For your term paper, you will need to determine which of the following topics you would like to investigate further:

1. Passwords vs. Passkeys

2. Multi-Channel Social Engineering Techniques

3. User Awareness and Ethical Decision-Making on the Use of AI and Language Models

Each of these topics are broad; your task is to determine which angle researchers haven't yet explored in-depth and help supplement that gap in the literature. As you begin the research process, your central research question may evolve, and more follow-up questions may arise. Be sure to write down any questions you may have as you go - they could provide a great basis for your 4 required research questions!

In order to better understand what unanswered questions there are about your topic, we can perform background research. There are a few different places that will be useful for gathering basic information. Write down anything you notice about the existing literature: are there angles that haven't been explored, or questions that arise as you search? These can provide the basis for your research questions, which may evolve as you learn more about your topic.

Your assignment requires 4 total research questions. This means you should focus on one central research question, and provide three more follow up questions that your research will address. Your question should be specific, focused, open to multiple possibilities, and answerable.

For instance, take these three research questions:

Too broad: “What is the impact of social media on young people?”
What kind of social media platforms? What age group? What kind of effect is being studied (grades, social life, emotions, etc.)? Too many possibilities make this question unanswerable.

Too specific: “What is the impact of using TikTok for 1 hour and 6 minutes on 16-year-olds named Rachel?”
With a question this narrow, very few people fit the criteria. It limits participation and follow-up research questions.

Just right: “How does daily Instagram use affect stress levels among undergraduate students at WMU?
Focused on one social media platform and one group (undergrad WMU students). Simultaneously, it is broad enough to allow multiple follow-up questions (e.g., differences by gender, time spent on the app, comparison to other platforms, etc.).

For another perspective, check out this video from the University of Guelph McLaughlin Library, licensed under CC BY-NC-SA 4.0.

Four Steps to Narrow Your Research Topic

A hypothesis is your educated prediction of the answer to your research question. It can't be "I don't know" or a random answer; it's a well-thought out guess based on what the literature on your topic leads you to believe. Your hypothesis must also be testable, since your research (in this case, the survey) will aim to either support or refute it.

For more information on crafting a strong hypothesis, watch "6 Steps to Formulate a Strong Hypothesis," courtesy of Scribbr.

Determining the Delivery Method

Surveys can be conducted in a variety of ways: in person, online, over the phone, or mailed. For your survey, you should determine how you are going to administer the survey to your participants, as this will be part of the methodology section of the research paper. When choosing a delivery method, think about who your participants are and how to reach them most effectively. 

In Person
Pros: Great for building rapport with the participants. 
Cons: Takes more time and effort to be on site; also must determine a good place that students visit, like the Student Center, Waldo Library, the Parkview Campus hallways or the cafe, etc.

Online
Pros: Participants can answer when they like, without having to interface with other people asking the survey questions. 
Cons: Technology can fail - links break, WiFi slows down or fizzles out, and people may have difficulty navigating the interface.

Conducting a Survey: A Quick Overview

For a quick recap of conducting surveys, check out this step-by-step video from Ayesha Mushtaq of Educational Hub.

Writing Solid Survey Questions

Coming up with questions may come easily to you, but it's easy to miss some of the subtle ways to improve survey questions to ensure the participants understand exactly what information you're trying to capture. Above all, be sure to respect your participants' privacy - only ask the identifying information you need.

As you're writing your survey questions, remember to Relax, Be Cool.

R: Relevant.

• Each of the survey questions (you need at least 20) should directly relate to one of your research questions. You can make a table to establish which research question they help answer, or simply write it down next to the question for your first draft of the survey questions. If the survey question doesn't have a corresponding research question, consider omitting or reframing the survey question.

• Example: Your research question is about the impact of secondhand smoke exposure on adults 30-39 with asthma... but one of your survey questions ask if they have ever visited Canada.

B: Balanced. 

• There are a variety of survey question types and responses. You may be familiar with the Likert scale - every time you've rated something 1-5 stars, or said you strongly agree or disagree with something, you've answered a question that employed the Likert scale. You've also encountered multiple choice questions on tests and quizzes, and open-ended questions where you had to write a sentence or essay. Having a mix of survey responses will capture both the quantitative (numerical, objective) data from participants, as well as the qualitative data (free form, subjective). This will help paint a fuller picture of your research.

