paired vs independent, which one

juca

New Member
#1
Hello,

For my PhD, I am analysing the results of a survey with n=232.

The survey asked the participants about:
- their previous experience with an activity (yes/no)
- how many people did they know related to the activity (normalized to 0-100)
- their future interest in the activity after a measure is introduced (normalized 0-100)

I would like to analyse using bayes or t-tests if:
- the previous experience is related to the amount of people they knew related to the activity
- the previous experience predicts their future interest after the measure is introduced

I am not sure which test can I use to analyse these data. It is my understanding that this would be considered a paired sample, but I am a bit unsure. Some information mentions that a paired sample would need a second sample asking again the same individuals. I do not have such sample.
In my view I just have two groups, the one that had experience before, and the one that had not.
Any help is appreciated.
 

Miner

TS Contributor
#2
Paired (dependent) is where you are evaluating the same subjects before and after (repeated measures).
Independent is where you are measuring different groups of subjects.
 

hlsmith

Less is more. Stay pure. Stay poor.
#3
To clarify with examples:

Paired: compare people's weights before and after an intervention.
Independent: compare people's weights that got an intervention to people that did not get the intervention.
Independent: compare people's weights that got an intervention to people that did not get an intervention while controlling for prior weight (if intervention was not randomized).
Shitshow: compare some people's weights before and after an intervention but you don't know which people got the intervention.
 
Last edited:

Karabiner

TS Contributor
#4
No harm intented with regard to the OP, but:
Assuming that the OP is among the most gifted academics in his field
(otherwise he wouldn't have been offered to become a PhD, yes?), the
academic system's reasoning is not clear to me. A formulation such as
"I would like to analyse using Bayes or t-test" (something like "I would
like to analyse these texts using hermeneutics or word for windows"),
or the incapability to identify one's very own data structure and study
design, leave me quite confused. Why does the system expect PhD
students to execute statistical analyses, while they lack basic understanding
and knowledge? Either they should provide appropriate infrastructure
(statistics consultants etc.), or they should not expect statistical analyses
to be part of PhD theses, from people who need to consult an anonymous
web forum do get an idea what they are doing.

Again, no harm intented

Karabiner
 
Last edited: