Karen Tumulty’s interesting story on The Washington Post’s website Thursday highlighted the fact that President Barack Obama is losing the confidence of key parts of the coalition that elected him. The story reports the changes in Obama’s job approval within subgroups in the latest Washington Post/ABC News poll, fleshed out by details from follow-up interviews with participants in those polls, and also results from a series of focus groups with women in Little Rock and Des Moines conducted by a GOP and a Democratic pollster.
The thrust of the quantitative part of the article was straightforward. Obama has lost support among women compared with previous points in time, and also among young people and Hispanics. The article highlights, “the degree to which the president’s approval has slipped among key parts of the Obama coalition -- the women, youth, and Latino voters most responsible for putting him into office.”
As far as women are concerned, the article notes that Obama’s approval/disapproval ratio among women is nearing an all-time low in the poll.
The article provides a good opportunity to look at a key issue in the analysis of a president’s support (or other similar measures) -- the value of relative comparisons. If a president’s overall approval rating has been rising or dropping significantly among the entire population, then we would hold out the expectation that his approval rating would also rise or drop concomitantly among each subgroup of the population. The question becomes one of -- in addition to looking at the absolute rise or drop in approval among a subgroup -- analyzing the relative change in approval among a subgroup compared with the overall change in approval.
By way of example, given the free fall in automobile sales in 2008 as the recession hit, we would expect that Ford Motor Co.’s auto sales would fall dramatically as well. The fact that Ford’s sales fell in 2008 in isolation wouldn’t be as interesting as the degree to which Ford’s sales fell relative to the drop among the industry as a whole. President George W. Bush’s job approval rating soared among women in September and October of 2001. That could seem remarkable, except that his approval rating also soared among every other subgroup in the population as a result of the rally effect following the 9/11 terrorist attacks.
Along these lines, The Washington Post article does provide a basis for making these types of comparisons. The story notes that Obama’s approval rating among women has fallen from 60% in January 2013 to 44% in the most recent poll. Although not mentioned explicitly in the article, Obama’s overall approval rating between these two polls is readily available, and the trends show that his overall approval rating over that period fell from 55% to 42%. Thus, and this is the important finding, his approval rating fell by 16 percentage points among women from January 2013 to September 2014 in the Post/ABC News poll, and fell by 13 points overall during that same time period. This allows one to make a very cautious conclusion that the president may be suffering slightly more among women (a drop of 16 points) than he is overall ( a drop of 13 points), but given the margin of error on the 500 or so women in each of the two polls, this conclusion must be treated with some caution.
Our Gallup Daily tracking data allows us to use very large sample sizes to look at these comparative trends in more depth. On a long-term basis, this comparison -- similar to the Post polls -- doesn’t show a lot of change over time in Obama’s relative standing among women.
Here is a graph of our Gallup data, based on 30,000 interviews a month in 2009 through 2012, and 15,000 interviews a month in 2013 and 2014. The graph shows Obama’s job approval rating overall and among women.
And this graph below distills the gender gap into one line -- the net difference in approval, on a monthly basis, between women and the overall average during the entire Obama administration.
Both of these graphs reflect the same data, in slightly different ways. Obama has, so far in his term (through August), averaged an approval rating among women that is 2.9 points higher than his approval rating overall, calculated on a month-by-month basis. So far this year, this gender gap has been 2.6 points, only marginally down from his term average. Women gave him a two point higher approval rating in May, four points higher in June, three points higher in July, and two points higher in August. So far in September, after 11 days of interviewing, his approval rating is two points higher among women than his overall rating.
So, the conclusion is that if Obama is losing support among women on a relative basis, it is very marginal -- to the tune of a drop in the gap between women’s approval and approval overall of less than one percentage point.
We will want to continue to monitor the difference in the weeks and months ahead. If the gender gap falls for example to one point, or no points going forward, then the conclusion will certainly be that the president is losing the relative edge among women that has been a key part of his coalition to date. But it will take time to monitor the degree to which that is happening. At the moment, it’s safer to say that women are still more supportive of Obama than the average American, even though women’s absolute support has eroded over time along with everyone else’s.
We will be reporting the results of similar analyses among other population subgroups at Gallup.com in the days ahead. I can say there is some evidence that Obama has, from late last year to the present, been losing his edge among Hispanics both on an absolute basis and relative to the general population, in part reflecting a drop from the very wide levels of support he attracted among that group during and after the 2012 election.