Assignment:
This project is a short paper presenting your findings on a subject of your choosing. Find 2 recent studies or experiments that present statistical findings on an issue. You must clear your topic with me (email, Canvas message, or in person) by November 7th, and link me to one study that you'll be using. Any topic for which you can find several sets of statistics is acceptable; I encourage you to pick a topic that you're interested in, even if it's controversial in some way. The purpose of the project is to present a statistical analysis of an issue; your grade is in no way dependent on any personal opinions that you might express in your writeup.
Your topic and conclusions will be treated as confidential. The main goal of the project is to compare the methodology and results of the studies. Often, even when a phenomenon is commonly accepted, different studies produce wildly different findings about its frequency or severity. This is especially true for politicized issues; it's extremely easy to create and present misleading statistics to push a particular agenda. Most studies provide a confidence interval (or margin of error) for their findings, and it's quite common for two studies of the same population to present mutually exclusive findings. This is an enormous red flag that something is seriously wrong with one (or both) of the studies.
Final project topic is "The effect of playing sports or video games on youth hand-eye coordination."
The research component of this project will require you to access journals that are not free for public access. If you're signed in to USF's campus WiFi, you should be able to get into most of these journals for free. Please reserve a few hours in the next several weeks for on-campus research.
The topic you choose will greatly affect how your final project looks, but you should follow some general guidelines:
• Identify the variable in question - what property or characteristic is being investigated?
• Identify the population (plants being given fertilizers, homes in a city, people in an age bracket...). The population is usually not "all U.S. adults."
• What are some confounding factors that could skew the results of the studies or experiments? How do the studies control for these factors? Does one study do an especially good (or bad) job at this? This is one of the most important things to consider when evaluating a study's reliability.
• Are the sample sizes used sufficient to make a conclusion? Is there evidence that the samples are selected sufficiently randomly?
• What kind of statistical tests do the studies use to form their conclusion (regression, 1-population hypothesis testing, etc). What p-value is used? We'll cover these concepts in the next few weeks. You can search for the word ‘test' in your articles to quickly identify what statistical tests the researchers used.
• What groups carried out the studies? What groups funded the studies? This kind of disclosure is usually presented at or near the bottom of the first page of the study, and omitted from press releases and news articles on the study. A study may be more reliable if funded by a neutral party (perhaps the NSF), and completely worthless is funded by the industry being studied.
• Most importantly, do the results in the studies agree or disagree? If the results contradict each other, can you tell why? Maybe one study was based on a much smaller sample, and is less likely to be accurate, or the underlying populations are different. This last point will vary greatly depending on what you find in your research.
If you start researching a topic, you will probably first find news stories about studies. These almost always link to the actual document written by the researchers, which may be paywalled but available for free elsewhere if you search for the paper's title or the names of the PIs. Many media outlets do a terrible job reporting on statistics, exaggerating or distorting findings to get you to click on their article. You'll need to look at the paper being referenced for all important details.
Your final project should be a paper that is long enough to present and compare the findings of the 2 studies (at least several pages). The bulleted list above is not a concrete list of requirements - different topics will have different needs. It should be formatted as a research paper using a reasonable citation style (MLA or APA, for example), and should cite as sources the studies that you use and any news articles about them that you reference. You will not be explicitly graded on how perfectly you adhere to a citation style, but I'll deduct points if your paper is extremely sloppy or your references are hard to follow.
You are welcome to submit a partial or complete draft of the paper for feedback as many times as you'd like until November 30th, after which no more rough drafts will be accepted. I suggest that you come to office hours at least once during the project to get feedback and suggestions; it might save you a lot of time.
Your grade will be determined by the following 3 factors:
1. Mathematical correctness: do you correctly use the appropriate statistical terms to communicate and compare the studies' findings?
2. Completeness: do you provide a satisfactory overview of the studies you selected? Is it clear from reading your paper if a study is deliberately biased, or unreliable in some way? If two studies present different results, do you point this out and present a possible reason why?
3. Writing quality: is your paper reasonably well-written? Is it easy to follow?