• Example: Your research question is investigating whether students who have visited at least 5 other U.S. states have higher grades than those that have never left Michigan. Your survey questions might include asking how many states they recall visiting (0, 1, 2, 3, or ranges like 0, 1-2, 3-4, etc.), the names of the states (open-ended), how long they stayed (date range), or the purpose of their visit (multiple choice). You may hypothesize that students who visited 5 or more states did better, but you found more nuance due to the different kinds of questions: they did better, unless the purpose of their visits were due to emergency or family circumstances.

C: Clear. 

• Ideally, the questions you are asking will only have one interpretation. This is one of the most difficult parts of writing survey questions, because we often assume what makes sense in our minds will make sense to others'. Providing definitions or more guidelines will help ensure your survey participants will understand exactly what you mean.

• Example of a vague question: How often do you use social media? (Multiple choice: A) All the time, B) A lot, C) Every once in a while, D) Not a lot

• Example of a clear question: In the last week, how many days have you used Instagram? A) 0 days, B) 1-2 days, C) 3-5 days, D) 6-7 days

• The clearer question provides a time frame (the last week), the unit of measurement (days), and the specific social media platform (Instagram).

Remember: your survey questions are what will give you the data to test your hypothesis - so make sure they're well-written!

Language Tips

Make sure to frame questions neutrally if possible (without outdated or gendered language). For instance, instead of "Congressman," use Member of Congress, Congressperson, Congressional Representative, etc.

Further Resources

For more information about how to craft excellent survey questions, check out Writing Survey Questions from the Pew Research Center, one of the foremost research groups around.

You've written at least 20 great survey questions - congratulations! Now, it's time to put those questions to work.

Many researchers will do something called a pilot survey, where they use a similar sample demographic to see if their questions make sense to respondents. It's also a great way of ensuring the survey isn't too long (leading participants to drop out before finishing), offending participants, or otherwise not working. If you choose to pilot your survey, ensure the demographic of participants you're using is similar; for instance, if you are researching senior citizens in assisted living facilities, you wouldn't ask your sibling to fill out the survey for you.

Once you’ve piloted (or revised) your questions, you’re ready to administer the full survey and begin gathering the data that will inform your results section.

First question: Do you have enough responses? Remember, you must survey at least 40 respondents for this assignment. If you've met that number, great! If not, resist the urge to fudge the numbers, and reach out to your instructor for guidance.

What you do with the survey data will depend on the kind of responses you asked for. For instance, if your responses include...

Numbers: Calculate the mean of this number. If your data is skewed one way (or if you have a couple of outliers), use the median instead of (or along with) the mean.

Multiple choice or categorical answers: Calculate the percentage of participants who selected each option. If it totals more or less than 100%, there's a problem!

Ranked answers (like the Likert scale): These are a bit tricky - Likert scales capture not just agreement or disagreement, but the intensity of that opinion. For this reason, report the median or mode, and consider noting the spread of responses (for instance, how many answered "always" vs. "never"). 

Open-ended answers: These can also be tricky, because it is more time consuming to read responses, and there is room for interpretation error when summarizing. Begin by pulling out short phrases or key words that the respondents have shared. When summarizing, look for common themes across many responses rather than focusing only on one or two outliers that support your expectations. For instance: "My favorite part of the survey was the fact that it wasn't too short and wasn't too long, and that I could eat my lunch while we were talking." We may summarize this as: "Survey length was just right" and leave out all the other fluff. Optionally, we could include the respondent enjoyed the flexibility of the delivery method (since they could eat during it), but if it doesn't appear often in survey responses, we can consider this an outlier.

Once you’ve summarized your data, the next step is to interpret what these results mean in the context of your research question and existing literature.

Now that we have the data and breakdown of what the participants responded with, we can begin to identify patterns, if any exist, in the data. However, be warned that it's easy to draw conclusions based on patterns that might not really exist. For instance, you may have heard of...

Correlation vs. Causation

We can't assume just because the data moves together/behaves alike that they cause one another.

