1
$\begingroup$

I have a question related to an experiment I conducted about giving behavior. I'm interested in how giving behavior can be manipulated by giving different types of information about charities and what type of information is most effective in doing so.

Every participant in the experiment is in the beginning randomly allocated to one of seven different conditions - one control condition + 6 conditions in which participants are influenced with different types of information about three charities. In the end they can decide which charity they want to donate money to. I want to compare those frequencies to significantly show that for example participants in one conditions preferred the 'Against Malaria Foundation' to the 'Books to Africa' charity. Attached to this post you can see the actual question, which is displayed to the participants at the end of each condition.enter image description here

Is there any statistical test, which can compare frequencies (whether people donated in one condition significantly more often to a particular charity compared to the control condition)?

I already did a Chi-square test (homogeneity of proportions), which was significant at the 1% level, showing that participants' donation decision clearly depended on the condtion they were grouped in. But is there some kind of test to compare a condition to the control group, showing f.e. that the Against Malaria Foundation was chosen significantly more often compared for example to Die Arche Kinderhelfswerk? I'm always struggeling with the fact that the variables 'condition' (which condition you are in) and 'choice' (donation choice) are both on a nominal scale.

$\endgroup$
  • 1
    $\begingroup$ This seems perfectly clear to me. $\endgroup$ – gung Mar 23 '18 at 13:32
0
$\begingroup$

I compared individually the means of the donations going to the Against Malaria Foundation in each condition with the donations in the control group going to the Against Malaria Foundation. The data is not normally distributed, so I used the Mann-Whitney-U Test because it is the non-parametric version of the two sample t-test since.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.