# Help choosing a statistical method for data analysis

#### Devinharris09

##### New Member
Hi everyone! I am working on a hypothetical research proposal for a sensation and perception class and I need help with the data analysis section of the paper. My hypothesis it that people who have been blind for longer than five years have better tactile acuity than those who have been blind for less than five years. In order to measure tactile acuity I am utilizing two methods: Braille charts and JVP domes. The 40 participants will be separated into two groups based upon how long they have been blind, and all participants will complete both methods for measuring tactile acuity. What I need help with is figuring out specifically which statistical method I would use in this hypothetical study in order to assess my hypothesis and back it up with numbers. Paired t-tests? ANOVA? I have no idea! Any help is greatly, greatly appreciated!

#### Miner

##### TS Contributor
As described, a 2-sample t-test may be appropriate provided the assumptions have been met. These do not appear to be dependent samples, so the pair t is not appropriate. If you change the study slightly to use the actual time instead of the < 5, > 5 categories, you could use a regression analysis. This would maximize the information content of your data.

#### Karabiner

##### TS Contributor
What I need help with is figuring out specifically which statistical method I would use in this hypothetical study in order to assess my hypothesis and back it up with numbers.
You'd need to control somehow for additional characteristics
which differ between shortly blind and longer blind and
might affect outcome (e.g.: age; age at onset of blindness,
whatever). So, an analysis which includes such additonal
characteristics should be an option, or matching
of participants. Moreover, putting people who have been blind for
1 week or 4.9 years into the same group, or putting
people who have been blind for 5 years or 60 years into the
same group, sounds not convincing. As Miner said, rather
use [multiple] regression analysis with the orginal duration of
blindness.

With kind regards

K.

#### Miner

##### TS Contributor
Agreed. This would be more difficult to do with a t-test, easier with regression.