Dear Mie,

My suggestion would be to live the question open for respondents and then group the answers in 5-6 big categories,

Hope that it helps.

Dear Luis, Napoleon,

Thank you very much for your advice. I think I understand what you mean and I agree that I should produce the question without any impartial and inoffensive items. It is better to motivate participants to answer.

Can I ask like following items as 5-6 big categories. i.e. 

Caucasian

Latino/Hispanic

African

Caribbean

Asian

Mixed multiple ethnic group

Other__________________________________

Does this make sense to participants?

I myself would favour open ended questions that encourage the participation of your purposeful sample/ Go for simple easy to answer questions first just to get them talking and then go for bigger categories of questions further on in the study. Or you could do a focus group and bring a cross section of those you want to speak to into your group. Get a group conversation going that will open up more concepts and ideas. It may end up giving your study the depth needed to successfully answer your key questions.

Hi Mie, My first impression is that Native American/Indigenous is missing and so I have no category from which to choose.  If your research is being conducted in the U.S., the Census categories should suffice.  In your dissertation, you can validate your categories from the U.S. Census.  If you are developing your own instrument for a quant study, developing your own instrument will greatly dilute your validity and thus, your research findings. But you may only be gathering demographics for the participant pool.  I suggest using a tested instrument if you are conducting quant research.  Finally, allow your committee to guide you on this, as you are a novice researcher and they are your supportive research experts.  Jan Vinita White, PhD, gerontologist

Dear Georgina and Jan,

Thank you very much for your opinion and advice. My research is Internet survey about the perception of public towards human directed aggressive behaviour in dogs. It is for people who would access the questionnaire, all over the world. I would like to identify the cultural difference as well, so the ethnic groups are quite important items. As you say that I should think about  the Census categories as well.

I will reconsider it deeply what the best way is. Thank you again. 

In general, it appears that the large categories you have formulated would work well for participants from across the world.  I see three exceptions to this.  First, I agree with Jan Vinita White that a category labeled something like "Native/Indigenous" is needed for those who self-identify in that way.  Second, the category "Pacific Islander" should be added to the list.  Third, because the categories are broad, I would also suggest another open-ended item at the very end that says something like "Additional information about your race/ethnicity you would like to add:____________________"  In this way, individuals who want to could add more specific information.  For example, if a participant checked "African," but was born and raised in the U.S., that person might want to add "African American" to clarify.  Best of luck, Mie, with your research! 

Dear Richard,

Thank you very much for your advice.

It is really useful advice for me too. After I have had useful advice from kind people, I decided to use a broad categories and follow's Jan's advice. As you say, it will be better to add "Pacific Islander" and also open-ended item at the end. 

I will do my best to obtain right information for my research and lead it to a good research! 

Mie, it all depends on your purpose. If you're hoping to compare your race distributions to a population statistic (e.g., Census), and DEFINITELY if you're planning to do any kind of adjustment on race (e.g., post-stratification) you should use the questions exactly as Census (or your comparison survey) does. Words, categories, and all. On the other hand, there are a lot of good reasons not to use Census race and ethnicity questions. One is that they don't get you very detailed sub-ethnicity classifications (see the California Health Interview Survey for more detailed questions on ethnicity), and the other is that there's good evidence (wish I could cite off-hand) that a lot of respondents don't think of Hispanic as an ethnicity separate from race (although official statistics continue to ask it that way for official reporting reasons). 

Good luck with the dissertation!