[Sytion06]

What changed

• Ethics discussion. I kept the team's version from main and added a concrete privacy example. One neighborhood has only one recorded sale, so we'll avoid publishing its parcel details or drawing conclusions about the neighborhood from that sale.
• Analysis plan. I tied the next steps to Figure 1 and the methods covered in class. We'll keep the living-area × dwelling-type interaction, use partial F-tests to compare nested models, and check residuals after each refit. The plan explains when we'd try log price or a squared living-area term and how we'd check whether the three unusual partial sales change our conclusions. It also sets dates for the analysis, poster, and recording.
• Introduction and preliminary results. I added an example of how an appraiser could compare an extra 100 square feet in a duplex versus a detached home in historical Ames sales. I qualified the interaction test because the widening residual spread makes its p-value and confidence intervals provisional.
• References and PDF. I added a reference for the Ames dataset and codebook, distinguished it from De Cock's article, formatted the journal references, and rebuilt the proposal PDF from the R Markdown.