Information
Photography
Syndicate This Blog
Friday, February 1. 2019
Misleading Survey Results
One of the pitfalls of data science is bad source data. I noticed a particularly egregious example at https://www.debate.org/opinions/can-torture-be-justified-as-an-added-form-of-punishment today.
Notice the actual question on the right "Can torture be justified as an added form of punishment?" versus the image above the yes/no selection, "Torture always wrong" and their placement. Beyond this being a study in bad interface design and the image itself introducing bias by the designer, which one is being answered by what percentages of those seeing this page?
Should the results of this survey be trusted? What if this was to be used as part of a data aggregate by a public policy maker to make a decision? Bad data of this nature can become terrifyingly influential as we rely on increasingly complex automated data tools to make decisions.
Notice the actual question on the right "Can torture be justified as an added form of punishment?" versus the image above the yes/no selection, "Torture always wrong" and their placement. Beyond this being a study in bad interface design and the image itself introducing bias by the designer, which one is being answered by what percentages of those seeing this page?
Should the results of this survey be trusted? What if this was to be used as part of a data aggregate by a public policy maker to make a decision? Bad data of this nature can become terrifyingly influential as we rely on increasingly complex automated data tools to make decisions.
Trackbacks
Trackback specific URI for this entry
No Trackbacks