Research paper

RESEARCH PAPER GUIDELINES Developed by Joe Turek

Your research paper should be structured as shown below. You are free to use whichever style manual you prefer, e.g., APA, MLA, or Chicago/Turabian, so long as you do so consistently throughout the paper. Length should be between 10 and 15 pages, exhibit can be interspersed throughout the body of the paper to permit easy reference, and all resources should be fully and accurately cited. You don’t need to include your data, but you might want to run Descriptive Statistics and include the results in an Appendix. I would like for you to double space the paper, (which leaves me some room for comments), and use a 12-point typeface with an easy-to-read font. COVER PAGE: title, your name, data. (Does not count in the page total). INTRODUCTION: A relatively brief description of your project. What question are you trying to answer? Why is this important? Include any evidence to suggest that this problem or question is becoming increasingly important. LITERATURE REVIEW: What does the literature (the articles you read and the sources you consulted) have to say about the topic you’ve selected? You are, essentially, describing what others have done, how they approached the topic, and what they found. How is your paper going to contribute to a better understanding of the topic? In other words, if there is something you’re doing that wasn’t done in the literature you consulted, you want to point that out, perhaps the inclusion of new variables, the use of different data (coverage or time period), or the adoption of different methodology. If there is any disagreement within the literature you reviewed, perhaps A found that gender was statistically significant while B found that it wasn’t (or A found that the coefficient on income was positive while B concluded that it was negative), you’ll want to point this out and explain why the difference is important. Even if one of the articles you consulted didn’t use regression analysis, it is likely that the author(s) implied a relationship between two or more variables. You might discuss why the methodology used was inappropriate or the reasoning incorrect. THE MODEL: In this section, present your regression model in the form Y = a + b1X1 + b2X2 + … + bnXn, with variable labels included. For example, you might write: Q = a + b1PRICE + b2INCOME + b3CHICKEN + b4GENDER Where Q is the quantity of steak consumed, PRICE is the price per pound of steak, INCOME is family income, CHICKEN is the price per pound of chicken, and GENDER is a dummy variable (GENDER = 1 if the buyer is female, 0 if the buyer is male). The names or labels you use are important because your reader will need to refer back to these when interpreting the regression output. Include, also, a very brief explanation for why you chose to include each variable in the model – you might justify your selection based on one of the articles in your literature review, the underlying theory, or maybe even personal experience. Sometimes – and I don’t suggest this for this paper – you would also include what are referred to as your a priori expectations for the sign of the coefficient attached to each variable. This is important because, if you have a sign expectation, which is actually pretty common, you would conduct a one-tail test of significance rather than a two-tail test. We can discuss this one-on-one in more detail if you’re interested.

DATA: In this relatively brief section, you are going to discuss your data sources and raise any issues or concerns you might have about your data. For example, you might start by explaining if you are using cross-sectional data (observations on multiple units at a given point in time, say, Tennessee’s 95 counties in 2010 or the 50 states in 2020) or time series data (observations on one unit at multiple points in time, say, enrollment at CBU and its determinants for the last 100 years), and why. You do not have to discuss your sampling met2hodology or the validity and reliability of your survey instrument if you are not using primary data. Since you are using secondary data, which are data that have previously been collected by some other person or entity, you need only identify the source and speak very briefly about data collection methodology, if there are relevant concerns. I would run the descriptive statistics for your data set and include as an Appendix. ANALYSIS: Before you actually run your regression, specify the significance level you’re going to use for your hypothesis tests. In this section, you will run and discuss your regression analysis. After you’ve run the regression, I would cut and paste directly into the body of the text to make it easy for your reader to follow your discussion (putting your findings in a technical appendix will require people to hunt for your results). You can cut out the ANOVA (the middle table) and include only the regression statistics (table at the top that contains the R and R2) and the estimates (the table at the bottom that contains the estimated coefficients, standard errors, t stat, etc.). You should write out your estimated regression model, substituting coefficient values for letters, e.g., Q = 5000 – 4 PRICE + .2 INCOME + 2 CHICKEN - .2 GENDER and interpret the results. Which variables are statistically significant? How do you know? Also, talk about the signs of the coefficients. What, for example, does a -4 coefficient on PRICE mean? What about a -.2 coefficient on GENDER? CONCLUSIONS: In this section, you should summarize your (important) findings and identify the ways in which they are consistent with, or deviate from, the research findings discussed in your literature review. You should also identify any policy implications that follow from your conclusions, e.g., “the negative correlation between exercise and myocardial infarction suggests that people should participate in 30 minutes of cardio at least 3 days a week.” In your introduction, you explained why your topic was important and worth investigating. This is an opportunity to extend those remarks in light of your findings. LIMITATIONS: In this section, you can do any number of things. If there are factors which limit the generalizability of your conclusions, you should discuss that. For example, using U.S. data might limit your ability to extend your conclusions to individuals or organizations in other countries. Using data from 1980 may limit your ability to extend your conclusions to 2020 if underlying conditions have changed. Gathering data from 11 unique companies may not tell you very much about the hundreds of thousands of companies that are dissimilar. Or maybe you were forced to use median household income because per capita income was not available – might that have distorted your results and biased your conclusions.

-------------------------------------------------- These guidelines are intended to provide you with a quick and easy-to-follow reference guide. If you’ve got any questions or if there are additional topics that demand coverage, please let me know and I will be glad to expand the document.