Relative rank statistics for dialog analysis

Huerta, Juan M.

doi:10.3115/1613715.1613840

Cited by 6 publications

(5 citation statements)

References 12 publications

(4 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We will compare the most frequently used keywords for two corpora in order to find out how the corpora (the chats) differ. To be more precise, we will use Huerta's () relative rank difference which tells us which keywords are comparatively more frequently used in corpus c relative to

c'

. Formally, we measure the keyness of word w in corpus c relative to

c'

by generating ranks

r_{c} (w)

for all words w in corpus c according to frequency (and in descending order).…”

Section: Resultsmentioning

confidence: 99%

“…As Huerta (, p. 967) points out, the rd score “denotes some sort of percent change in rank. This also means that this function is less sensitive to small changes in frequency in the case of frequent words and to small changes in rank in case of infrequent words.” In other words, we are not concerned about cardinal measures, but ordinal measures, making the analysis distribution‐free.…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Why factors facilitating collusion may not predict cartel occurrence — experimental evidence

Fonseca¹,

Li²,

Normann³

2018

Southern Economic Journal

View full text Add to dashboard Cite

TITLEWhy factors facilitating collusion may not predict cartel occurrence -experimental evidence AUTHORS Fonseca, M; Normann, HT; Li, Y JOURNAL Southern Economic Journal DEPOSITED IN ORE September 2018This version available at http://hdl.handle.net/10871/33888 COPYRIGHT AND REUSEOpen Research Exeter makes this work available in accordance with publisher policies. A NOTE ON VERSIONSThe version presented here may differ from the published version. May 2018Abstract Factors facilitating collusion may not successfully predict cartel occurrence: when a factor predicts that collusion (explicit and tacit) becomes easier, firms might be less inclined to set up a cartel simply because tacit coordination already tends to go in hand with supra-competitive profits. We illustrate this issue with laboratory data. We run n-firm Cournot experiments with written cheap-talk communication between players and we compare them to treatments without the possibility to talk. We conduct this comparison for two, four and six firms. We find that two firms indeed find it easier to collude tacitly but that the number of firms does not significantly affect outcomes with communication. As a result, the payoff gain from communication increases with the number of firms, at a decreasing rate.JEL classification: L42, C90, C70

show abstract

c'

. Formally, we measure the keyness of word w in corpus c relative to

c'

by generating ranks

r_{c} (w)

for all words w in corpus c according to frequency (and in descending order).…”

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

Why factors facilitating collusion may not predict cartel occurrence — experimental evidence

Fonseca¹,

Li²,

Normann³

2018

Southern Economic Journal

View full text Add to dashboard Cite

show abstract

“…In Figure 9, we depict the 50 most frequent tokens in treatments NoSanction and Sanction and their relative rank differential (see Huerta, 2008;Fischer and Normann, 2019;Özkes and Hanaki, 2020), 38 with the most frequently used word having rank 1. Following Fischer and Normann (2019), we define tokens as more frequent in one treatment than in the other if the relative rank statistic is larger than or equal to one.…”

Section: Indirect Communicationmentioning

confidence: 99%

“…To the best of our knowledge, ours is the first study using LDA to understand how communication affects behavior in experimental markets. 6 The relative rank differential statistic due to Huerta (2008), which we use to analyze the communication content in different market settings, is also employed in Moellers et al (2017), Odenkirchen (2018), andFourberg (2018).…”

Section: Introductionmentioning

confidence: 99%

How Communication Makes the Difference between a Cartel and Tacit Collusion: A Machine Learning Approach

2022

View full text Add to dashboard Cite

“…Common approaches are, among others, the extraction of relevant keywords and the number of messages sent (see e.g. Huerta (2008); Moellers et al, (2017)). Besides analyzing keywords and message counts, economists have recently started to investigate whether certain types of communication (e.g.…”

Section: Introductionmentioning

confidence: 99%

How to Analyze Communication Data from Laboratory Experiments Without Being a Machine Learning Specialist

Wegener

2021

JEBS

View full text Add to dashboard Cite

Recently, the analysis of communication has gained attention in experimental research. One important question is whether certain types of communication affect decisions differently than others. In this regard, Houser & Xiao (2011) present an approach for the classification of natural language messages. The primary limitation of their approach is its limited applicability to large message datasets. Therefore, Penczynski (2019) extends the methodological instruments by applying a machine learning classifier to experimental communication data. This is accompanied by the problem of a dearth of machine learning knowledge among experimenters. Hence, this paper presents an approach that employs a publicly available machine learning text analysis application. This makes it possible to analyze larger datasets based on small training datasets classified beforehand by human evaluators. As a first step, I use primary communication data reported by Charness and Dufwenberg (2006) to generate both training and test datasets. Following this approach, I am able to substantially replicate the original classification results obtained by Charness and Dufwenberg. The second step again involves messages from Charness and Dufwenberg as training data, while I take messages from a related trust game published by Deck et al. (2013) as a test, dataset. Promisingly, I am also able to replicate the classification results obtained by the external evaluators, as reported by Deck et al. The findings suggest that machine learning can be used to analyze large message datasets, both if the artificial intelligence is trained with data from the very same experiment and if it is trained with message data from a comparable experiment.

show abstract

Relative rank statistics for dialog analysis

Cited by 6 publications

References 12 publications

Why factors facilitating collusion may not predict cartel occurrence — experimental evidence

Why factors facilitating collusion may not predict cartel occurrence — experimental evidence

How Communication Makes the Difference between a Cartel and Tacit Collusion: A Machine Learning Approach

How to Analyze Communication Data from Laboratory Experiments Without Being a Machine Learning Specialist

Contact Info

Product

Resources

About