Basic question about pre and post measures

#1
I am new to stats and am getting confused, please can someone tell me what it the best approach to analysing two separate groups (intervention and control) at two time points (very small sample size) any help appreciated!
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
Was intervention group assignment randomized? If not - things get trick. If so, you can model post = pre + treatment_group. Controlling for pre can pick up any residual confounding that slipped through the randomizing cracks.
 
#3
Was intervention group assignment randomized? If not - things get trick. If so, you can model post = pre + treatment_group. Controlling for pre can pick up any residual confounding that slipped through the randomizing cracks.
Thank you, they were not randomised but matched as close as poss, but there were small baseline differences?
 

Karabiner

TS Contributor
#4
How large are your sample sizes? What is your dependent variable, and how was it measured? What kind of intervention is it?

With kind regards

Karabiner
 
#6
How large are your sample sizes? What is your dependent variable, and how was it measured? What kind of intervention is it?

With kind regards

Karabiner
Thank you. sample sizes at present are 5 intervention and 4 control (all I have for my thesis - more time 2 data to collect for future write-ups)
Dependent variable is resilience measured using validated Connor Davidson CD_RISC scale. It is an individualised resilience intervention
 

Karabiner

TS Contributor
#8
So the most easy way would be to calculate the individual differences
between pre- and post-scores, and compare them between groups using
Mann-Whitney U-test (because of the small sample size, I am not sure
whether t-test can be recommended).

With kind regards

Karabiner
 
#9
So the most easy way would be to calculate the individual differences
between pre- and post-scores, and compare them between groups using
Mann-Whitney U-test (because of the small sample size, I am not sure
whether t-test can be recommended).

With kind regards

Karabiner
Thank you Karabiner. I will look this up in my SPSS survival manual. Your help is really appreciated, Rowan
 

hlsmith

Less is more. Stay pure. Stay poor.
#10
Pre-value meant pre-score or pre-measure - if not balanced or addressed; baseline differences can be erroneously attributed to intervention status!
 
#13
We're they matched based on Pre-value?
They were not matched based on their resilience scores. Is that the Pre-value? They were matched based on their age, gender, the severity of the offence they had committed and ethnicity. I've just looked at their baseline resilience scores and apart from one they are fairly similar?