Make no mistake, president Trump is still polling underwater, but what's remarkable is how little underwater he is given the deluge over the past year plus of negative stories that the media continues to purvey about him regardless of substance. The unspoken secret however is that these polls mean zero to the president, zero to his supporters and zero to his re-election chances as well as nearly zero to the Republican controlled congress.
Nevertheless it's all worth a brief debunking. Here's what the latest RealClearPolitics polling average of job approval for the president looks like.
In the recent average there are a total of 12 polls. Gallup (having the president at -17%), Reuters (-18%), CNN (-19%), CNBC (-14%) and CBS News (-20%) all are polling all adults. That's been proven to be unreliable with respect to electoral results. They are often unreliably skewed as well. For example take CNN's poll. On page 6 it suggest 37% of Democrats approve of the president's job. In a poll of 1037 respondents and a confidence interval of +/- 3.5%, and the standard 95% confidence level, reverse engineering the statistics, it means the sample contained 429 Democrats. That's a poll that contains 41.4% Democrats. Similarly the poll contained 235 Independents (22.6% of those polled), leaving 373 Republicans (36% of the polled population).
While Democrats tend to be more numerous a +5.4% skew towards Democrats and a much larger proportion of Independents than a lot of historical polls. The latter point could be a real variance from the past but the skewness of the poll is still highly questionable. [If you are interested in reverse engineering the polls minus all the stats, there's a good basic calculator here.]
Much of the remaining polling - in fact all but Rasmussen - look at registered voters. It's historically more reliable but based on that sort of polling, president Obama would not have won re-election in 2012. The merits of each of those polls require their own individual assessments, something I'm not prepared to do here but it is worth noting that one pollster is a Democrat firm, and as I've pointed previously, Marist polling has issues (something the election results bore out). While everyone is now saying most all of the polling in 2016 was correct within the margin of error nationally, there were very few polls that actually had Donald Trump ahead state-by-state in the electoral college. Yet that's exactly what happened. Issues with polling is nothing new.
Putting stock in the polls of the day is a dangerous and misleading game. All that said, there is a continuity of evidence when even Rasmussen has president Trump at -9%. Rasmussen was so off in 2012 that in 2016 I tended to disregard it. Nevertheless Rasmussen has Trump much better off than his predecessor was for much of his first term. If I were president Trump I would not be discouraged by the polls. It still really is about getting things he's promised, actually done.