Dear @Google_Surveys, The Statistics Addict In Me Wants To Hug You


The data scientist in me has a new crush: Google Surveys. Last night, Google’s consumer marketing division released a deliciously rich dataset of citizen reaction to the Democratic National Convention speeches with a rapid response survey. Unlike the notoriously young, liberal bias of social media sites, nearly everyone with an internet connection uses Google (it is a verb, after all). Downloadable data can be sliced and diced by age, gender, and income, and has already revealed some intriguing insights. Bill Clinton’s speech, in seems, was a bipartisan bananza: both young hipsters and old money-bags agreed with him that no president could turn around the economy in 4 years. It also seems that both men and women equally prioritize the issue of women’s rights. Data experts and stat lovers can play around with the slick interactive user interface here.

Of course, self-reported surveys do have their limitations. For instance, research shows that people are dreadfully bad at predicting whether they will vote. But, users do know how they feel about issues, especially if asked right away. So, rapid response surveys, combined with Google Politics’ new search trend monitor, makes the search titan a welcome source of near realtime and (relatively) unbiased data on the impact of political campaigns.

However, we politely beg the Google Survey team to work on their question methodology. For instance, the question, “Do you agree with Bill Clinton’s claim that no president could have repaired the economy in four years?” is a survey no-no: so-called “double barrel” questions make it hard to distinguish respondents who agree with Bill Clinton, his opinion, or both. Nor is giving respondents 27 different ways of characterizing his speech particularly useful (is there a difference between “good”, “positive,” and “rousing”?). But, methodological belly-aching aside, this is a wonderful tool that we hope to make use of in our articles soon.