something of particular substance or exerting a specific effect — from overlapping and non-specific “noise.” Central to this effort is reliable modeling of probabilities and uncertainties, as well as disciplined reasoning.

For this paper, do not embed questions just put number then answer ex:
1. answer
2. answer
This Online Week has two foci. The first is to continue our focus on important aspects of quantitative thinking. Nate Silver discusses quantitative approaches to detecting signal — something of particular substance or exerting a specific effect — from overlapping and non-specific “noise.” Central to this effort is reliable modeling of probabilities and uncertainties, as well as disciplined reasoning.
The second focus of this Online Week is to introduce Latent Variable Structural Equation Modeling (SEM). While many students become quite anxious at the appearance of the word “equation” in its name, and try to steer away from SEM as fast as they can, it 1) is actually just a combination of factor analysis and regression (two ordinary advanced statistical procedures), 2) Meyers et al. comments aside, it actually has no special requirements in what data is collected or how it is collected (for example, a questionnaire or other survey instrument is administered in the same way with SEM as it would be with any other statistical procedure in mind), and 3) requires data to be entered item-by-item in the dataset (so each item of a measurement instrument is a variable in the dataset) but this is no different than what would normally be done so that reliability analyses can be conducted. Adding to the scare (as in “scary”) factor, persons using SEM often describe complex and fancy-looking models with mind-bending illustrations involving many ovals, squares, and lines that can appear quite intimidating. This is commonly combined with the presence of many Greek letters and matrix descriiptions. But the reading and interpretation of the statistics output is actually quite straightforward and the statistical software does all of the drawing and illustrating for you (so no problem!). But what about the equations, you ask. While the statistical software finds itself quite busy with all sorts of equations in SEM, the user does not have to contend with any equations. Even required sample sizes are quite doable, depending on the complexity of the particular model being tested using SEM analysis. So while the presentation of SEM can involve a lot of bark, it actually has no more bite than any other advanced statistical procedure. And for many research questions, SEM has highly important advantages.
Your reading for this Online Week is from Nate Silver and Meyers et al. (2nd edition of Meyers et al., NOT the third edition that you have as a textbook), see Course Syllabus for details.
Submit a response, stated in your own words, to each of the following ten questions:
Nate Silver reading:
1) As described in Chapter 4, why does the National Weather Service still employ human beings to interact with the data before forecasts go out? What do they do? How are they value-added? What does this suggest about your interacting with — getting your eyes on — your Dissertation data (descriiptive statistics, checking for assumptions being met, etc.) before conducting formal statistical analyses? Describe why it is important that you “get to know your data well” before actually conducting any statistical analyses.
2) In Chapter 4, Silver notes that what makes a forecast good is accuracy, honesty, and economic value. Review these concepts in Silver’s chapter, and discuss how honesty applies to your statistical analyses for your Dissertation (to your review and understanding of the literature; to the particular statistical procedures you utilize based on that review and understanding; to the analyses you conduct and how you conduct them; to how you read, interpret, and report the results; and to how you present and discuss the findings). Is the statement “A good Dissertation is a done Dissertation” consistent with an honest approach? Flipping that statement on its head, how about the statement “A done Dissertation will be a good Dissertation”?
3) As described in Chapter 6, what is uncertainty and why is it important to communicate? Offer an example from Chapter 6 to make your point clear.
4) So much data, but so little causation. As described in Chapter 6 (in the section “Correlations Without Causation”), why is it difficult to make predictions about the economy even though there are so many variables available to work with? What does this suggest about the importance of advanced statistics and the inclusion of control variables for any serious analysis of data, including for your Dissertation?
5) After reading Nate Silver (Chapters 2, 4, and 6), why do so many predictions fail — but some don’t? In answering this question, it is important to include your understanding of quantitative thinking. How is it special or distinct from other ways of thinking? It is also important to include your understanding of foxes and hedgehogs, probabilistic thinking, modeling uncertainty, and their importance for detecting the signal from the noise.
Meyers et al. (2nd edition) reading:
6) A structural equation model is composed of a measurement model and a structural model. In your own words, provide a brief but specific descriiption of the measurement model and the structural model.
7) Although structural equation modeling (SEM) contains elements of path analysis, in highly important ways SEM is very different from path analysis. Describe these differences, as discussed in Meyers et al. (2nd edition).
8) Because of structural equation modeling (SEM) and its particular capabilities, the use of path analysis (by itself, as its own stand-alone procedure) has notably declined. As a consequence of the development of SEM, describe the most important reason(s) for the decline. Be specific. (Instructor’s note: While there has been a decline in the use of path analysis with the advent of SEM, path analyses continue to be conducted and make an important contribution to our understanding of phenomena. So it’s all relative here, yes SEM gets the gold medal as an analytic approach, path analysis gets the silver medal, which is still good.)
9) The assumption of “no measurement error” is often made in statistical analyses. It is not a realistic assumption and basically is never met. How does structural equation modeling view and respond to measurement error, and how does this response represent an important advance compared with other statistical procedures?
10) Regarding Dissertation Proposals, can you envision how structural equation modeling may be an appropriate statistical analysis? If yes, please describe. If no, why not?