Which statistical test

I have two data sets: they pertain to the same thing except one is before and the other after an intervention

I am seeing what percentage of patients with a shoulder dislocation where seen within 1hr after admission. Before my intervention 4/56 were seen within 1hr. After it, 3/10 (a different sample of patients all together) were. Which statistical test can I use to see if there is a significant improvement?
NOTE the groups of 100 are NOT related

Can I use chi square. If not what
(or should I use the Fisher Exact as the second population size is very small)
Thanks

(edited 6 years ago)

Reply 1

ladamesansmerci

9

I think you'd use a paired t-test as you are comparing linked data sets.

Please wait for someone else to confirm this though as it's been a good few years since I've done any statistics.

Reply 2

username459260

OP

16

Original post by ladamesansmerci

I think you'd use a paired t-test as you are comparing linked data sets.

Please wait for someone else to confirm this though as it's been a good few years since I've done any statistics.

So the data sets are not linked
I tried the paired t-test but the results for each patient is binary: Yes or No (0 or 1)

Reply 3

Gregorius

14

Original post by jsmith6131

I have two data sets: they pertain to the same thing except one is before and the other after an intervention

I am seeing what percentage of patients with a shoulder dislocation where seen within 1hr after admission. Before my intervention 4/56 were seen within 1hr. After it, 3/10 (a different sample of patients all together) were. Which statistical test can I use to see if there is a significant improvement?
NOTE the groups of 100 are NOT related

Can I use chi square. If not what
(or should I use the Fisher Exact as the second population size is very small)
Thanks

You want a test for the "difference of proportions". In "R" you can do this with the function "prop.test".