Sample size matters when calculating Pillai scores

Sample size matters when calculating Pillai scores

Pillai scores, an output statistic from a MANOVA model, are used to measure acoustic overlap between two merging, splitting, or shifting vowels (Hay et al., 2006; Nycz & Hall-Lew, 2013, inter alia). They range from 0 (complete overlap) to 1 (complete separation). In this paper, we demonstrate th...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:The Journal of the Acoustical Society of America 2021-10, Vol.150 (4), p.A70-A70
Hauptverfasser: Stanley, Joey, Sneller, Betsy
Format: Artikel
Sprache:eng
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page A70
container_issue 4
container_start_page A70
container_title The Journal of the Acoustical Society of America
container_volume 150
creator Stanley, Joey
Sneller, Betsy
description Pillai scores, an output statistic from a MANOVA model, are used to measure acoustic overlap between two merging, splitting, or shifting vowels (Hay et al., 2006; Nycz & Hall-Lew, 2013, inter alia). They range from 0 (complete overlap) to 1 (complete separation). In this paper, we demonstrate that sample size matters in two important ways when calculating Pillai. First, relatively small datasets (
doi_str_mv 10.1121/10.0007656
format Article
fullrecord <record><control><sourceid>scitation_cross</sourceid><recordid>TN_cdi_crossref_primary_10_1121_10_0007656</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>jasa</sourcerecordid><originalsourceid>FETCH-LOGICAL-c716-1fed758d7cf4e3f22f6dab83d95a37855233aa81a8ba27ac847a1873107731763</originalsourceid><addsrcrecordid>eNp9j01LxDAQhoMoWFcv_oKclWomaT56lMUvWFBw72U2TTSStktSEf31Ztk9e5mHeXkY5iXkEtgNAIfbQsaYVlIdkQokZ7WRvDkmVUmhblqlTslZzp9llUa0FeFvOGyjozn8OjrgPLuU6feHG6nFaL8izmF8p68hRgw02ym5fE5OPMbsLg5ckPXD_Xr5VK9eHp-Xd6vaalA1eNdraXptfeOE59yrHjdG9K1EoY2UXAhEA2g2yDVa02gEowUwXYZWYkGu9mdtmnJOznfbFAZMPx2wbld2x0PZIl_v5WzDXH6exv_sPzAjUvA</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Sample size matters when calculating Pillai scores</title><source>AIP Journals Complete</source><source>Alma/SFX Local Collection</source><source>AIP Acoustical Society of America</source><creator>Stanley, Joey ; Sneller, Betsy</creator><creatorcontrib>Stanley, Joey ; Sneller, Betsy</creatorcontrib><description>Pillai scores, an output statistic from a MANOVA model, are used to measure acoustic overlap between two merging, splitting, or shifting vowels (Hay et al., 2006; Nycz &amp; Hall-Lew, 2013, inter alia). They range from 0 (complete overlap) to 1 (complete separation). In this paper, we demonstrate that sample size matters in two important ways when calculating Pillai. First, relatively small datasets (&lt;60 per sample) tend to underreport mergers. In simulated data, we draw two samples (5–100 tokens in each) from the same normal distribution to represent two vowels known to be merged. We find that a sample size of at least 60 tokens is required for at least 95% of the iterations to return a Pillai score less than the commonly used threshold of 0.1. s, Pillai scores are sensitive to the difference in sample sizes across two samples, with large differences resulting in Pillai scores that overreport mergers for samples drawn from two distinct distributions. If Pillai scores are used to measure vowel overlap, we recommend (1) using the same number of tokens across both samples and (2) using p-values from the MANOVA model to accompany them to assess whether the difference between two distributions is significant.</description><identifier>ISSN: 0001-4966</identifier><identifier>EISSN: 1520-8524</identifier><identifier>DOI: 10.1121/10.0007656</identifier><identifier>CODEN: JASMAN</identifier><language>eng</language><ispartof>The Journal of the Acoustical Society of America, 2021-10, Vol.150 (4), p.A70-A70</ispartof><rights>Acoustical Society of America</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://pubs.aip.org/jasa/article-lookup/doi/10.1121/10.0007656$$EHTML$$P50$$Gscitation$$H</linktohtml><link.rule.ids>207,208,314,777,781,791,1560,4498,27905,27906,76133</link.rule.ids></links><search><creatorcontrib>Stanley, Joey</creatorcontrib><creatorcontrib>Sneller, Betsy</creatorcontrib><title>Sample size matters when calculating Pillai scores</title><title>The Journal of the Acoustical Society of America</title><description>Pillai scores, an output statistic from a MANOVA model, are used to measure acoustic overlap between two merging, splitting, or shifting vowels (Hay et al., 2006; Nycz &amp; Hall-Lew, 2013, inter alia). They range from 0 (complete overlap) to 1 (complete separation). In this paper, we demonstrate that sample size matters in two important ways when calculating Pillai. First, relatively small datasets (&lt;60 per sample) tend to underreport mergers. In simulated data, we draw two samples (5–100 tokens in each) from the same normal distribution to represent two vowels known to be merged. We find that a sample size of at least 60 tokens is required for at least 95% of the iterations to return a Pillai score less than the commonly used threshold of 0.1. s, Pillai scores are sensitive to the difference in sample sizes across two samples, with large differences resulting in Pillai scores that overreport mergers for samples drawn from two distinct distributions. If Pillai scores are used to measure vowel overlap, we recommend (1) using the same number of tokens across both samples and (2) using p-values from the MANOVA model to accompany them to assess whether the difference between two distributions is significant.</description><issn>0001-4966</issn><issn>1520-8524</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2021</creationdate><recordtype>article</recordtype><recordid>eNp9j01LxDAQhoMoWFcv_oKclWomaT56lMUvWFBw72U2TTSStktSEf31Ztk9e5mHeXkY5iXkEtgNAIfbQsaYVlIdkQokZ7WRvDkmVUmhblqlTslZzp9llUa0FeFvOGyjozn8OjrgPLuU6feHG6nFaL8izmF8p68hRgw02ym5fE5OPMbsLg5ckPXD_Xr5VK9eHp-Xd6vaalA1eNdraXptfeOE59yrHjdG9K1EoY2UXAhEA2g2yDVa02gEowUwXYZWYkGu9mdtmnJOznfbFAZMPx2wbld2x0PZIl_v5WzDXH6exv_sPzAjUvA</recordid><startdate>202110</startdate><enddate>202110</enddate><creator>Stanley, Joey</creator><creator>Sneller, Betsy</creator><scope>AAYXX</scope><scope>CITATION</scope></search><sort><creationdate>202110</creationdate><title>Sample size matters when calculating Pillai scores</title><author>Stanley, Joey ; Sneller, Betsy</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c716-1fed758d7cf4e3f22f6dab83d95a37855233aa81a8ba27ac847a1873107731763</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2021</creationdate><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Stanley, Joey</creatorcontrib><creatorcontrib>Sneller, Betsy</creatorcontrib><collection>CrossRef</collection><jtitle>The Journal of the Acoustical Society of America</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Stanley, Joey</au><au>Sneller, Betsy</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Sample size matters when calculating Pillai scores</atitle><jtitle>The Journal of the Acoustical Society of America</jtitle><date>2021-10</date><risdate>2021</risdate><volume>150</volume><issue>4</issue><spage>A70</spage><epage>A70</epage><pages>A70-A70</pages><issn>0001-4966</issn><eissn>1520-8524</eissn><coden>JASMAN</coden><abstract>Pillai scores, an output statistic from a MANOVA model, are used to measure acoustic overlap between two merging, splitting, or shifting vowels (Hay et al., 2006; Nycz &amp; Hall-Lew, 2013, inter alia). They range from 0 (complete overlap) to 1 (complete separation). In this paper, we demonstrate that sample size matters in two important ways when calculating Pillai. First, relatively small datasets (&lt;60 per sample) tend to underreport mergers. In simulated data, we draw two samples (5–100 tokens in each) from the same normal distribution to represent two vowels known to be merged. We find that a sample size of at least 60 tokens is required for at least 95% of the iterations to return a Pillai score less than the commonly used threshold of 0.1. s, Pillai scores are sensitive to the difference in sample sizes across two samples, with large differences resulting in Pillai scores that overreport mergers for samples drawn from two distinct distributions. If Pillai scores are used to measure vowel overlap, we recommend (1) using the same number of tokens across both samples and (2) using p-values from the MANOVA model to accompany them to assess whether the difference between two distributions is significant.</abstract><doi>10.1121/10.0007656</doi><tpages>1</tpages></addata></record>
fulltext fulltext
identifier ISSN: 0001-4966
ispartof The Journal of the Acoustical Society of America, 2021-10, Vol.150 (4), p.A70-A70
issn 0001-4966
1520-8524
language eng
recordid cdi_crossref_primary_10_1121_10_0007656
source AIP Journals Complete; Alma/SFX Local Collection; AIP Acoustical Society of America
title Sample size matters when calculating Pillai scores
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-20T05%3A00%3A51IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-scitation_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Sample%20size%20matters%20when%20calculating%20Pillai%20scores&rft.jtitle=The%20Journal%20of%20the%20Acoustical%20Society%20of%20America&rft.au=Stanley,%20Joey&rft.date=2021-10&rft.volume=150&rft.issue=4&rft.spage=A70&rft.epage=A70&rft.pages=A70-A70&rft.issn=0001-4966&rft.eissn=1520-8524&rft.coden=JASMAN&rft_id=info:doi/10.1121/10.0007656&rft_dat=%3Cscitation_cross%3Ejasa%3C/scitation_cross%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true