I am going to collect authentic data for my statistics class so the sampling does not have to be a rigorous as if I’m writing a research paper. The question is “What is the biggest barrier to equity in the United States.” Each group has 50 members and is broken down by culture/race.
Anglo, African, Asian, Latinx. Middle Eastern.
What are some ideas on how to gather my data.
Oof. For a statistics class? What kind of data are you looking for? Income and other dollar amounts? Socioeconomic stuff about parents and childhood environment? Years of schooling?
You might be able to get some data from the US census.
No. Going to a person, recording their culture and their answer to the question, “What is the biggest barrier to equity in the United States.”
First, define “equity”. My initial thought was that it was about investments. I assume you actually mean equality. Or am I misunderstanding?
You should say “Social Equity” then.
Diversity, equity, and inclusion is a catch phrase now. I understood what he meant by equity.
It’s an interesting illustration of inadvertent social bias that for some people, “equity” means something only the investor class would be aware of.
For the overwhelming majority of the human race, “investing” is making sure you can make the next rent payment or secure your next meal.
I did too. I see that @bob_2 is English–wonder if this is one of those “cultures divided by a common language” things where the most common US meaning of “equity” is simply not used in the UK.
Ok but let’s get back to the OP because the highjack on “what is equity” is not the question I asked.
How can I find 50 people from each group to answer whatever question I ask?
If it makes it easier to stop hijacking the thread. assume I need to ask 50 members from each group: Anglo, African, Asian, Latinx. Middle Eastern. “What’s your favorite ice cream flavor.”
First, define “ice cream”.
Sorry, couldn’t resist.
I think it matters what you expect your explanatory variables to be. The fact that you are stratifying by ethnicity suggests that you are expecting ethnic group to be a determining factor of the response to the question. If there are other explanatory variables, for example income, gender, or education, that will complicate things.
You are basically interested in a face to face poll or a simple written survey. It should ideally be administered under a constant set of conditions to avoid other biases. Given the relatively large number of people you will need to reach, I would suggest a short written survey. Perhaps you can target a large public event? If you really need to end up with at least 50 respondents for each ethnic group, that will elevate the group size a lot as you will be limited by the least represented group. So it will end up being more than just 5 x 50 people.
And please be careful about aggressively chasing down people of a particular ethnic group. That might not go over well 
Finally, you will also need to think about the format of your question and how you intend to analyse the results. Multiple choice options will be a lot easier to analyse than open-ended responses, but will give less insight.
Oh, one additional thought. You mentioned that this is for a statistics class.
Which statistics techniques/skills is your teacher expecting you to demonstrate? I would imagine a stats class being interested in tests of significance etc. For that your question really should ideally produce a quantitative response for which a mean and standard deviation can be calculated. An open-ended question is OK for social research, but doesn’t lend itself well to usual statistical analysis. Multiple choice questions could be OK, but the responses should fall along a continuous spectrum, which would be hard in your case. Have you vetted your research question with your teacher?
Way way way too complicated for what I’m doing. This is not research project. Don’t care about other variables, etc. This is for a high school class to analyze. The survey is
Go to person. What race are you? “Middle Eastern” What your favorite ice cream flavor?" Strawberry.
Go to person. What race are you? “Anglo” What your favorite ice cream flavor?" Rum Raisin.
And so on until I get 50 people from each group. Doesn’t even have to be in person. Could be a Google Form.
So please stop with the advice on methodology or definitions or whatnot. I know you are all trying to help but if this were a publishable study I know how to do all that. You are not helping so please … I’m begging can you all just stick to the question I asked and stop STOP STOP with the methodology.
How can I conduct the survey to get responses from 50 people in each group?
That’s it
That’s all I’m asking
Please answer that question
Just that question
Do not make up a new question to answer
Just stick to that question
IT DOESN’T MATTER! MY QUESTION IS HOW TO ADMINISTER THE QUESTION TO GET 50 RESPONSES FROM EACH GROUP!
How can I ask my FQ any clearer so it gets answered and not all this crap I don’t care about?
