You're completely right. I would just point out that it's very close even when the question is explicitly phrased to specify politicians and civil servants (45 - 46%). It's likely that those with more trust in politicians are most more likely to be actual voters.