In the winter, many people wear hats to keep themselves warm. In the winter, fewer people eat ice cream. This doesn't mean that wearing hats doesn't impact our sweet tooth! Here's another fun example that you can find on Spurious Correlations, a website by Tyler Vigen. While it might look like people living in Maine can avoid divorce simply by cutting margarine out of their diet, this is another example of numbers that just so happen to behave similarly, while other forces push and pull their impact.

Mismatched Data

Sometimes, participants can feel shy being honest. We call this "social desirability bias," and it might happen without it even being a conscious decision! These are little fibs that people say, consciously or unconsciously, that skew the data. For instance, if your survey asks someone whether they have a gym membership, some participants may say "yes" because it makes them appear to be living a healthy lifestyle, or stronger, or associated with fitness. Sometimes, these data skews always depend on context: if someone responds that they don't spend a lot of time of social media, but also share they spend 4 hours every day on Instagram, that data doesn't necessarily line up.

Where to Look for Patterns

Instead of drawing conclusions based on what looks or feels right, let's turn to the data. Some places we can look to find patterns:

Demographic Information

Were men or women more likely to respond they strongly disagreed with your question? How about those who were older or younger than 20 years old? Do students from a certain area of study (e.g., computer science students) feel more or less confident of their technology skills compared to their peers? Since we can group demographics based on the pieces of these identifiers, it can help us create a fuller image as to the norms of that particular group in terms of the survey responses. This is called crosstabulation (AKA "crosstabs" in data lingo) and it helps us understand which ____ is more likely to _____, or if X group tends to think Y thing.

Repeated Phrases or Words in Open-Ended Responses

It's especially promising when respondents use the same words to describe their attitude or position without multiple choice answers pushing them into a box. Maybe they actually feel "angry" and not dissatisfied, or maybe several respondents have the same kind of question. Identifying similar phrases and concepts gives us another lead on the connections we can find.

Finally, remember: patterns are only meaningful if they help answer your research question. Always ask: how does this connect back to what I wanted to know?

This is when we start putting everything together and put pen to paper (or the digital equivalent of that). Here are a few tips to help keep you focused:

Don't Write Chronologically - Start With an Outline!

Resist the urge to begin with the abstract or introduction. While they come first in the paper, they are the hardest to write before you know what your study actually shows. Instead, try this order:

• Methods: This is an easy place to start, since you're just describing what your process is, not drawing any conclusions.
• Discussion: Once you’ve identified patterns, explain what they mean to your research questions.
• Conclusion: Summarize why your findings matter, any limitations you had, and where future research could go.
• Introduction: Includes your background research (you can copy + paste your notes here as a starting point), your hypothesis, and your research questions. Be sure to also highlight the gaps in the literature; this shows why your study matters and sets up the need for your research.
• Abstract: A one-paragraph summary of the whole paper. That's why it should come last!

Synthesize, Don't Summarize

You, and your readers, should have a strong understanding of what happened (discussion) and how it happened (methods). Your conclusion should not just rehash everything they just read, but rather the case for why it matters and helps fill in the research gap. Take care to connect the dots between the data and how it connects to the bigger network of literature on your topic.

Review Format Guidelines

For your term paper, you must use IEEE/ACM standards - which means that your paper must be in two columns on the page, among other things. For more information on formatting in IEEE standards, check out the IEEE Format libguide page or visit the IEEE General Format guidelines from Purdue OWL (Online Writing Lab).

This is the home stretch - you may be tempted to call it "good enough" and turn in your assignment before a final check, but it can make the difference between a good, and a great, paper. Here are some things to look for:

• Have all the necessary parts been included? This means all the parts of the research paper are written, the requisite number of questions were written and included in the appendix section, each group member's contributions were listed in the appendix, etc. Review the assignment rubric careful and check off what elements are included as you go. Give yourself at least a few days before the deadline to finish cleaning everything up.
• Do the sections answer the questions they're supposed to? Each section of the paper has a specific purpose. Make sure you're checking all the boxes on what they include, which you can read more about in Parts of a Research Paper.
• Does the paper make sense when it's read aloud? When reviewing a paper for clarity, try reading the paper out load or having someone read it back to you. Reading it out loud makes us experience the paper more like a first-time reader, who is much more likely to spot issues with grammar and clarity.

For more information, check out this video called "Tips for Revising & Editing Final Papers" from Arizona State University.