I get my OP may have been ambiguous but later I posted this
Can we please answer that question
There are a few ways. Create a free Qualtrics account and create a survey that includes questions on race, ethnicity, and “ice cream.” Distribute links to the survey in diverse areas. You probably will need to explain you are asking this as part of a high school class. You could also use Google Forms or similar.
I recommend you make the survey extremely brief, as people are just doing it as a favor, they probably won’t put up with more than a few questions.
You could use Amazon’s Mechanical Turk and pay for responses. I don’t think you can pre-filter by race, but I think you can by location. So you should be able to distribute to multiple countries to increase the odds of getting the groups you want. Unequal group sizes are an important concept in statistics!
Simulate the data.
OK. Maybe out to my fellow teachers then we could all share the data
Literally the two questions Ethnicity and Flavor
I promise my students I use authentic data whenever possible.
First - is the statistic meaningful if you specificaly target certain groups by certain factors?
Classic statistics issues:
Dewey defeats Truman happened because the polls were done by telephone when many rural and lower income people did not have telephones.
I think it was Trump polled extremely high among black voters in California in 2016, apparently because the sample of voters was small, the sample of Republican black voters smaller still, and one dedicated black Trump supporter made the poll for black support look much higher than it was. (Exacerbated by polling the same people over time).
The debate about “equity” is not irrelevant. Many people have heard the word, like me, and are aware of its dual applications. But I had to read one of the links to understand exactly what the difference was between equity and equality. So the problem is, even if you find enough “test subjects” is their answer meaningful? Or are they describing barriers to equality? Or something else? Plus, answers will vary depnding on other factors - educated, employed blacks or native Americans will have a different point of view than poorer ones. Poor white people may have different views, some offensive. People in more integrated areas will have a different viewpoint than say, in Chinatown or other less diverse areas. Can you tell by sight who is Turkish? Are they considered “middle eastern”? What about Uzbekis? Are Nubians black or middle eastern? ("what’s a “Nubian”?) Also, I don’t see a distiction between south Asian and East Asian, two distinct ethnic classes.
So, how can you get 50 people of each sample? Good luck. Keep in mind if you stake out a geographical location, the bias is toward that location. Poor area? Business district? Different answers at Coney Island and Wall Street, even from the same ethnic groups. The ideal way is (used to be) picking names at random from a phone book… Back when people answered unkown callers and there were such things as phone books. Online is biased toward young and educated (Says this old fart) , depending on the forum and you have no idea how meanigful the group or answer is.
Finally, some people may be offended if you try to ask their ethnic group. Are you instead going to guess their group? Or correct it based on visual cues if they are difficult? What about mixed? Keep in mind, Tiger Woods for example is the most successful Asian-American golfer. Plus, some people may be offended that you seem to be focussing on specific ethnic groups, so you can’t be obvious standing on a corner and asking only, say, apparently black people.
You could stake out an assortment of shopping malls covering multiple income/class strata or negibourhoods - but would need permission from the mall owners. So, street corners may be the best choice. Change locations frequently to ensure no bias by neighbourhood or income strata.
Note I’m assuming you live near an area where there are in fact an assortment of ethnic types so you can make your quota.
That is a good thought, as hopefully it will limit the amount of explanation and convincing required.
Getting diverse samples is extremely difficult due to lots of reasons, ranging from language barriers to active distrust of researchers from the target population. I know Qualtrics makes it fairly easy to create alternative surveys in different languages, so it might be worth the effort of doing translations to reach non-English speakers.
If you have contacts with teachers at schools with large minority populations, they might be good places to start.
Also, as your collection scheme gets more complex, the chances of having it come to the attention of someone at your school/district who is upset by it increases. I know in higher-ed doing something like this may have to go through IRB and human subject approval, though it could probably be exempted if no personally identifiable information is requested, but in that case getting the exemption is the important step.
Another source of data is to look for free datasets. Many government and census type stuff is available for free. I don’t have any examples in mind that get at your specific question, but it might be possible to find something that is useful.
This only matters if you are trying to actually answer questions about equity. If you are trying to create a statistical lesson based on real data and a topic the students are interested in, then the validity of the questions might not matter too much—it is just a really complicated and haphazard simulation, not a graduate thesis.
There’s a subreddit
https://www.reddit.com/r/SampleSize/
where people volunteer to provide survey data.