A level psychology help

When you're evaluating a study and you're talking about the sample, do you link it to generalisability, internal validity or external validity? Internal because a good sample size means that any individual differences won't skew the findings, producing more accurate results, external because it involves linking the findings of the study to people outside the original sample, or generalisability - but my teacher said that a varied sample doesn't always mean high generalisability?

I'm a bit confused about this as my teacher didn't really explain this properly - does anyone have an idea?

Reply 1

username6164978

Original post by srb_16

It may be that she was describing a random sample (which is supposedly varied). A random sample may, by chance, not have high generalisability.

Let’s say you randomly select 20 people using a random number generator, and your study aims to measure running speeds (dependent variable)- I’m not naming the independent because it makes it more confusing. It is a repeated measures design . Everyone has an equal chance of selection. However, it turns out that everyone you have selected, just so happens to be really bad at running, and thus your dependent variable is skewed because of this. Your conclusion and results may also be different than if you took a larger sample with there being a higher chance of people having a greater variation in running speeds (However this is of course not certain, and you may end up with more people who are really bad at running).Overall, it may end up not being representative of the general population, where people have a greater range of running speeds. You have just so happened to ‘randomly’ select people who are all of similar running abilities.

Think of it like a dice, you can roll a dice 6 times and get 2 every single time, or get a mix. This is what a random sample is like, and is not always representative and thus generalisable, and nothing can help this other than taking larger and larger samples. Even then you cannot guarantee that it will be generalisable.

Hope this helps.

(edited 6 months ago)