Hello everybody. Welcome to this video on sampling.

Before we discuss anything more about how to collect data,

it is crucial to understand that in your research,

you will not be able to research your entire target population.

As you cannot talk to every stakeholder or expert involved in your research,

you have to make a selection.

This is what we call sampling.

In this video, we will discuss which methods of sampling you

can use for your research. Let's get into it.

The two most important methods for

sampling are probability sampling and purposive sampling.

When you use probability something,

the sample is drawn from a population based on probability,

so random, with the aim of being representative.

Probability sampling is used in quantitative data collection methods.

The other method for sampling is called purposive sampling,

where you specifically select those people that are supposed

to have the most relevant knowledge on your research subject.

According to theory and to your knowledge of the situation,

this method is quite often used in qualitative data collection methods.

In this video, we will discuss both types of

sampling and provide you with several approaches.

Let's start with probability sampling.

I said the sample is drawn from

a population based on probability with the aim of being representative.

This means that your sample is random.

There are several approaches to create a random sample.

In this video, we will focus on simple random sampling and stratified random sampling.

An example, let's assume that you wish to

study the behavior of master students specializing in

urban developments at the Institute for Housing and Urban Development Studies

in Rotterdam through quantitative analysis.

You require a random sample to do statistical analysis.

The first option would be a sample random sample,

which is selected in a way that everyone in

the sample population has an equal chance to be picked,

making the process indiscriminate or unbiased.

For instance, drawing a name from a bowl containing all the names of

the students in the program is an example of a simple random sample.

Each name that is in the bowl has an equal likelihood to be picked.

Another option is what we call stratified or segmented random sample.

This is obtained on the basis of a representative segment or strata.

It involves separating the population into

mutually exclusive sets or strata and then drawing the samples from each group.

In other words, it requires the population to be divided

into the smaller groups or categories which we call strata,

as you see the image displayed here.

It is still random,

but will be better distribution among the specific groups.

It may give you more relevant information and also might be more cost effective.

Let's say for example,

you have a selection of masters students in IHS Erasmus University,

you may choose to first divide them into the strata,

like two different master specializations,

and within the strata, draw your sample.

This would ensure an equal distribution over to different specializations,

making it a better reflection of the population as a whole.

Both options discussed now,

the simple random sample and a stratified random sample,

will lead you to a random sample based on probability.

This means it will enable you to generalize the results to

a target population as a whole and do statistical analysis.

Therefore, your sample should sufficiently be large to represent the entire population.

A representative sample can be calculated from the size of the entire population.

Sometimes it is not possible to reach a representative sample.

In that case, bare in mind that

statistical analysis requires a minimum number of observations,

usually set at 30.

However, if you want to analyze differences between subgroups,

you should be aware that the sample size should be allowed at

each subgroup has at least 30 observations or respondents.

Indeed, the last option will provide you with a stratified random sample.

The first option is actually a purposive sample,

which we will discuss more in a bit.

The second option would be a simple random sample.

Okay, let's continue.

Random sampling should never be confused with convenience sample,

which is very common among researchers.

Samples chosen through convenience sampling are conducive to

circumstances such as time of the research or self selection of the researcher.

For example, you decide to carry out research during a rainy day,

you may choose to interview people at a cafe.

This is convenience based on the weather conditions,

but it affects your sample.

The subjects are selected because they are easiest to find or to recruit.

This produces a sample bias.

Definitely it is not fully representative of the whole population

hence limiting the generalization of the data

and making inference about the entire population.

Furthermore, it results to a low external validity.

When using convenience sampling,

it is necessary to describe how your sample

differs from a sample that is randomly selected.

You may also need to indicate individuals who may be left out doing

the selection process or are over represented in the sample.

For qualitative research, this is different,

and purposive sample is better suited.

The aim of qualitative data collection is to gain new insights and knowledge.

Therefore, it is important to find

those respondents that can tell you most about a certain situation,

process and et cetera.

To illustrate the difference between probability and purposive sampling,

we maintain the quality of public space in a city as an example.

If you would use a probability sample,

you randomly question a number of respondents through,

for example, a questionnaire or structured interview.

You are done and able to generalize the results,

but what you cannot describe from these results is why they view it like this,

and what are their opinions of perceptions,

and what are the processes taking place.

To do that, a purposive sample can be useful

selecting respondents based on their knowledge of the quality of public space.

In this example, you could choose,

for example the employees of a municipality or local governments working on

public space or a key figures in

the local society or academics who have researched the quality of public spaces.

They may all give you more insights.

That is why for qualitative methods,

a purposive sample is most often use.

Purposive sampling include snowball sampling.

Which means that you use your respondents to find or recruit other possible respondents.

Your first respondents will therefore lead you to the next ones and so on.

Another option is quota sampling,

which is also purposive but it looks like

a non-random version of the stratified random sample that I mentioned before.

With quota sampling, the population is also divided into

different groups according to specific characteristics,

but the sample is taken from these groups based on the researcher's expectations.

For example, if we go to the quality of the public spaces,

an interviewer may want to sample key figures in community,

local governments and researchers,

then these three categories form the groups within which several respondents are chosen.

This way, the researcher is able to reach a certain quota within the three groups.

As you may already understand,

in any research you conduct,

it is important to think about which method you use to sample your respondents.