Waseda Institute of Political Economy早稲田大学 現代政治経済研究所

News

10月24日現政研セミナー(ヨハン・コスキネン先生)のご案内 the WINPEC Seminar on October 24th.

※For English please scroll down.

 

10月24日(火)に行われます、現政研セミナーの日程、場所などにつきましてご案内申し上げます。なお、本セミナーはSGU(スーパーグローバル大学創成支援)との共催となります。

【日 時】10月24日(火)10:40~12:10

【場 所】早稲田大学 早稲田キャンパス3号館10階 第一会議室

【報告者】ヨハン・コスキネン先生 (マンチェスター大学 社会科学部 社会統計学 講師)

【タイトル】Analysing network data from unstructured sources

【概要】

The network perspective has proven a fruitful conceptual framework and analysis tool for understanding how societies and communities are organised and connected. Social networks afford the notion that behaviours and attitudes and other outcomes of individuals are potentially contingent on their personal contacts as well as their individual properties and properties of the contexts in which they are embedded. The canonical form of network data is collected through a roster method that prompts respondents to report their interactions to everyone in a predefined set of people. This requires a census of the members of, or frame for, the relevant set and in many research settings this is not a priori possible. Various link-tracing designs have been proposed in the literature for eliciting network data (or to simply recruit respondents) in environments where no such frame is available. We propose here to not only allow for tracing links between individuals but also to trace the affiliations to organisations and settings of individuals, as well as using information of how these organisations and settings are connected. The target of inference is to infer the tie-formation processes and to that effect we assume a multilevel exponential random graph model. A Bayesian approach has recently been proposed for accounting for the sampled nature of data and here we extend this procedure to handle the different types of ties of the overall network. We use two illustrative examples. The first example is an analysis of a network collated from newspaper reports on terrorist activities in Indonesia that we treat as a gold standard for appraising the veracity of the inference scheme. The second example is based on a network collected in the wild in the Eastern Democratic Republic of Congo, stemming from a project aimed at investigating the post-conflict livelihoods of combatants, civilians, and demobilised people. The key types of ties are personal and job-seeking ties of individuals, the affiliations with civic institutions and different armed groups, as well as alliance ties between armed groups. The collected data consist of a number of partially overlapping personal networks with individual affiliations. In addition to the substantive interest in modelling the ties between individuals, the inference scheme provides predictions of missing ties that can be used for studying centrality of different actors (people and armed groups) in the population.

 

【招聘教員】 戸堂 康之 先生

【使用言語】 英語

現政研セミナーは広くみなさまにひらかれたセミナーです。

皆様の御参加をお待ちしております。

 

 

*English

This notice is to inform you of the Waseda Institute of Political Economy (WINPEC) Seminar on Oct. 24th. This seminar is co-hosted by Top Global University Project (TGU).

【TIME】24th, October Tuesday, 10:40-12:10

【Venue】The meeting room (conference room) #1, BUILDING 3 – Floor 10.

【Presenter】Johan Koskinen  (University of Manchester, School of Social Sciences, Social Statistics, Lecturer)

【Title】Analysing network data from unstructured sources

【 Abstract 】

The network perspective has proven a fruitful conceptual framework and analysis tool for understanding how societies and communities are organised and connected. Social networks afford the notion that behaviours and attitudes and other outcomes of individuals are potentially contingent on their personal contacts as well as their individual properties and properties of the contexts in which they are embedded. The canonical form of network data is collected through a roster method that prompts respondents to report their interactions to everyone in a predefined set of people. This requires a census of the members of, or frame for, the relevant set and in many research settings this is not a priori possible. Various link-tracing designs have been proposed in the literature for eliciting network data (or to simply recruit respondents) in environments where no such frame is available. We propose here to not only allow for tracing links between individuals but also to trace the affiliations to organisations and settings of individuals, as well as using information of how these organisations and settings are connected. The target of inference is to infer the tie-formation processes and to that effect we assume a multilevel exponential random graph model. A Bayesian approach has recently been proposed for accounting for the sampled nature of data and here we extend this procedure to handle the different types of ties of the overall network. We use two illustrative examples. The first example is an analysis of a network collated from newspaper reports on terrorist activities in Indonesia that we treat as a gold standard for appraising the veracity of the inference scheme. The second example is based on a network collected in the wild in the Eastern Democratic Republic of Congo, stemming from a project aimed at investigating the post-conflict livelihoods of combatants, civilians, and demobilised people. The key types of ties are personal and job-seeking ties of individuals, the affiliations with civic institutions and different armed groups, as well as alliance ties between armed groups. The collected data consist of a number of partially overlapping personal networks with individual affiliations. In addition to the substantive interest in modelling the ties between individuals, the inference scheme provides predictions of missing ties that can be used for studying centrality of different actors (people and armed groups) in the population.

 

【Organizer】Prof. Yasuyuki Todo

【Language】 English

We encourage your participation.

 

 

Page Top
WASEDA University

早稲田大学オフィシャルサイト(https://www.waseda.jp/fpse/winpec/)は、以下のWebブラウザでご覧いただくことを推奨いたします。

推奨環境以外でのご利用や、推奨環境であっても設定によっては、ご利用できない場合や正しく表示されない場合がございます。より快適にご利用いただくため、お使いのブラウザを最新版に更新してご覧ください。

このままご覧いただく方は、「このまま進む」ボタンをクリックし、次ページに進んでください。

このまま進む

対応ブラウザについて

閉じる