Frequently Asked Questions
What is Fat Finger?
Fat Finger is a free website that publishes a topical three-question mini-poll every Tuesday, Thursday, and Saturday, and offers a variety of other means for users to provide their views.
In exchange for their opinions, users gain access to real-time interactive poll results and personalized information about where their responses place them among the wider public.
In exchange for their opinions, users gain access to real-time interactive poll results and personalized information about where their responses place them among the wider public.
What is Fat Finger’s mission?
To increase public participation in public opinion.
How is Fat Finger different from other polling?
We contextualize the results: Our poll results link to past polls on the same topic, show users how views may have changed (or not) over time, and generally create a more complete picture of public opinion. Results are breakable by political affiliation, age and gender, so users can better understand who thinks what.
We personalize the product: Panelists can see how their responses place them along different demographic groups, and how responses they’ve submitted predict their views on other issues. Users may not only better understand others, they may better understand themselves.
We demystify polling: We make public the process behind the generation of each question and post insights to help explain select results.
We keep it brief: Each poll is only three questions along a common theme. This doesn’t mean we don’t put a lot of effort into or polls, we do. With apologies to Twain (or Franklin, or Thoreau…) we never have to say, “if we had more time we would have written a shorter poll.” We take the time so respondents, visitors, and followers don’t have to.
We personalize the product: Panelists can see how their responses place them along different demographic groups, and how responses they’ve submitted predict their views on other issues. Users may not only better understand others, they may better understand themselves.
We demystify polling: We make public the process behind the generation of each question and post insights to help explain select results.
We keep it brief: Each poll is only three questions along a common theme. This doesn’t mean we don’t put a lot of effort into or polls, we do. With apologies to Twain (or Franklin, or Thoreau…) we never have to say, “if we had more time we would have written a shorter poll.” We take the time so respondents, visitors, and followers don’t have to.
Does Fat Finger have a representative panel?
Yes. Participants across the electoral spectrum participate in our polls because questions are written to discover opinion, not to support an agenda. For over a decade, the political affiliation of our panel has been within a few percentage points of Gallup’s party affiliation statistics
Are poll results weighted?
Yes. In any given poll the demographics of the respondents may be skewed, so we weight each poll’s responses based on who replies to the poll. For instance, if 30-44 year old, male, Republicans make up 12% of the US adult population, and in a given poll that demographic comprised only 10% of our respondents, then each of their responses would get a weight of 12/10 = 120%. Conversely if in a given poll that demographic was 14% of our respondents, then each of their responses would get a weight of 12/14 = 86%
Can you say a bit more about weighting?
We use Iterative Proportional Fitting. IPF works by adjusting the weight of each response based on the demographic characteristics of the respondent. Specifically, the IPF algorithm begins by establishing a set of target demographics, such as age, gender, and political affiliation, that we want to accurately represent in survey results. The algorithm then iteratively adjusts the weights of each response until the overall distribution of demographics matches the target distribution. This process continues until the weights converge to a stable solution. The result is a representative weighted sample that accurately reflects the target population.
How does Fat Finger handle smaller demographics?
Some demographic options available to users during signup comprise a relatively small portion of the country, such that breaking results by that group may mislead rather than illuminate. For instance, it is estimated that people who identify as non-binary make up about 2-10% of young adults in the US but only 1% of the country as a whole. As such, in a poll with 1000 respondents we could expect 10 respondents to be non-binary. The results of a 10-person poll are not statistically reliable to represent that group, and as such we don’t break results by groups that size. A group must be at least 10% of the US voting age population for us to break by it.