# Thread: Measuring effect - urgent

1. ## Measuring effect - urgent

Hi Guys,

For my thesis, I am researching the influence of colour (IV) on product attractiveness (DV). Respondents in my experiment choose one of 7 products they like most. 129 people participated in the experiment.

Outcome per product in frequency:
A - 2
B - 10
C - 1
D - 6
E - 69
F - 6
G – 35

Now is there a way to measure if colour (IV) has a significant effect on product attractiveness (DV) for this specific example?

2. ## Re: Measuring effect - urgent

where is the IV variable !

i think you d drive a chi square test of association but needs IV variable

gd luck !

3. ## Re: Measuring effect - urgent

That IV is colour, which are the different options: A, B, C, D, E, F, G
Can you make something out of that?

4. ## Re: Measuring effect - urgent

Can you rule out other features or attributes of the products as influencing selection? Are there any characteristics of the respondent you would want to control for (e.g., age, sex, education, etc.)?

5. ## Re: Measuring effect - urgent

Gender / Nationality etc. has all been taking care off. And I know the tools to analyze that.

The issue I currently have is how to measure the core effect, as described above.

6. ## Re: Measuring effect - urgent

hi,
obviously E was preferred by far the most participants. Now, as hlsmith asked, how certain are you that this preference is due to the color and not some other charactetistic of the product? If other characteristics are confounded with the color then obviously the data will not allow you to draw any conclusions about the influence of the color, so this would be the first thing to ask.

E.g. if you look at chewing gum - color might be confounded with the flavor. So, a preference of the red colored product might show in reality a preference of the strawberry flavor..
regards

7. ## Re: Measuring effect - urgent

Ignoring all of the previous concerns, we assume the null hypothesis is color and selection are independent. So each product has an expected uniform probability of being selected, probability = 5.4%. I believe you can then address the problem using a binomial test.

Side note, how do you control for gender, nationality,... weighting?

8. ## Re: Measuring effect - urgent

hi,
the probabilities should sum to 1 I guess so p=0,14. It looks more like a multinomial to me - maybe a permutation test would work?

9. ## Re: Measuring effect - urgent

yes, 14.3%, I was looking at the wrong line on my calculator

 Tweet

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts