McDaniel uses “regression analysis” to claim that instead of losing to incumbent Sen. Thad Cochran by 7,667 votes, he really won by 25,000. Regression analysis is a method of estimating or predicting the relationship between variables, such as what effect a price increase might have on sales of an item.
McDaniel’s legal team didn’t do its own regression analysis of black voters. It submitted as “Exhibit E” a column from the FiveThirtyEight blog posted the morning after the runoff. It’s headlined “It Looks Like African Americans Really Did Help Cochran Win.”
FiveThirtyEight does respectable work crunching election numbers. But the post includes disclaimers such as “It’s still too early — after the vote and in the morning — to definitively answer this question.” And the column notes more data would be needed for accuracy.
McDaniel likewise did not hire his own poll but submitted a Chism Strategies automated telephone survey from mid-July of 463 self-identified Democrats who said they voted in the GOP primary. The poll says 71 percent intend to vote for someone other than Cochran in November and that black Democrats would be 10 percent more likely to do so.