CLUE – Method Section
3.1 Materials
For this study, an online experiment with six conditions was conducted. The independent
variables of this study included prejudice control instructions (direct vs indirect vs no control)
and non-native accent strength (slight vs moderate). To begin the experiment subjects were
given one set of prejudice control instructions with varying levels of directness. Participants
were either given implicit instructions which focused on discrimination in general, explicit
instruction that specifically addressed discrimination based on accent, or no instructions which
was used as a control condition. While the content of the instruction documents was slightly
varied in terms of content, researchers ensured that the document design was kept constant to
limit possible confounding variables, as seen in Appendix A. This Intervention design was
developed through a pre-test. After being exposed to one of the intervention conditions,
participants were asked to listen to one of two audio fragments from a marketing lecture. This
stimulus material was created and used by Hendriks et al (2016). A verbal guise technique was
used in the creation of the audio fragments, whereby the same text (see Appendix B) was
presented by different speakers with either a moderate or slight Dutch accent. Participants were
then asked to evaluate the assigned audio fragment in an online questionnaire.
3.2 Subjects
In total 194 Dutch participants took part in the study. Over half of the participants were female
(54.1%), while 44.3% were male and 1.5% identified themselves as non-binary. The Chi-
squared tests for both the accent strength conditions and prejudice control groups were
insignificant (p>.810). The age of the participants ranged between 18-30 years (M=24.60,
SD=3.44). To test the distribution of age between both the accent strength and prejudice control
groups, an independent t-test (t (191.79) =.259, p =.796) and F-test (F (2, 191)=2.57,p=.079)
were conducted, showcasing an equal distribution across all conditions. In terms of education
level, 125 (64.4%) participants were currently attending or had completed an HBO education
level whereas 69 (35.6%) subjects were enrolled in a WO programme. Again a Chi-squared
test was conducted for both the accent conditions (χ2(1)=.53,p=.465) and prejudice control
conditions (χ2(2)=.30, p=.862), demonstrating an equal distribution among all groups.
Participants were also asked to indicate their perceived English proficiency on a 7-point
Likert scale (M= 6.01, SD= .77) based on Krishna & Alhuwalia (2008). For both accentedness
(χ2(2)=1.99, p=.369) and Prejudice control (χ2(4)=1.16, p=.884), self-evaluated proficiency
, was found to be distributed equally. Simultaneously, participants were asked to take the
LexTALE test (Lemhöfer & Broersma, 2012; www.lextale.com) to evaluate the actual English
proficiency of subjects. Again, the independent t-test results yielded insignificant results for
both the accent conditions (p=.184) and prejudice control condition (p=.500). To control for
possible pre-existing attitude prejudice, participants were asked to indicate to what extent they
agree to several statements on a scale from 1-to-7, based on Ura et al. (2015). The results
indicated no significant differences for both independent variables between the groups in terms
of attitude prejudice (Accent strength: t(190.14)=1.41, p=.159; Prejudice control: F(2,
191)=.53, p=.590).
Another background variable that was determined was the participant's familiarity with
the Dutch language. Hereby participants had to indicate to what extent they agree with three
different statements on a scale from 1-to-7, based on Hendriks et al. (2021). Again, Dutch
language familiarity seemed to be distributed equally among the different conditions for each
variable (Accent strength: t(190.44)=1.03,p=.305; Prejudice control: (F(2,191)=.75,p=.476).
Additionally, participants were asked to indicate whether they have work experience and hiring
experience. In terms of work experience, it was shown that the majority of participants (84.5%)
had between 1 to 15 years of working experience while 30 participants (15.5%) had no working
experience. Nevertheless, work experience was shown to be equally distributed among accent
strength conditions (t(190.53)=.18, p=.093) and prejudice control groups (F(2,191)=.13,
p=.879). Regarding hiring experiences, the descriptive analysis revealed that 81.4% of subjects
never worked as part of a hiring panel in contrast to 18.6% of participants who had previous
experience. Lastly, participants were asked to indicate the main language(s) of their study
programme (Hendriks et al., 2021). Hereby it was shown that most students followed a study
programme in English (95.5%) and/or Dutch (93.8%), whereas only 51 students (26.3%)
followed a study programme conducted in another language (German, French, Russian).
Nevertheless, an independent t-test demonstrated that the language of the programme was
equally distributed across all accent strength groups (p=.730) while an F-test indicated similar
results for the prejudice control groups (p=.178).
3.3 Design
The study employed a 3 (Prejudice control measure: Implicit, Explicit, No Control) x 2 (Non-
native accent strength: slight, moderate) between-subject design, giving 6 possible conditions
overall.
