Reliability and Validity - PLEASE HELP!

[DylanBurrowes]

So lost, could someone please explain with examples the different types of reliability and validity? I just don't get it... Thanks!

[HiddenUser]

Validity = It measures what it should be measuring "It is a valid measure"
Reliability = It does this consistently "It is a reliable measure"

[emilyhobbes]

There's only a couple listed on the study design, so it's not too bad
So, reliability is how consistent the research and data is, so like, if you do the same experiment twice, you'd want to get pretty similar results. The type of reliability listed on the study design is internal consistency which is basically that within the measurement instrument, separate sections get pretty similar results as well, for example, odd numbered items get similar results to the even numbered items

Validity is how accurately the research tool is testing what it claims to measure. So, like, if a scale measures me as weighing 3kg less than I actually do, but it does this consistently, then it's not valid (it's measuring inaccurately) but it is reliable (it says the same thing each time)

The two types of validity listed on the study design are construct validity and external validity
Construct validity is if the content of the research tool relates to the theory or concept the tool's meant to be testing.
External validity refers to the extent that the findings can be generalised to a wider (external) population

hope that makes sense!

[DylanBurrowes]

The type of reliability listed on the study design is internal consistency which is basically that within the measurement instrument, separate sections get pretty similar results as well, for example, odd numbered items get similar results to the even numbered items.

Do you have an example? I just don't get the whole odds and evens thing, the rest was awesome though, thanks!

[emilyhobbes]

Hmm, let's see..
Okay, so say you had a 50 question math test yeah?
Out of the 25 even numbered questions, you get 22/25
Out of the 25 odd numbered questions, you get 23/25
That's internal consistency, like, half of the questions weren't ridiculously hard and the other half weren't really easy, both halves produced pretty similar results (reliable)

 

[DylanBurrowes]

Ahhhh okay, so would the consistency be through comparing the 22/25 to the 23/25 or by seeing people get 22/25 and 23/25 each time?