Students’ interaction anxiety and social phobia in interprofessional education in Hong Kong: mapping a new research direction

Abstract

Background

Interprofessional education (IPE) has been promoted as a breakthrough in healthcare because of the impact when professionals work as a team. However, despite its inception dating back to the 1960s, its science has taken a long time to advance. There is a need to theorize IPE to cultivate creative insights for a nuanced understanding of IPE. This study aims to propose a research agenda on social interaction by understanding the measurement scales used and guiding researchers to contribute to the discussion of social processes in IPE.

Method

This quantitative research was undertaken in a cross-institutional IPE involving 925 healthcare students (Medicine, Nursing, Social Work, Chinese Medicine, Pharmacy, Speech Language Pathology, Clinical Psychology, Food and Nutritional Science and Physiotherapy) from two institutions in Hong Kong. Participants completed the Social Interaction Anxiety Scale (SIAS-6) and Social Phobia Scale (SPS-6). We applied a construct validation approach: within-network and between-network validation. We performed confirmatory factors analysis, t-test, analysis of variance and regression analysis.

Results

CFA results indicated that current data fit the a priori model providing support to within-network validity [RMSEA=.08, NFI=.959, CFI=.965, IFI=.965, TLI=.955]. The criteria for acceptable fit were met. The scales were invariant between genders, across year levels and disciplines. Results indicated that social interaction anxiety and social phobia negatively predicted behavioural engagement (F = 25.093, p<.001, R²=.065) and positively predicted behavioural disaffection (F = 22.169, p<.001, R²=.057) to IPE, suggesting between-network validity.

Conclusions

Our data provided support for the validity of the scales when used among healthcare students in Hong Kong. SIAS-6 and SPS-6 have sound psychometric properties based on students’ data in Hong Kong. We identified quantitative, qualitative and mixed methods research designs to guide researchers in getting involved in the discussion of students’ social interactions in IPE.

Key Messages

• The Social Anxiety Scale (SIAS-6) and Social Phobia Scale (SPS-6) scales have sound psychometric properties based on the large-scale healthcare students’ data in IPE in Hong Kong.

• Social interaction anxiety and social phobia negatively predicted students’ behavioural engagement with IPE and positively predicted behavioural disaffection. The scales are invariant in terms of gender, year level and discipline.

• Quantitative, qualitative and mixed methods studies are proposed to aid researchers to contribute in healthcare education literature using the SIAS-6 and SPS-6.

Keywords:

• Construct validation
• interprofessional education
• social interaction anxiety

Interprofessional education (IPE) is a team-based instructional model in healthcare education that trains students from diverse healthcare expertise to develop interprofessional teamwork by learning about, from and with [1] each other as a means of optimizing integrated and seamless healthcare. Developing collaborative interpersonal relationships among students from diverse fields of expertise becomes an important means of achieving the goal of cultivating knowledge co-creation and interprofessional teamwork for the safe delivery of patient care. This assertion is vital because developing members’ smooth social relationships during the early forming stage is construed as a mechanism to achieve desired goals during the team’s performing stage [2]. However, earlier reports show that approximately 10% of sample undergraduate students in the United Kingdom reported marked to severe social anxiety [3]. In Hong Kong, of 1,119 undergraduate students surveyed, more than 54.4% reported some degree of anxiety symptoms [4], which, if not attended to, might undermine their optimal engagement and achievement in school.

Social anxiety is construed as a ‘fear of social situations in which embarrassment may occur (e.g. making conversation, meeting strangers, dating) or there is a risk of being negatively evaluated by others’ (e.g. seen as stupid, weak, or anxious [5]). Phobia, on the other hand, is an intense and irrational fear of certain objects and situations. Studies have shown that student social interaction anxiety has adverse effects on various outcomes: academic achievement [6,7], school completion [8,9] and well-being [10]. In IPE, social phobia is considered a barrier to promoting teamwork and collaboration among students [11,12]. Furthermore, reports indicate that students with Asian heritage tend to experience higher levels of social anxiety compared with their European and American counterparts [13,14].

