Data in this study came almost entirely from content analysis performed by Pew Research Center staff of petitions and responses downloaded from the White House’s “We the People” website.
Content Analysis
In total, 4,799 archived petitions were downloaded from the API of “We the People” on Aug. 2, 2016, covering the period from Sept. 22, 2011, the day the site was created, through July 3, 2016. The website only archived petitions that received at least 150 signatures. Therefore, petitions that were posted on the site but did not get at least 150 signatures in 30 days were excluded. Along with the text of the petitions, the downloads included metadata such as the number of signatures, the date created, the URL of the White House response (if any) and the tags used by the petitions’ authors.
Human coding of the petitions
Each petition was coded by an experienced researcher for two variables: whether the petition dealt with U.S. foreign policy and the subject matter. Because there were so many petitions that crossed over in subject matter, each petition could be assigned one or two unique subject categories.
Researchers created a list of 135 subject categories, although many of those categories were combined for the sake of analysis. To choose the categories that best fit each petition, researchers referred to the title along with the text written by the authors. The tags chosen by the authors were not used to influence the Center’s coding because those tags were not consistent and because different authors may interpret the tags in different ways.
Intercoder Testing
To test the validity of the coding scheme, two researchers each coded the same 325 petitions. The percent of agreement for the two variables were as follows:
Foreign policy: 95%
Subject matter: 85%
Human coding of the White House responses
The responses written by White House staff were also downloaded using the site’s API and covered the same time period as the petitions. There were 227 responses during the time period studied. Of those, three were responses to petitions that did not meet the signature threshold.
Some responses written by the White House were connected as answers to multiple petitions, so there was not a one-to-one correlation between petitions that reached the signature threshold and responses.
In addition to the text of the responses and the accompanying metadata, researchers categorized each response to see if the text of the responses referred to specific policies and if the responses included a declaration by the White House that it was unable to comment on that specific issue.
