![Sample our Economics, Finance,Business & Industry journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5931/TOC-Subject-Sample-2021-Economics-Finance-Business-Industry.jpg"})
ABSTRACT:
While crowdsourcing is an increasingly common method of open-government practices to strengthen participatory democracy, its impact on governance is unclear. Using data from a crowdsourced city-plan update by the City of Palo Alto, California, this article examines the impact of a crowd’s input on policy changes. We used an enacted policy change to quantify government’s response to crowd suggestions, whether crowd suggestions are adopted in the policy changes or not. While the city responded to less than half of the crowd’s suggestions, the likelihood of its doing so increased by 51.42 percentage points when the crowd’s ideas were amplified by a citizen advisory committee (CAC), a panel of residents working with the city in the policy update. We also found that the government is more likely to respond to crowd suggestions that are perceived as actionable. These two factors—CAC and the perceived data quality—constitute a filter which the crowd’s suggestions have to pass to make into the policy. This filter created a hierarchy in the participatory practice. Although crowdsourcing intends to create equality and inclusiveness in policymaking, our findings reveal that the civic data overload and the filter hierarchy complicate the adoption of crowdsourcing as a democratic innovation in governance.
Notes
Notes
1 For the purpose of this article, we do not analyze the CAC’s responsiveness to the crowd’s ideas. We focus on the relationship between the government and the crowd, because that is the relationship between the crowdsourcer and the crowd.
6 Please note that some policies from the original policy document were deleted, which means that they did not appear in the final document. This is why the number of policies that were changed can be larger than the number of policies in the final document (211 > 202).
7 We web-scraped all of the Palo Alto Online news articles (https://www.paloaltoonline.com/square/) from the time it was founded (year 2006) to the time this data was accessed in May, 2017 (N = 6,366). To identify whether or not a news article was about transportation, we used Tf-idf to develop all of the words and phrases that are related to transportation issues. Since the crowdsourcing period is from August 2015 to June 2016, we compared the percentage of transportation news coverage from August–December for the years 2010–2016 and we did not find that August–December 2015 had a significantly higher percentage of news coverage on transportation. Similarly, we also compared the percentage of transportation articles from January–June for the years 2010–2017. We also did not find that January–June 2016 had a significantly higher percentage of news coverage on transportation. The total number of news article that are about transportation during the crowdsourcing months is 103.
8 For how scholars in the field of public management use natural language processing techniques, see Pandey, Pandey and Miller (Citation2017). For a more general use of machine learning techniques on text collection and analyses, see Manning, Raghavan, and Schütze (2008).
9 For more details on the meaning and computing of Tf-idf algorithm, see: http://www.tfidf.com.
10 Some readers might wonder how the approach of Tf-idf is similar or different from LDA, which is a popular topic modeling method in text mining. The purposes of LDA and Tf-idf are different. The goal of LDA is to help researchers generate the main topics and their associated words so that researchers can label the topic to classify a document. The purpose of Tf-idf is to give researchers weighted term frequency for each document and to inform researchers how important a word or a phrase is to a document. When we compare the three datasets—citizen suggestions, representative suggestions, and the policy changes—we are more interested to learn about the key words (phrases) for the three datasets and therefore we chose Tf-idf. When we tried to develop the category to classify citizen suggestions, we tried both LDA and Tf-idf. We found that the topics generated by LDA did not help us label the document in a meaningful way. Using the words generated by Tf-idf was better able to help the researchers to develop the category. Therefore, although both methods are helpful text analysis tools, for our data, using the word list Tf-idf generates is more useful than the topics (and associated words of each topic) from LDA.
11 As David Kenny (2015) explained, “when X and M are dichotomies f2 equals the d2/4 where d is the d difference measure described above. Cohen (1988) has suggested that f2 effect sizes of 0.02, 0.15, and 0.35 are termed small, medium, and large, respectively.” For details, see: http://davidakenny.net/cm/moderation.htm. In our case, d = 1.07, thus f2 = 0.29, which is closer to 0.35. Thus, it is a large effect.
13 In Midland Michigan, http://www.cityofmidlandmi.gov/125/E-CityHall?pd_url=https%3A%2F%2Fwww.peakdemocracy.com%2Fportals%2F242%2FIssue_3878#peak_democracy, the government has adopted a similar online crowdsourcing platform as the city of Palo Alto to crowdsource citizen ideas. In Los Altos (http://www.losaltosca.gov/community/page/open-city-hall), Klamath Falls, Oregon (https://www.klamathfalls.city/i-want-to/find/city-hall/administration/open-city-hall,) Decatur, Georgia (http://www.decaturga.com/whats-new/open-city-hall), etc., the government has used Open City Hall to call for crowd suggestions on policy issues and government budget. Beyond the United States, crowdsourcing has also been used in Finland (www.thefinnishexperiment.com) and Iceland (https://democracyoneday.com/tag/iceland/). As scholars (Aitamurto 2015; Aitamurto and Chen Citation2017) pointed out, governments across the world are facing an increasing civic data analysis challenge.
Additional information
Notes on contributors
Kaiping Chen
Kaiping Chen ([email protected]) is a Ph.D. candidate at the Department of Communication, Stanford University. Her research uses big data, machine learning, and text analysis to study civic engagement and government responsiveness across regimes. Her research has appeared or is forthcoming in journals such as American Political Science Review, International Public Management Journal, Theory and Practice of Legislation, and the Russell Sage Foundation Journal of the Social Sciences.
Tanja Aitamurto
Tanja Aitamurto, Ph.D. ([email protected]), is a postdoctoral fellow at the School of Engineering at Stanford. She examines the impact of civic technologies on human behavior and society. Tanja’s work has received a number of awards and has been published in highly ranked academic journals, including New Media & Society, Information, Communication & Society, and International Journal of Communication, and in venues such as the Conference on Human Factors in Computing Systems (CHI) and Computer-Supported Collaborative Work and Social Computing (CSCW). Tanja has attended meetings and given talks about her research at the White House, the United Nations General Assembly, the Wikimedia Foundation, OECD, the Council of Europe, and in several parliaments and governments.