The Social Interaction Anxiety Scale (SIAS-6) and Social Phobia Scale (SPS-6) [15], which originated from 20-item SIAS and SPS [16], are companion scales the researchers and clinicians have used in the diagnosis and evaluation of the treatment of social fears/anxiety. Because of the importance of these scales, their psychometric properties using exploratory or confirmatory factor analysis have been examined in clinical and non-clinical student samples, including students from Australia [16], Brazil [17], USA [18], Germany [19], Sweden [20], United Kingdom [21], Holland [22], Japan [23] and China [24], among others.

Across clinical and non-clinical settings, SIAS-6 and SPS-6 have been useful to both researchers and clinicians. For example, in clinical settings, the scales are used to examine symptom reduction for participants aged 18–75 after treatment [25,26]. For non-clinical settings, the scales are used as diagnostic tools to measure intervention outcomes such as engagement [27] or kindness [28], among others. In medical education, these scales have been used to investigate medical students’ social anxiety and their link with communication attitudes [29]. To our knowledge, there has been no study that has explored the validity of these scales in Hong Kong in the context of inter-institutional IPE – a programme that especially leverages students’ unique and divergent professional expertise as important characteristics to develop interprofessional expertise. Furthermore, IPE is a social performance context where students will need to communicate and interact with peers from different disciplines that they are unfamiliar with; thereby, social interaction anxiety could be even more prevalent in this particular learning context.

Noting these research gaps, we aim to contribute to IPE literature in various ways. Practically, we hope to contribute to the identification of psychometrically valid scales to aid researchers in conceptualizing important studies that will advance our understanding of IPE implementation design which takes into account students’ different traits or states. Conceptually, we aim to distinguish the nomological network of social interaction anxiety with theoretically-relevant constructs (e.g. behaviour engagement and disengagement). Using the scales, we hope to understand potential gender, year level and discipline level social interaction anxiety and social phobia that may arise in a large-scale inter-institutional IPE. We conceptualized research topics using SIAS-6 and SPS-6 to actively engage in the conversation of social interaction or sense of relatedness as a meaningful research focus in IPE. Validating these scales in IPE involving a sample of Hong Kong students enables us to prepare for the future, where there are available psychometrically sound tools to screen students who may not be able to participate fully in IPE and provide remedial support to ensure optimal learning.

IPE in one government-subsidized university in Hong Kong, where this study was conducted, is a large-scale cross-institutional and cross-country interprofessional training programme [30]. IPE uses a spiral three-tier approach that is captured in 3 C’s: Tier 1 (Connect), Tier 2 (Capacitate) and Tier 3 (Collaborate). This progressive design is consistent with assumptions of Bloom’s Taxonomy [31] and Miller’s pyramid [32]. Using team-based learning [33] and case-based learning [34], this programme aims to facilitate the development of essential collaborative practice competencies and team-based competencies through simulation cases (e.g. COVID-19 infection control and management) in a small group. While the lecture was a common teaching strategy in IPE implementation (e.g. Japan [35], Philippines [36]), simulation and small-group discussion have been used in Hong Kong.

There are a number of reasons to support the conjecture that these scales are important and valuable in IPE. Considering that healthcare educators are interested in supporting students from diverse backgrounds to work together to be better collaborators, understanding potential social interaction issues helps spark a discussion for mapping a theoretically-driven research direction in promoting interprofessional collaboration. This discussion can lead to the identification of best practices in the IPE community of practice in general and managing students who are not comfortable socially in particular. Furthermore, self-determination theory [37] posits that students’ autonomous or intrinsic motivation derived from the satisfaction of one’s need for autonomy, competence and relatedness is linked with engagement, achievement and well-being. Validating scales on social interaction anxiety and phobia is a step closer to promoting a sense of relatedness in IPE, which will have desirable effects on their engagement and well-being.

