@inproceedings{kalra-etal-2026-predictions,
title = "Where Do Predictions Go? Class-Level Shifts and Carbon Cost in Stance Detection",
author = "Kalra, Nityaa and
Attema, Jisk J and
Shterionov, Dimitar",
editor = "Atwell, Katherine and
Borah, Angana and
Dementieva, Daryna and
Kreiss, Elisa and
Kotonya, Neema and
Liu, Jiarui and
Olson, Liz and
Wan, Ruyuan and
Zhao, Jieyu",
booktitle = "Proceedings of the Fifth Workshop on {NLP} for Positive Impact ({NLP}4{PI})",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.nlp4pi-1.43/",
pages = "592--604",
ISBN = "979-8-89176-495-8",
abstract = "Stance detection is widely used to summarize public discourse by measuring support, opposition, and neutrality towards societal issues. Evaluation pipelines rely on aggregate metrics such as macro-F1 that can mask how models actually distribute predictions across stance classes. We study this limitation by comparing large language and (statistical) supervised models across three stance detection datasets with different definitions of neutrality. Our evaluation considers predictive performance, class-level behavior, and environmental impact. We find that language models do not consistently outperform supervised baselines, and some models exhibit class-level priors that persist across datasets and prompting strategies. In addition, performance gains tend to saturate in the 4B{--}8B parameter range. These findings suggest that aggregate performance scores alone are insufficient for responsible stance detection. When model outputs are used to measure public opinion at scale, class-level behavior and deployment cost matter as much as the overall accuracy."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="kalra-etal-2026-predictions">
<titleInfo>
<title>Where Do Predictions Go? Class-Level Shifts and Carbon Cost in Stance Detection</title>
</titleInfo>
<name type="personal">
<namePart type="given">Nityaa</namePart>
<namePart type="family">Kalra</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jisk</namePart>
<namePart type="given">J</namePart>
<namePart type="family">Attema</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dimitar</namePart>
<namePart type="family">Shterionov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the Fifth Workshop on NLP for Positive Impact (NLP4PI)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Katherine</namePart>
<namePart type="family">Atwell</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Angana</namePart>
<namePart type="family">Borah</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Daryna</namePart>
<namePart type="family">Dementieva</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Elisa</namePart>
<namePart type="family">Kreiss</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Neema</namePart>
<namePart type="family">Kotonya</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiarui</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Liz</namePart>
<namePart type="family">Olson</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ruyuan</namePart>
<namePart type="family">Wan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jieyu</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-495-8</identifier>
</relatedItem>
<abstract>Stance detection is widely used to summarize public discourse by measuring support, opposition, and neutrality towards societal issues. Evaluation pipelines rely on aggregate metrics such as macro-F1 that can mask how models actually distribute predictions across stance classes. We study this limitation by comparing large language and (statistical) supervised models across three stance detection datasets with different definitions of neutrality. Our evaluation considers predictive performance, class-level behavior, and environmental impact. We find that language models do not consistently outperform supervised baselines, and some models exhibit class-level priors that persist across datasets and prompting strategies. In addition, performance gains tend to saturate in the 4B–8B parameter range. These findings suggest that aggregate performance scores alone are insufficient for responsible stance detection. When model outputs are used to measure public opinion at scale, class-level behavior and deployment cost matter as much as the overall accuracy.</abstract>
<identifier type="citekey">kalra-etal-2026-predictions</identifier>
<location>
<url>https://aclanthology.org/2026.nlp4pi-1.43/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>592</start>
<end>604</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Where Do Predictions Go? Class-Level Shifts and Carbon Cost in Stance Detection
%A Kalra, Nityaa
%A Attema, Jisk J.
%A Shterionov, Dimitar
%Y Atwell, Katherine
%Y Borah, Angana
%Y Dementieva, Daryna
%Y Kreiss, Elisa
%Y Kotonya, Neema
%Y Liu, Jiarui
%Y Olson, Liz
%Y Wan, Ruyuan
%Y Zhao, Jieyu
%S Proceedings of the Fifth Workshop on NLP for Positive Impact (NLP4PI)
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-495-8
%F kalra-etal-2026-predictions
%X Stance detection is widely used to summarize public discourse by measuring support, opposition, and neutrality towards societal issues. Evaluation pipelines rely on aggregate metrics such as macro-F1 that can mask how models actually distribute predictions across stance classes. We study this limitation by comparing large language and (statistical) supervised models across three stance detection datasets with different definitions of neutrality. Our evaluation considers predictive performance, class-level behavior, and environmental impact. We find that language models do not consistently outperform supervised baselines, and some models exhibit class-level priors that persist across datasets and prompting strategies. In addition, performance gains tend to saturate in the 4B–8B parameter range. These findings suggest that aggregate performance scores alone are insufficient for responsible stance detection. When model outputs are used to measure public opinion at scale, class-level behavior and deployment cost matter as much as the overall accuracy.
%U https://aclanthology.org/2026.nlp4pi-1.43/
%P 592-604
Markdown (Informal)
[Where Do Predictions Go? Class-Level Shifts and Carbon Cost in Stance Detection](https://aclanthology.org/2026.nlp4pi-1.43/) (Kalra et al., NLP4PI 2026)
ACL