Help!! (Generalising data)

[radl223]

Hello,

I have my ERA sac next week and i was wondering that judging from the results and data, how do I know if it can be generalised to the population? Is it if the hypothesis is supported?

Also, when it says 'can it be generalised?', which group does it refer to? your target population or the public in general?

[brenden]

Can the sample data be attributed to the research population?

If the sample is stratified and properly representative and there were no confounding variables, then you could generalise the data to the research population.

I never had your last question answered for me, I've only ever assumed only the research population. My logical justification of this is - if we wanted to test the influence of punishment on behaviour modification in children between the ages of six and eight years of age, then we wouldn't generalise the sample's data to the entire population because that would just be plain silly. It follows, then, that we apply sample data to the research population (if we agree that it can be generalised).

Does that sort your conundrum?

[radl223]

Yes it does to a degree! So regardless of whether the results support the hypothesis etc... If it is full of confounding variables (and the fact that i used convenience sampling) it cannot be generalised to the target population?

Sorry for so many questions and thankyou so much!

[brenden]

Yes it does to a degree! So regardless of whether the results support the hypothesis etc... If it is full of confounding variables, it cannot be generalised to the target population? Also, just making sure, can we ever conclude that it is 'statistically significant' (despite no way of calculating it) in our own ERA sacs simply because the hypothesis is supported?

Sorry for so many questions and thankyou so much!

Well, if it's full of confounding variables, it's a rubbish study, so you wouldn't even ask that question. (Note well the difference between extraneous and confounding variables) Pretty much the things you should be looking at for things to be generalised is the sample size (if it's way too small, you can't generalise) and if it's stratified/representative.

If they don't present you with a p-value, you cannot infer statistical significance from descriptive data. be careful. You can say that it's statistically significant, just... don't fuck up. I mean, if you're testing reaction times of X People and Y people, and X people react 10 seconds faster than Y people, pretty safe to say that there's a statistical significant.

[radl223]

Okay thankyou so much! Helped me a lot 

[brenden]

You're welcome Good luck on your ERA

[Russ]

If they don't present you with a p-value, be careful. You can say that it's statistically significant, just... don't fuck up. I mean, if you're testing reaction times of X People and Y people, and X people react 10 seconds faster than Y people, pretty safe to say that there's a statistical significant.

You can?

[brenden]

I believe VCAA would accept that, however, I will verify this in a moment.


Edit: I was wrong. I got marks in a SAC once for saying descriptive stats were significant and my teacher has been an assessor for more than a decade. (I often suspected she marked SACs over-quickly). Thank you for pointing out my mistake, Russ.