As there are a few statistical experiments that I'm interested in carrying out this summer, I've decided to create this page to show the results of these experiments as soon as I finish them. Even though I find these experiments to be somewhat interesting, I'm sure that most people will think that they're either useless or not worth doing. Before anyone jumps to conclusions and accuses me of "not having a life," these experiments, along with mostly everything else on this website, are done during the little free time that I have.
If a statistician happens to stumble across this page, I'd be very grateful if he/she is willing to offer some assistance and verify that my calculations are accurate. You can be reasonably certain that they're correct because I did score above 100% in my statistics class, but I'm still just a student and it doesn't mean that I know the answers to everything. I also haven't taken a math course in a while, and since my knowledge is rusty, I'm likely to make a few mistakes here and there.
The Experiments
Experiment #1: Asians and their popularity on xanga.
Null hypothesis: # of Non-Asian Users = # of Asian Users (The proportion of Non-Asian users is equal to the proportion of Asian users on xanga; there is no significant difference between the proportion of Asian and non-Asian users on xanga.)
My hypothesis: # of Non-Asian Users > # of Asian Users (More than half of xanga users are not Asian; there are more Non-Asian users on xanga than Asian users.)
Information
Many people are probably familiar with the stereotype that "most xanga users are Asian," and this reference is thrown around by so many people that I'm beginning to wonder if Asian users truly make up the majority of users on xanga. Is there any validity behind this statement, or are people just assuming that most xanga users are Asian based on what other people are saying? By the term "majority," I'm implying that Asian users constitute at least 50.1% of the population of users, which seems a tad too large when you think about it.
Could it be possible that more than 50% of xanga users are Asian? Out of the 40 million users on xanga, at least 20 million of them would have to be Asian, which is nearly twice as large as the current population of Asians in the United States. If every single Asian person in the United States had his or her own xanga site, there would still be a few million other xanga sites that would have to be taken into account. Perhaps these few million accounts can come from Asian users from countries besides the United States, which seems to be a fairly reasonable estimate. Now, there are always people who have more than one xanga site, but considering that not every single Asian person in the United States has a xanga site, it's more than balanced out in the end.
There is obviously a sizeable population of Asian users on xanga, but by using logic, I disagree that they are the largest group of users on xanga. My alternative hypothesis states that there are more Non-Asian users on xanga than Asian users, which also means that Asians do not make up the majority of users on this site if my alternative hypothesis proves to be true.
Sampling Method
How will the experiment be carried out? Since this experiment has to be unbiased in order to be valid, I'll obviously have to avoid selecting a group of users from particular areas of the site where my information isn't going to be generated in a random manner. For instance, I can't look at blogrings to do my sampling because that's likely to produce biased results. After all, not everyone is a member of the same blogring, and there are many blogrings on xanga that are mainly comprised of one specific group of people. If I were to retrieve samples from blogrings such as Asian Diaspora or New York City Asians, for instance, it's pretty obvious that all of my samples are going to be Asian. On the other hand, if I decide to do my sampling by looking at the top 5 most popular blogrings, very few of them will be populated with Asian users.
Surveying random people and asking them about their ethnic background is another option, but biased results will also occur. Since they may lie about their ethnicity on purpose to throw off the results, it's better if they don't know what the purpose of the experiment is about. That, and I highly doubt that people would willingly participate in the experiment. Another option is to simply take a look at the majority of xanga users who leave comments for various "xanga celebrities," but even this would produce biased results because it simply isn't representative of the rest of the population. Perhaps this is how people get the idea that "most xanga users are Asian," as there are quite a few "xanga celebrities" who attract predominantly Asian readers.
Although all of the above methods of gathering information will produce biased results, I know of one possible way that can produce unbiased results to make this experiment valid. I can gather my information from xanga's Newly Updated Sites page, which randomly generates 50 of the most recent updated blog sites on xanga.
Additional Notes
This experiment is partly inspired by a study that was carried out by a student at Stanford University two years ago. For his final project, he had to survey random people and come up with a conclusion as to whether or not there was a correlation between bloggers of a specific ethnic background and their preferred blogging service. He concluded that while a majority of Livejournal users were White, a majority of xanga users were Asian. I don't remember any other details about his study and I'm not sure if his methods of gathering samples were biased.
Result
I haven't actually carried out the experiment yet, but if you're really interested in knowing the answer, just keep badgering me to do it every week and I'll get to it eventually. With that said, I'm a little surprised that no one has actually tried to carry out this experiment themselves instead of just assuming that the statement is true. If I happen to have more spare time this summer than expected, I may conduct this experiment in more than one way.
