to point to the target population. Every element has the same chance of being selected.
benefit:
– lack of prejudice;
– It is easy to use and can be used for further analysis as a representative of the general population.
Unlike the previous one, hierarchical selection presupposes segmentation based on predefined target information.Segmentation is based on explicit feature (age, nationality, education, etc.), elements can be classified based on their similarity.
These subgroups are called stratification, and each of these subgroups is further sampled by randomly selected subjects.
As before, we can take the same example here. Although, there will be a modification. This time we are at a party. All players here are subdivided into small groups (classes) with common interests. People still play together, but only with friends. Then over time, the bottle randomly selects some group members.
benefit:
– higher precision
– Save money (high cost performance)
– proper representation
System selection uses arbitrary selection in the first stage and selects the following elements at regular intervals. The process functions like a chain with overarching elements that have the same opportunity for selection.
benefit:
– lack of bias (order of selection elements);
– Simple to use
There are three types of cluster options:
– one-stage cluster selection (select clusters arbitrarily = no specific features are considered before sampling);
– Two-stage cluster selection (1-arbitrary cluster-2 arbitrary element selection);
– Systematic clustering (the first element in the segmentation process is arbitrary; the following elements are sampled according to the predominant o-regular gaps between elements in the subgroups).
– multi-stage selection (using complex selection methods: 1 or a combination of methods);
benefit:
– Immutable
– Expanded sampling range
– easy to exploit
-Very reliable
2) non-probabilistic choice (‘Simple Sampling’; ‘Purpose Sampling’; ‘Quota Sampling’; ‘Recommendation/Snowball Sampling’);
This means that convenient/available resources are considered. This is the method for fallible elements (segments). Specific segmentation is achieved by the accessibility of groups within the population. The selection is based on the accessibility of the source.
In everyday life, it’s like we buy department-specific products in a store and then pick products from the shelves at our eye level, it’s very convenient, but you might miss bread or salt because they’re placed differently. The downside is that such selection results do not properly represent the general population.
benefit:
– Quick data collection;
-cheap;
– easy to use;
This approach is oriented towards the goals and intent of the research. Purpose sampling corresponds to qualitative analysis. It’s called the “judgment process” because the investigators involved in the study excluded the sample from the general population. My choice based on the purpose of the research.
benefit:
-flexible
– The purpose of accelerating research
-Easy to carry
= method performed using specific acquired features. Researchers usually create properties themselves.
= No randomization in this case.
= These self-created features are soliciting elements/samples and classifying them according to this or that principle.
= special function required
benefit:
– Cost-effective (can be used on a limited budget)
– This selection process is faster and saves time
– easy to execute
- Recommended/Snowball option.
This method is suitable for unspecified populations.
As with the recruiting process, a part of a recommendation sampling study (like people) can address surrounding elements (people from a close environment if the recruiting process is involved) and ask them about it. It helps to reach more designated members. Therefore, the unknown population (without detailed information) is segmented into more accurate samples.
(from abstract to concrete = from little snowball to big snowman)
The recommended sampling method is similar to building a snowman. If we want to make a giant snowman, we need to start rolling it with a small snowball (add the surrounding snow, take advantage of the surrounding space). When the snowball is big enough, we can shape the snowman.
benefit:
– Find samples quickly
– Simple use
– Precise data
For an in-depth look at modern scientific trends, please visit College page And watch video lectures.
What is the difference between probability sampling and non-probability sampling?
First, we should note that in probability sampling methods, every unit in the population has an equal chance of being selected. Furthermore, probability sampling is intended for statistical analysis. Therefore, it is very suitable for quantitative research.
Unlike probabilistic methods, non-probabilistic methods have different goals and advantages. This approach has nothing in common with the Arbitrary Principle. Also, not all members included in the population are usually selected.
It depends on the topic of interest/research purpose (research). Self-selection is also a feature of non-probability sampling. This technique is not suitable for quantitative research, but it is very suitable for qualitative research.
Post views:
9
