# Thread: Is this approach permissible?

1. ## Re: Is this approach permissible?

Originally Posted by Dason
Haha, no not really. Thanks for notifying

2. ## Re: Is this approach permissible?

All of a sudden, it's just greek to me

Sorry guys to bother you again. I read in my statistiscs book that I need more than one independent continuous variable to create a regression model.

I mean, I receive one continuous variable (Index of Social Media) and have one categorial variable. Is this not perfect for a independent t-test?

3. ## Re: Is this approach permissible?

I would like to investigate, whether successfully funded initiatives use more social media.
So "sucessfully funded" is your grouping variable
and number of channels used is your dependent
variable. A beginner's choice could be using the
U-test, then.
I read in my statistiscs book that I need more than one independent continuous variable to create a regression model.
You need more than 1 independent variable to create a MULTIPLE
regression model. For a SIMPLE regression model you just need
1 IV.

4. ## The Following User Says Thank You to Karabiner For This Useful Post:

julia89 (07-17-2012)

5. ## Re: Is this approach permissible?

Originally Posted by Karabiner
So "sucessfully funded" is your grouping variable
and number of channels used is your dependent
variable. A beginner's choice could be using the
U-test, then.
Well, my statistiscs book suggests a T-test for that...

Quoting:

Two variables:
• one categorical, independent variable (e.g.males/females);
• one continuous, dependent variable (e.g. self-esteemscores).

Am I wrong now?!

6. ## Re: Is this approach permissible?

The issue is that it makes more sense to think of your social media index as the independent variable and the binary "Funded" as the dependent variable.

7. ## Re: Is this approach permissible?

Am I wrong now?!
I do not know how large your samples are and/or whether assumptions
of the t-test are fulfilled. Did you already check assumptions? U-test is
a robust alternative if you should encounter problems with the t-test.

8. ## Re: Is this approach permissible?

Originally Posted by Karabiner
I do not know how large your samples are and/or whether assumptions
of the t-test are fulfilled. Did you already check assumptions? U-test is
a robust alternative if you should encounter problems with the t-test.

Well, I collected some cf initiatives that fall under the category "music" on Indiegogo. Is this sample random?! I mean a took a sample that is from the overall crowdfunding pool "population"

I checked the normality assumption and skewness and kurtosis are under 2, which my book says is good

Thanks