Stalking State Statutes: A Critical Content Analysis and Reflection on Social Science Research

2019 ◽  
pp. 1-22
Author(s):  
Caralin Branscum ◽  
Seth Wyatt Fallik ◽  
Krystal Garcia ◽  
Breanna Eason ◽  
Kayla Gursahaney
Author(s):  
Leeann Bass ◽  
Holli A. Semetko

This chapter explains content analysis, which is a social science research method that involves the systematic analysis of text, media, communication, or information. The source, the message, the receiver, the medium, and the influence of the message are all topics that have been studied using content analysis and in combination with other methods. There are deductive and inductive approaches to content analysis. Two widely cited studies using content analysis take a deductive approach: using predefined categories and variables based on findings and best practices from prior research. Studies taking an inductive approach to content analysis, by contrast, have an open view of the content, usually involve a small-N sample, and are often based on a qualitative approach. Meanwhile, much has been written on methods and approaches to measuring reliability with human coders. Traditional content analysis uses human coders, whereas a variety of software has emerged that can be used to download and score or code vast amounts of textual news data. The chapter then identifies key benefits and challenges associated with new computational social science tools such as text analysis.


2019 ◽  
Author(s):  
Christian Pipal ◽  
Martijn Schoonvelde ◽  
Gijs Schumacher

Sentiment is an important concept in social science research to study issues such as public opinion, media tone, or campaign dynamics. While automated content analysis methods such as the application of sentiment dictionaries provide reasonably good results overall, these methods do not take into account that the sentiment dimension of a word can depend on its context. Adopting novel methods from computational linguistics, we present a semi-supervised sentiment-topic model (JST and reversed JST) to estimate topics and sentiment simultaneously in political text. Validating our results with a crowd coded data set of UK prime minister speeches, we show that taking topic specific sentiment into account improves the accuracy of the generated sentiment scores compared to commonly used dictionaries. Our findings highlight the connection between topics and sentiment, and have important implications for the analysis of emotional content in political texts.


2019 ◽  
Vol 13 (1) ◽  
pp. 93 ◽  
Author(s):  
Adesoye Isiaka Mustapha ◽  
Ikponmwosa Ebomoyi

Sign in / Sign up

Export Citation Format

Share Document