English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Open

that is oriented to graphing low-dimensional distributions and model results. But this notion of graphing bivariate relations as predictions-and-data-points becomes more and more difficult to follow as the number of dimensions increases. There are two ways of responding to this problem. The first, one that many statisticians will favor and that, I think, Healy implicitly leans toward, is to be extremely suspicious of multivariate models and to shake one’s head sadly when results from highly leveraged models are interpreted. There is much to be said for this response. It is not only statistically conservative, but it fits a way of doing social science—one that is oriented to crisper questions, more convenient research designs, and proof via visual inspection—that is widely appreciated outside of sociology. (If you want to publish in Science, say, bivariate relations are often preferred if possible!) However, sociology has a way of thinking, going back to Durkheim’s Suicide, in which contenders battle it out by adjudicating between possible interpretations of one relation between variables by adding another variable. We could split the sample, but when the number of contenders gets large, or each brings many variables, then this becomes impracticable. Moving sociology away from this paradigm will not be easy. If Healy thinks this is the way to go, he will need to bang a drum more vigorously than he has done here. Pushing the field in that direction probably cannot hurt; even if we don’t really go there, we need repeated caution about multivariate models. But it seems that it might be a lost opportunity if we were to consider only data sets possessing such elegant simplicity. There is another approach, which is to use visualization to try to better understand the nature of our high-dimensional data before attempting to fit models. There is valuable work coming from the computer science realm, broadly understood, on ways of constructing such visualizations, including ones that end up printable in black and white on regular paper. These have their own difficulties and obscurities (for example, the ‘‘perplexity’’ parameter for the t-SNE algorithm), but this is probably not the time for sociology to do an aboutface and flee multivariate data structures. What we now need is a practical introduction to high-dimensionality data visualization, one that builds on the work here on low-dimensionality data visualization. Healy seems like just the person to do it.

What Should Historical Sociologists Do All Day? Starving the Beast , the Reagan Tax Cuts, and Modes of Historical Explanation