The present study

We set out three objectives for this study involving healthcare students in Hong Kong. The first objective was to examine the within-network construct validity of SIAS-6 and SPS-6 to test whether the two-factor solution of SIAS-6 and SPS-6 applies to Hong Kong healthcare students’ sample. Since these scales are used together, examining their construct validity through a statistical approach provides us with the assurance that they measure the construct adequately. The second objective was to examine the between-network validity of the scales by exploring the relationships of social interaction anxiety and social phobia with theoretically relevant constructs (e.g. behavioural engagement and behavioural disaffection to IPE). Previous conceptualizations identified behaviour engagement and behaviour disaffection as outcomes of several educational variables: learning goals [38], self-determined motivation [39–41] and peer victimization [42]. Given that empirical data indicated that anxious individuals are perceived to avoid social interactions [10,43], in this study, we hypothesized that behavioural engagement to IPE would be negatively associated with social interaction anxiety and social phobia, while behavioural disaffection to IPE would be positively correlated with them (H₁). Our third objective was to establish the scale invariance or equivalence of the constructs by distinguishing if SIAS-6 and SPS-6 work well between genders (e.g. male and female), across year levels (e.g. Years 3, 4) and disciplines (e.g. MBBS vs Nursing). Previous studies examined scale invariance to examine scale equivalence [44,45]. We end by proposing research areas involving healthcare students in IPE using the newly validated scales to help map a new direction for research in IPE.

Methods

Participants and procedures

We utilized data from 925 (85.8%) students from a total of 1,078 health and social care students enrolled in a ‘Ten-day online IPE program’ [46]. Developed during the COVID-19 pandemic, this online learning program replaced the conventional face-to-face model. The participants were from three higher education institutions (HEIs) in Hong Kong, of which one was government-subsidized, while the two were self-financed HEIs. They were from nine disciplines: Medicine (n = 160, year 4), Nursing (n = 289, years 2 & 4), Social Work (n = 128, years 4 & 5), Chinese Medicine (n = 47, year 3), Pharmacy (n = 89, years 2 & 3), Speech Language Pathology (n = 52, years 3 & 5), Clinical Psychology (n = 12, Masters level), Food and Nutritional Science (n = 51, years 3 & 4) and Physiotherapy (n = 55, years 2 & 3). Some students failed to indicate their discipline (n = 42). The average age of the participants was 21.96 years old (SD = 2.89), with 34.6% of males and 65.4% of females (Table 1). Grouped into teams of five to seven members following team-based learning principles [47], these participants completed team-based tasks which consisted of four parts: preparation, readiness assurance, application exercise and enrichment activity (Figure 1). Students’ participation in this research was entirely voluntary. We sought ethics approval in conducting this research from Human Research Ethics Committee (EA210423). Pilot testing was conducted involving an earlier cohort of IPE students. Using Qualtrics, data collection took place at Time 2 (post-test) which was the 10th day of IPE implementation. We explained to the students that their participation would not affect their course grades.

Figure 1. IPE simulation model; MCQ: multiple choice questions; iRAT: individual readiness assurance test; tRAT: team readiness assurance test; RAT: readiness assurance test.

Table 1. Demographics of participants.

	N	Percentage (%)
Gender
Male	320	34.7
Female	605	65.4
Total	925
Year level
Second year	219	23.7
Third year	156	16.9
Fourth year	321	34.7
Fifth year (post graduate)	125	13.5
Missing	104	11.2
Total	925
Discipline
Medicine	160	17.3
Nursing	289	31.2
Social work	128	13.8
Chinese medicine	47	5.1
Pharmacy	89	9.6
Speech language pathology	52	5.6
Clinical psychology	12	1.3
Food and nutritional science	51	5.5
Physiotherapy	55	5.9
Missing	42	4.6
Total	925

Measures

Social Interaction Anxiety Scale (SIAS) and the Social Phobia Scale (SPS). The short versions developed by Peters et al. [15] are composed of 12 items from the original 20-item SIAS and SPS scales [16]. The SIAS-6 is used to assess general anxiety in terms of initiating and maintaining social connections (6 items, e.g. ‘I have difficulty making eye contact with others’; α=.87). The SPS-6 scale is designed to assess anxiety when performing various tasks while being observed by others (6 items, e.g. ‘I can feel conspicuous standing in a line’; α=.89). The response format ranges from 0 (not at all characteristic or true of me) to 4 (extremely characteristic or true of me).

