A Good, Reliable Sample – Possible?

Just what is sampling?

As psychologists we use different sampling methods to find trends within a population. Testing the whole population can be both expensive and time consuming, so for this reason it is much easier to test a representative sample of the population. From the results of testing this sample we can make an assumption across the rest of the population and progress our research from there.


So what can go wrong?

Well for a start we have to look at ‘sampling biases. If the participants were not selected randomly, they may have been chosen with the aim of generating the best results relevant to the test – this can seriously undermine the validity of the test. For this reason it could be said we should aim to use a ‘random sampling method’
While more reliable in one respect, it generates the issue. As the participants are chosen at random, the minority of the population could be chosen entirely and outweigh the majority in the sample. For example, if we were to look at the amount of people with a learning difficulty in a small town of 100 people, we could end up with a sample of 30 with a learning difficulty and 20 without – Whereas true results should show 30 with learning difficulties and 70 without. This problem can be helped by selecting a much larger sample, but it will still exist, albeit to a lesser extent.


But what about individual differences?

Everyone is different. This is something we should be taking into account as Psychologists – After all, it’s what our work as all about. But does this mean that the results from a tested sample cannot apply to the rest of our population? Well, in my view yes and no. We can say that there is a ‘good chance’ our results will also apply across the rest of the population, but we cannot confirm this to 100%.


Okay, but that’s just a sample. What if we tested the whole population?

While testing a whole population would be both time consuming and expensive, we could gain some much more accurate information from the results. However, the problem would exist again in that we could not necessarily apply these results across any more than the population tested. Simply put, to apply results to the rest of the world, or even one country, we would have to take into account individual differences and cultural biases on a much larger scale.


So is there any point?

Yes. Even though all of the above seems to run the idea of sampling a population into the ground, the act of sampling does give us a very good idea of how to progress our research. Sampling helps us to understand trends in a population and points us in the right direction for further research; without it we would be pretty much ‘stabbing in the dark’ to find a conclusion.


About DJPsych

... And now I find my self in my first year at Bangor University (Wales, United Kingdom), studying Psychology with Neuropsychology With the long term goal of gaining my Doctorate. In the shorter term however, I intend on sharing with you all, my thoughts and opinions across a wide variety of subjects within the vast realm that is 'Psychology'. However, if you want to hear (Or indeed read), my opinion on wether Psychology can be called a 'Science', then it would probably be best to email me (: View all posts by DJPsych

4 responses to “A Good, Reliable Sample – Possible?

  • ihmsl

    Hey DJPsych, liking your blog very much, it shows that you have carefully considered the the pro’s and con’s when carrying out research. Yes I agree that a sample is effectively a representation of the population. Yes also to random sampling as the best possible option with the least bias but even without researchers input this can sometimes introduce bias without realisation. Whilst I agree with the fact that we are all individuals, there are common behavioural characteristics that we also share. Of course sampling is a necessary method on balance as the alternative would be completely unrealistic. Lastly sampling is completely worth it, as you have identified it gives us a window through which to view what may or may not be the case in terms of human behaviour.

  • csg3bu

    I agree with you that sampling is extremely important but surely stratified sampling would be a better method than random sampling? That way the groups of the population are looked at in numbers that reflect how they actually exist in reality as well as having the random selection within those groups to avoid bias as much as possible.

  • daniking1234

    I think a good, reliable sample is possible. For a sample to be a good sample I believe that, like you said, the sample should be picked completely at random as this does not face any biases as if say, volunteer sampling was used, the people who volunteered are likely to be similar to eachother and the results could not be generalised. For a sample to be good and reliable it should be a large sample as it would contain many different types of people and that should make the findings generalisable.

  • rsmpsychblog

    the question you are asking is a tricky one! instead of random sampling i would use stratified sampling then we gain a representative sample of all the different types of behaviour. personally although time consuming i believe that this is the only way to gather a representative, and reliable sample, because with a random sample you always have the possibility of gathering all the participants displaying one type of behaviour.

You must be logged in to post a comment.

%d bloggers like this: