<?xml version="1.0" encoding="UTF-8"?>

<?xml-stylesheet type="text/xsl" href="/static/oaitohtml.xsl"?>

<!--
<?xml-stylesheet type="text/xsl" href="/oaitohtml.xsl"?>
-->

<OAI-PMH xmlns="http://www.openarchives.org/OAI/2.0/" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/ http://www.openarchives.org/OAI/2.0/OAI-PMH.xsd" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
    <responseDate>2026-10-14T15:50:49Z</responseDate>
    <request verb="GetRecord" metadataPrefix="oai_dc" identifier="10.57760/sciencedb.45626" >https://www.scidb.cn/oai</request>
<GetRecord>
    <record>
    <header >
    <identifier>10.57760/sciencedb.45626</identifier>
    <datestamp>2026-08-07T10:39:12Z</datestamp>
</header>
    <metadata>
        
<oai_dc:dc xmlns:oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/ http://www.openarchives.org/OAI/2.0/oai_dc.xsd">
  <dc:date>2026-08-07</dc:date>
  <dc:title>Survey Dataset for &amp;ldquo;The Promise&amp;ndash;Threat Paradox of Conversational AI in Arts Education: How AI Identity Threat and AI Career Anxiety Shape the Intention&amp;ndash;Use Relationship</dc:title>
  <dc:identifier>doi:10.57760/sciencedb.45626</dc:identifier>
  <dc:language>en</dc:language>
  <dc:description>This dataset contains anonymized cross-sectional survey responses concerning undergraduate arts students&amp;rsquo; use of conversational artificial intelligence (C-AI). Data were collected through the Wenjuanxing online survey platform between April 11 and May 30, 2026, from undergraduate arts students at a university in Fujian Province, China. A total of 973 responses were collected. After excluding questionnaires with excessively short completion times, missing responses to core items, or evident long-string and invariant response patterns, 946 valid cases were retained, yielding a valid response rate of 97.23%.&amp;nbsp;The dataset contains 946 observations and 34 variables. The first column provides a sequential respondent identifier. Five variables record participants&amp;rsquo; major, gender, year of study, duration of C-AI use, and frequency of C-AI use. The remaining 28 measurement items assess nine constructs: task&amp;ndash;technology fit, cognitive offloading, C-AI interaction quality, AI interaction positivity, perceived AI empathy, C-AI behavioral intention, C-AI actual use, AI identity threat, and AI career anxiety.&amp;nbsp;All measurement items were assessed using a five-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The variables are numerically coded, and the dataset contains no directly identifying information or missing values among the 946 valid cases. The data can be used for replication and secondary analyses involving partial least squares structural equation modeling, fuzzy-set qualitative comparative analysis, artificial neural networks, technology adoption, AI-assisted arts education, and career-related risks. The study was approved by the Academic Ethics Committee of the School of Film and Communication, Xiamen University of Technology (Approval No. XUT-AEC-2026-01; January 17, 2026).</dc:description>
  <dc:subject>Conversational artificial intelligence; Arts education; Task–technology fit; Cognitive offloading; AI interaction positivity; Perceived AI empathy; AI identity threat; AI career anxiety; Behavioral intention; Actual use</dc:subject>
  <dc:creator>Sihang Liu</dc:creator>
  <dc:rights>PUBLIC</dc:rights>
  <dc:rights>https://creativecommons.org/licenses/by-nc-nd/4.0/</dc:rights>
  <dc:type>dataset</dc:type>
  <dc:publisher>Science Data Bank</dc:publisher>
</oai_dc:dc>

    </metadata>
</record>
</GetRecord>
</OAI-PMH>