totally lost...

#1
Hey,

I need some help. I'm a full on Arts-and-Humanities person, but I need some stats!


Here's the story:
1. Two groups of participants: Group 1 got the treatment, Group 2 didn't.
2. I've calculated the difference in their scores from the pre-test and the post-test (on 5 different tests). I've also been able to work out the means in an independent t-test.
3. There are 7 variables that I can think of at the minute that might also need to be included at some stage (I've coded them all).

This is what I need help with
  • Can someone advise how I identify the p-value?
  • And suggest how I go about identifying if there's any correlation between my variables and the scores? Do I just keep doing the independent t-test but change the va
riables each time?

Thanks!
 

Karabiner

TS Contributor
#2
Here's the story:
Far from complete, I'm afraid.

1. Two groups of participants: Group 1 got the treatment, Group 2 didn't.
What is the topic of your study? What is the research question?
Where do these group come from? Was allocation to groups
randomized? How large were the groups?

2. I've calculated the difference in their scores from the pre-test and the post-test (on 5 different tests).
Seemingly, the measurements were on an interval scale, then?

3. There are 7 variables that I can think of at the minute that might also need to be included at some stage.
What do you mean by this?

Can someone advise how I identify the p-value?
From the t-test mentioned above? Or what do you mean?

how I go about identifying if there's any correlation between my variables and the scores?
Each variable separately or all of them jointly? In associtation with
the variable "groups", or not?


With kind regards

K.
 
#3
Hi there,

1. My aim is to chart changes in ability in perceptual ability, sound/letter knowledge, phonological awareness, and word decoding ability in 4-5 year old children following a treatment phase for half of them.

2. The participants were taken from one class and were matched into two groups of 8 - sample size is 16. I know it's small, but I've been advised to look at the stats anyway.

3. If an interval scale is the equivalent of percentages, then yes.

4. Current variables that I've accounted for are
- age
- whether they had the treatment or not
- library membership,
etc


And all of these are to be
- perceptual ability
- sound/letter knowlege
- phonological awareness
- word decoding ability

5. Regarding the p value- I think I'm asking whether a t-test is best, or if there's something that can help me to demonstrate that there may be a significance in my figures.

6. I'm not sure at this stage how many of my variables I might group together.

Hope this explains it a bit more! Thanks so much
BT
 

Karabiner

TS Contributor
#4
With a sample size < 20 it is often recommended to use "nonparametric" (rank based) tests. In you case, a Mann-Whitney U-test could be used in order to compare change scores between groups.

With kind regards

K.