3+4+3 pages

A Detailed Guide for Individual Projects

Cool car features: people could use but can’t find in their cars (observation). This can be a exploratory study (qualitative) based on observing other people’s car and interviewing them what features are their dream additions in a car. You can summarize the popular ones and provide some possible explanations.

The project topics are described above. The final report should include a report with results of your analysis (description of your objective and how you collected data, did the analysis, your results, interpretation of results, conclusion, and any limitations of the study that you can identify.

Here are the steps.

1. Have a good understanding of what your project goal is. The list of topics provides brief descriptions of the objectives and some suggestions. Every project starts with a question that needs to be answered or a hypothesis needing to be tested. The question of the project could be as simple as measuring some consumer behavior or opinion, or testing some relationship between variables. If you are not clear, see me during my office hours or call me during office hours. The management is hungry for insights and it is a risky business for them to base decisions on untested assumptions or gut feelings. Consider the goals of the following projects.

2. Collect Data and organize them for analysis. For descriptive research using data from primary or secondary sources, the analysis includes averages, percentages, cross-tabulations, correlations, and regression results to name a few. We are not doing analysis for the sake of analysis, but use it to generate the insights needed to answer the project question(s).

Data should be entered such that each row contains information about the unit of analysis. If the data is coming from a survey of individuals, then each line of data is representing all the information collected from that individuals. The questions in the survey then becomes the column of the data. Consider the examples below.

a. Sample Used Car Data

b. Sample Gas Price Data

3. Check data for accuracy and make sure you have all the variables for statistical analysis. Certain statistical analyses need recoding of variables. For example, regression analysis requires you to recode nominal (also called categorical) variables such as brands, city, color, into dummy variables (as shown in class). The number of dummy variables you need per nominal variables is the number of categories in that variable, minus one. For example, there are four colors in the Car data. You need dummy variables for Black, White, and Gray. Which one you leave out is optional. The table below shows what the dummy variables look like.

As you can see, the dummy variable (in red font) simply tells you presence (value of 1) and absent (0) of that feature. You will need to create dummy variables for Make and model, in similar fashion.

In the gas price data, location and brand are nominal variables, so they also need to be represented using dummy variables, as shown in the table below.

4. Analyze data. The statistical procedure to use will depend on the project goal. If you picked topic #1, we would only need to analyze text data (transcripts of interviews, focus groups etc.). Since the gas price and car price data deals with prediction (descriptive research design), we need numerical analysis and regression model seems best suited for the task.

1