The Power of Choice

Random sampling is, without a doubt, very useful. But what you get with non-random sampling is the power of choice. Instead of leaving things up to chance, you are taking ownership of the finite resources you have and making a clear decision on who should participate in the research. With random sampling, unless you're contacting a large portion of your sampling frame, you have to hope that your sample is representative and informative. Other powers or benefits that come with non-random sampling are listed below (it’s weaknesses are covered below).

One of the strongest reasons to use non-random sampling is that it doesn't have a lot of requirements. For most researchers,  non-random sampling is the default sampling technique because they don’t  have a complete sampling frame.

For most researchers, non-random sampling techniques are all they can use.

But some issues come with non-random sampling that you should be aware of.

‍

Non-Random Sampling Issues

Think about the grid that represents a complete sampling frame. With non-random sampling, you don't have something so neatly defined. You have to swap out the grid for something like this: an undefined, boundaryless shape where you have no idea how many people are there or the probability of contacting them. People can move in and out of this shape, so it's even harder to know who will or can participate at any single time.

With non-random sampling, you can only control where and who you look for. The issue is that you define both where and who. If you only ever look for participants on a Facebook group, this undefined shape covers the people who go in and out of that group. If you don’t use non-random sampling techniques in places where informative participants are, you’ll lose its power of choice.

When you are personally involved in choosing who and who not to recruit, you also unintentionally bring bias. People have real, emotional , and evolving perspectives and outlooks on life (see Collection 3, Handbook 1: Strengthening your Qual Research for more on constructivism). Yours might lead to focus or look at one person while unconsciously ignoring others. Be cognizant about who you turn away — they might be more informative than you assume.

Findings from non-random samples are also hard to generalize back to your population or segment. You have no way of determining a margin-of-error or estimating how far off from the true/correct answer your findings were, as you can with random sampling.

Sadly, one of the most common – if not the most common – mistakes new/inexperienced researchers make is assuming that findings from non-random samples automatically generalize to the larger population. What's even worse is that these researchers are confidently telling their stakeholders about these "generalized" findings. You have to know when and how to use non-random sampling. You also have to educate your partners on what non-random sampling can and can't do.

Let's finish this handbook by looking at various non-random sampling techniques and when to use each one.

‍

Types of Non-Random Sampling

You can group the various non-random sampling techniques into three distinct groups: purposeful sampling, quota sampling, and haphazard sampling. Each group has its strengths and weaknesses. You might also choose to use multiple non-random sampling techniques within a single study. Let's start with quota sampling.

Quota sampling is when you knowingly select enough participants to match a known or expected population proportion. For example, if you knew that 50% of your target segment was older than 18 years old, then you'd want to make sure your sample had similar characteristics.

Think of quota sampling like riding a roller coaster. If there are only are only 10 seats on the entire roller coaster, you have to make sure to put informative people into all 10 seats to match what you know about your population or segment. Use this approach when creating or growing your own sampling frame of people to contact (commonly known as a research panel).

The second group is for purposeful sampling. Purposeful sampling is when you select participants based on your explicit judgment and reasoning. If you want to learn what it's like to order a pizza from the current online ordering platform, then you'd set out to talk to people that have recently ordered a pizza using the platform. You are the one who decides who is and isn't the most informative participant for your research and set out to recruit them (after aligning with stakeholders).

One common way to use purposeful sampling is when you want to study a very diverse population or segment. You’re looking for maximum variation to see what patterns and trends exist within and across each type of person. Use purposeful sampling when you have a very specific MIP definition and/or with qualitative research questions.

The last group contains the most dangerous of all the non-random sampling techniques: haphazard sampling. Haphazard sampling isn’t so much about choice when sampling, but desperation. These techniques are dangerous because they're based on random luck or opportunity, not by reason or math. Haphazard sampling should only be used in situations with limited time or resources. Otherwise, it can be a careless and poor way to recruit participants.

Within haphazard sampling, there are three common variations of haphazard sampling: convenience sampling, voluntary response sampling, and chain sampling.

Convenience sampling is probably the most common haphazard sampling used. Convenience sampling is when you recruit participants that are easy or fast based on your limited resources. If you care about understanding what it's like using public transportation, you can wait at the bus station and talk to whoever is there. Convenience sampling is limited by the number of resources that you have. If you have a lot of help and time, you can actually cover a lot of distance or participants.  Use this haphazard technique as a last resort.

Voluntary response sampling is where you let people opt-in or self-select themselves to participate in your studies. This happens when you have a way to get contacted by potential participants, like posting a recruitment link on your company's website or sharing a general survey link on your company's social media. People participate because they're motivated or interested in what you're studying.

This approach is similar to putting up mailboxes wherever you’re trying to recruit. Whoever notices and cares enough about the mailbox will use it, while others might see them and do nothing. Use this approach to hear from very vocal or passionate people or those who identify with the purpose or questions of your study.

The last haphazard sampling technique covered here is chain sampling. Chain sampling is when you ask current participants if they can suggest or recommend other people that might be interested in taking part in your research. This is a very effective approach if you are working with very small populations (such as a startup) or when recruiting from niche populations (such as wanting to learn from a very specific type of dentist who uses a specific type of dental software). It’s also an effective way to grow the number of people you have access to.

Non-random sampling isn't necessarily bad; it just depends on when, how, and why you apply any one of these techniques. Try different approaches to see which one works best for your research needs.

‍

Closing Thoughts

Sampling and recruiting for your studies is more complex than just choosing some people to interview or sending a survey. If you can understand the underlying concepts or the math that fuel sampling, then you can make smarter decisions about who and how to select participants for your studies.

But this begs a very important question: how many people should you select? In the next chapter, let’s discuss, determine, and justify practical sample sizes for your research.