“If we don’t know who you researched, we are not interested in what you found out”.

Think about it…… “we surveyed 400 people to find out where the most beautiful beaches in the world are and 98% agree that Brazil has the most fantastic shoreline”. If we don’t know who you asked, we can’t do anything with the result.

How did you select who could answer the questionnaire, to what extent can you claim your data is representative?

If your survey was posted in Brazil and mostly answered by Brazilians, then, what a surprise, most people are patriotic when it comes to describing their own country. If you’d done the survey elsewhere, there’s a very big chance you’d have got different results. Ask the Colombians who has the most beautiful beaches!

The media is full of these headlines: 88% of a randomly selected group, stated that university education should be free. Well, what they don’t tell us, is that this group of people were 20 in total and randomly selected outside a uni and on a uni campus.
So.. hey presto, a few (not most) students (studying at this university) think uni education should be free. As you can hopefully see, without that extra bit of info, the results are pretty meaningless and in actual fact we can call it bogus reporting.

Ok, so enough of the stories, now let’s get down to what you should do…….

When carrying out primary data collection i.e. typically survey or interviews, you need to decide who to collect the data from. There are several ways of doing this and they form a vital part of your methodology description as the reader needs to know the source of the data.

I’m not aiming to give you a list of definitions, but here is a basic overview of what you need to know and include.

The vital questions I will address to help you are: What is a population? What is a sample? What is sampling?

Here we go!

1. What is a Population of a Study?

“The set of entities about which a researcher wishes to draw conclusions” Easterby-Smith et al, 2018. (Don’t forget to check out this source and many others on our list of recommended research methodology publications).

The population is who you are actually researching i.e. you are researching the attitudes of Germans, aged between 18 and 30, to Brazil as a holiday destination. The population for this study is Germans aged between 18 and 30. Obviously, you can’t contact ALL Germans between 18 and 30, due to time (would take forever), access (no phone numbers/mail addresses available etc) and money (would cost the earth). So….

2. What is a Sample?

…. you need to sample.

“The segment of the population that is selected for research.” Bell et al, 2018.

Most importantly is that you want the sample to be as representative as possible and to be able to generalize from your results as much as possible.

You can’t do this, if in the above (attitudes to Brazil study), you ask only 20 people who are all women, 18 and students etc etc etc. You could then end up with results which actually only tell you what twenty, 18 year old female German students think. Interesting results, but not reflective of your population or able to fulfill the aims of your study or answer your research question.

3. What is Sampling?

Don’t forget the size of the sample matters. As shown above, twenty people isn’t representative of very much, if the total population is 20,000. Check out the margin of error to see if your sample size is adequate – you can look this up online and use many sites including the margin of error calculator from Survey Monkey.

The actual process of sampling involves you checking out some definitions and remembering that fundamentally there are two types:

A) Probability sampling

Description: Is the type of sampling when you can calculate the chance of people being selected to take part and it is more than 0.

This always involves a SAMPLING FRAME i.e. a list of the population. If you are asking German students at a particular university, then the list of students registered at that uni is the sampling frame (i.e. the list you sample from) and you could use a random number generator to select those who will be asked to participate.

• …Imagine you want to know the attitudes to Brazil of the local volleyball club, you can use their membership list as the sampling frame.

• …Imagine you want to know the attitudes of google employees to Brazil, the list of employees is the sampling frame.

In this case, you need to use one of the following:

• Simple random sampling: each case in the population has an equal chance of being selected (meaning you have a list, stick a pin in it, put names in a hat, use random number sampling).

• Systematic random sampling: select the point in the sampling frame to start selecting and then select at random intervals (for example, on a list of 10,000 employees, you start at employee 1 and select every 10th thereafter).

• Stratified random sampling: divide the population into two or more relevant strata (for example, gender, origin, those who’ve been to Brazil, those who haven’t) and then select at regular intervals.

• Cluster sampling: divide the population into groups or clusters before you sample (for example, divide google by its locations before sampling employees). Then carry out random sampling (simple or systematic).

• Multi-stage sampling: This is the next stage of cluster sampling. Take cluster samples and use random sampling within each of them.

Don’t forget, all of these, can only work with a sampling frame!

B) Non-probability sampling

Description: Is the type of sampling when you cannot calculate the chance of someone being selected to take part.

• …Imagine you want to find out the attitudes of people walking down the high street in your town to Brazil as a tourist destination. Nowhere is there a list of who will be walking down the high street.

• …Imagine you want to research the attitudes of fans of Brazilian music to Brazil, you’re not going to find a list of these fans to select from.

In this case you need to use one of the following:

• Quota sampling: Check that the sample contains certain characteristics of the population you have chosen (for example, you ask X number of males, females and stop when the quota has been filled).

• Convenience sampling: Choose a sample that is easiest to reach (for example, you put your survey on social media).

• Purpose (or judgmental) sampling: Use your judgment as to who makes up the sample (for example, to achieve enough diversity/similarity in your sample as you don’t want only people from Germany to answer your questions, but you also don’t want 60 different countries in a sample of 100. Similarly, you might need people with specific knowledge, if you want to ask about the attitudes of Brazilian music fans, you need to sample, at least some people who are part of this group).

• Self-selection sampling: Let the participants contact you (you put out a note asking people to take part, if they have certain characteristics, for example, are fans of Brazilian music).

• Snowball sampling: Get participants from other participants (for example, you don’t know many fans of Brazilian music, but those you know can provide information to contact others).

Final Thoughts

These are the basic definitions, but I’m afraid you still need to check out why each makes sense, which problems may arise and which sampling method is best for your project. I’ve used Saunders et al, 2015 for the definitions.

So to end up, hopefully it’s now clear that:

• Results can only be interpreted with clear information on sampling.
• A sample aims to represent the whole population.
• The population is the focus of the research.

Ps: Where are the best beaches in the world? # Brazil.