3.1 Materials
For this study, an online experiment with six conditions was conducted. The independent
variables of this study included prejudice control instructions (direct vs indirect vs no control)
and non-native accent strength (slight vs moderate). To begin the experiment subjects were
given one set of prejudice control instructions with varying levels of directness. Participants
were either given implicit instructions which focused on discrimination in general, explicit
instruction that specifically addressed discrimination based on accent, or no instructions which
was used as a control condition. While the content of the instruction documents was slightly
varied in terms of content, researchers ensured that the document design was kept constant to
limit possible confounding variables, as seen in Appendix A. This Intervention design was
developed through a pre-test. After being exposed to one of the intervention conditions,
participants were asked to listen to one of two audio fragments from a marketing lecture. This
stimulus material was created and used by Hendriks et al (2016). A verbal guise technique was
used in the creation of the audio fragments, whereby the same text (see Appendix B) was
presented by different speakers with either a moderate or slight Dutch accent. Participants were
then asked to evaluate the assigned audio fragment in an online questionnaire.
3.2 Subjects
In total 194 Dutch participants took part in the study. Over half of the participants were female
(54.1%), while 44.3% were male and 1.5% identified themselves as non-binary. The Chi-
squared tests for both the accent strength conditions and prejudice control groups were
insignificant (p>.810). The age of the participants ranged between 18-30 years (M=24.60,
SD=3.44). To test the distribution of age between both the accent strength and prejudice control
groups, an independent t-test (t (191.79) =.259, p =.796) and F-test (F (2, 191)=2.57,p=.079)
were conducted, showcasing an equal distribution across all conditions. In terms of education
level, 125 (64.4%) participants were currently attending or had completed an HBO education
level whereas 69 (35.6%) subjects were enrolled in a WO programme. Again a Chi-squared
test was conducted for both the accent conditions (χ2(1)=.53,p=.465) and prejudice control
conditions (χ2(2)=.30, p=.862), demonstrating an equal distribution among all groups.
Participants were also asked to indicate their perceived English proficiency on a 7-point
Likert scale (M= 6.01, SD= .77) based on Krishna & Alhuwalia (2008). For both accentedness
(χ2(2)=1.99, p=.369) and Prejudice control (χ2(4)=1.16, p=.884), self-evaluated proficiency
, was found to be distributed equally. Simultaneously, participants were asked to take the
LexTALE test (Lemhöfer & Broersma, 2012; www.lextale.com) to evaluate the actual English
proficiency of subjects. Again, the independent t-test results yielded insignificant results for
both the accent conditions (p=.184) and prejudice control condition (p=.500). To control for
possible pre-existing attitude prejudice, participants were asked to indicate to what extent they
agree to several statements on a scale from 1-to-7, based on Ura et al. (2015). The results
indicated no significant differences for both independent variables between the groups in terms
of attitude prejudice (Accent strength: t(190.14)=1.41, p=.159; Prejudice control: F(2,
191)=.53, p=.590).
Another background variable that was determined was the participant's familiarity with
the Dutch language. Hereby participants had to indicate to what extent they agree with three
different statements on a scale from 1-to-7, based on Hendriks et al. (2021). Again, Dutch
language familiarity seemed to be distributed equally among the different conditions for each
variable (Accent strength: t(190.44)=1.03,p=.305; Prejudice control: (F(2,191)=.75,p=.476).
Additionally, participants were asked to indicate whether they have work experience and hiring
experience. In terms of work experience, it was shown that the majority of participants (84.5%)
had between 1 to 15 years of working experience while 30 participants (15.5%) had no working
experience. Nevertheless, work experience was shown to be equally distributed among accent
strength conditions (t(190.53)=.18, p=.093) and prejudice control groups (F(2,191)=.13,
p=.879). Regarding hiring experiences, the descriptive analysis revealed that 81.4% of subjects
never worked as part of a hiring panel in contrast to 18.6% of participants who had previous
experience. Lastly, participants were asked to indicate the main language(s) of their study
programme (Hendriks et al., 2021). Hereby it was shown that most students followed a study
programme in English (95.5%) and/or Dutch (93.8%), whereas only 51 students (26.3%)
followed a study programme conducted in another language (German, French, Russian).
Nevertheless, an independent t-test demonstrated that the language of the programme was
equally distributed across all accent strength groups (p=.730) while an F-test indicated similar
results for the prejudice control groups (p=.178).
3.3 Design
The study employed a 3 (Prejudice control measure: Implicit, Explicit, No Control) x 2 (Non-
native accent strength: slight, moderate) between-subject design, giving 6 possible conditions
overall.
