Monkey Cage post on support for the Confederate battle flag
The post is here.
Data for the Hutchings and Walton study are here and code is here.
Data for the Southern Focus Poll are here and code is here.
Here are factor analysis results for the Hutchings and Walton study and for the Southern Focus Poll.
---
UPDATE (July 7, 2015)
Corrected the code link for the Southern Focus Poll.
The Monkey Cage post is discussed in a scatterplot post.
More code to support this claim about Southern choice for words to describe food, as mentioned here.
Dr. Zigerell, would you mind explaining how you arrived at approximately 28% disapproval of interracial marriage for the Hutchings & Walton study? Looking at the raw data and at your coding, I can't quite figure it out. For all respondents approval/depends/disapproval was 288/201/212 for a total of 701 respondents with non-missing responses. For white respondents only (and it seems as though you're analyzing only the white respondents in the post), approval/depends/disapproval totals were 192/148/188 out of 528 respondents. That's 36% disapproval among whites and 30% disapproval among all respondents.
Your coding suggests that at one point you dropped respondents who selected "depends" for the marriage question for some of your analysis, but dropping those respondents increases rather than decreases the proportion of valid responses that were disagree, so that can't explain the discrepancy.
Thanks for your help!
Hi Andrew,
The text and graphs for the Monkey Cage guest post went through several revisions. The graphs that the Monkey Cage posted were based on the second draft of the post, which had this passage:
"Each heritage and hate measure was placed on a 0-to-1 scale. The figures report the difference in means for each measure: larger values indicate measures with a larger difference between persons who preferred the flag with the confederate battle emblem and persons who did not."
The line in the Stata code of "ttest nomar01, by(flag)" will return the 0.27 value for the black-white marriage item and [the other t-test commands will return] the plotted values for the remaining measures.
The Monkey Cage editor suggested revising the graphs to show percentages instead of mean differences, to make the results more intelligible to readers. For some reason, the updated graphs with the percentages weren't posted, but I just published the post that I had in the queue about the R code for the heritage percentages graph here: https://www.ljzigerell.com/?p=3440.
Update (July 8, 2015)
Added the text in square brackets.
Ah, much clearer now. Thanks!
No problem!