Based on the decision tree provided in Canvas, which statistical test should you run for each of the 3+ research questions you chose? Why?

Questions to use for Data
1. Is there a relationship between amount of television watched and achievement in school?

2. Do students that spend more time on education make better grades?

3. Do males or females spend more time on education?

Data Collection Instructions
For this assignment, take a minimum of three of your research questions (that can be solved using the 2018 data file [see Canvas for it]) and answer the following:

Based on the decision tree provided in Canvas, which statistical test should you run for each of the 3+ research questions you chose? Why? Tell me about the “path” you took to get each solution. The fuller the answer for each, the better. In some cases, this will involve you knowing whether the data in your research questions are nominal, ordinal, interval, or ratio. I will be checking all of them for accuracy. If you are unsure about the difference between nominal/ordinal/interval/ratio data, I admonish you to look at the PPTs I’ve given you, the class notes, and the text. If you are truly stumped and/or believe your question cannot be answered using the tree, contact me ASAP so I can help you get on the right path. Note: I will not answer it for you unless you’ve chosen something so exotic that it is not on the tree.

I do not expect you to know how to do all of these tests (yet), only that you can use the decision tree to make decisions as to which test to use.

Data Analysis Instruction

The time has come to put your new skills into practice and analyze the data you’ve been given.

To this point, most of you have told me what variables you plan to use and which statistical paths you think are correct. I have looked at both and given each of you my individual input, corrections, and recommendations. I have also built a set of “how-to” guides for the most common statistical tests chosen to this point.

Now, it is your turn: One of the big goals of this class is to get you into doing your own analyses and knowing the basics of choosing which tests to run. I hope these do not come across as bossy or paternal.


1. This should be old news to most of you, but if you have not (1) chosen your GSS data or (2) selected which statistical tests you will use to analyze it using the path model I’ve provided, you cannot proceed in this project. These two items must be completed and signed off by me before this assignment can be done.

2. Using the recommendations I’ve given you, use my “how-to” guides to complete the statistical tests for each of the three statistical questions you’ve asked. Start early: You are probably going to have some questions.

3. One of the questions might be, “John, do I need to use all 2,348 lines of data?” Answer: Not necessarily. The reason I included all the GSS data for 2018 for all the questions everyone asked is that this is what data can look like in the “real world”: it is often large, messy, and incomplete. With the new world of “big data,” datasets can now contain millions or billions of lines of data. To answer the question: It depends. Excel doesn’t like some computations where there is any missing data, and it is OK with others. Your “how-to” guides will inform you as to which ones Excel has problems with. If you’ve chosen a statistical test where Excel is allergic to missing data, follow my guides and do a bare minimum sample size of 20 lines where all data are available.

4. It is strongly advised to copy the variables you are working with into a new, smaller spreadsheet before manipulating them. Plus, you avoid damaging the larger database you are working with. If you mess up and ruin the larger dataset, just erase it and download a new copy from Canvas.

5.For each statistical test you run, I want you to (1) tell me how many participants you used, (2) show me your results, and (3) in a sentence or so, tell me what you found and–if appropriate–if it is significant (significance is not necessary if all you are doing is summarizing some variable data. The best way to do these things is to send me the Excel file you used and note where the results are. Please, only send worksheets that contain the variables you are working with.

6. A few of you have inquired about whether you can use SPSS instead of Excel to complete your analyses: the answer is, “yes.” Email me and I will send you instructions regarding how to do each of your tests.

7. Another question: “John, how will I know what to say about what I’ve found?”