Behavioural engagement and behavioural disaffection [48]. We used the two subscales of the Engagement versus Disaffection in Learning Questionnaire – Student Report to measure learners’ behaviour engagement (5 items, e.g. ‘I try hard to do well in school/IPE’ ; α=.90) and behaviour disaffection (5 items, e.g. ‘I don’t try very hard at school/IPE’; α=.83). The scale is answerable from 0 (not at all true) to 4 (very true). This scale has been used in previous studies in healthcare education [49]. We used the English version of all the scales used in this investigation, as English was the medium of instruction in the institutions where the participants were enrolled.

Approach to scale validation

This study adopted a construct validation approach [50,51] with two approaches. The first approach is within-network construct validation which pertains to the factor structure and factor correlation matrix examination. The second approach is between-network construct validation which examines relationships between the scales (or their factors) with external theoretically relevant constructs. We took advantage of the strengths of both approaches which had been used in previous studies [52,53] and applied them in this study.

Data analysis

We first examined the normality of data by examining the skewness and kurtosis. To address our first objective, we examined the within-network construct validity (internal structure) of SIAS-6 and SPS-6 involving healthcare students. We performed confirmatory factor analysis (CFA) to examine whether the scales were appropriate for use with our current sample and whether the original factor structures of the scales worked well with our new sample [54]. We used a comprehensive array of goodness-of-fit indices to indicate the model fit: comparative fit index (CFI), Tucker-Lewis index (TLI), goodness of fit index (GFI), incremental fit index (IFI), Bentler-Bonett normed fit index (NFI), the root mean square error of approximation (RMSEA) and standardized root mean square residual (SRMR). As a rule of thumb, for an acceptable model fit, all indexes need to be greater than or equal to 0.90, except for RMSEA and SRMR, which must be less than or equal to 0.08 [55–57]. Also, we conducted independent sample t-tests to examine gender differences in SIAS-6 and SPS-6 and one-way analysis of variance to clarify year-level and discipline differences.

To achieve our second objective on between-network validity, we performed correlation analysis and linear multiple regression to examine the relationship of social interaction anxiety and social phobia with behavioural engagement and disaffection to IPE. We used Analysis of Moment Structures (AMOS Version 23) [58] for the factor analysis and Statistical Package for Social Sciences (SPSS Version 26) for t-test, ANOVA, correlation and regression analysis.

To achieve our third objective, which was to clarify if the scales work well between genders and across year levels, we used multi-group CFA to check for measurement invariance. Previous work on SIAS and SPS indicated that these scales were invariant. Analytical comparisons between nested models were used. Using the configural model, which calculates measurement and latent construct parameters across groups, we first investigated configural equivalence (see [58]). By requiring that both the factor structure and factor loadings be the same across groups, we checked for measurement equivalence by the metric model [59]. Thirdly, we constrained that the factor structure, loadings, variance and covariance be the same across groups to evaluate further measurement equivalence by the scalar model [59,60].

Results

Nine hundred twenty-five health and social care students who responded to the scales obtained an average score of 5.29 (SD = 5.26) in social interaction anxiety and 5.15 (SD = 5.70) in social phobia, respectively. The SDs are big, suggesting dispersion of the data in relation to the mean. The demographic data are presented in Table 1.

Within-network study

Finney and DiStefano suggested that skewness and kurtosis absolute values greater than 2 and 7, respectively, indicate the absence of univariate normality [61]. Our data had a skewness of 1.328 to 1.942 and a kurtosis of 1.671 to 5.163, indicating normality and therefore meeting the assumptions critical in conducting CFA.

To determine the best model for our data, we compared the two-factor solution with an alternative one-factor model. Table 2 shows the fit indices for the one- and two-factor solution SIAS-6 and SPS-6 models. Our data from 925 healthcare students indicated that the SIAS-6 and SPS-6 models had an adequate fit for the data. RMSEA and SRMR were less than the suggested threshold values of .08. Using maximum likelihood estimation, CFI, NFI, IFI and TLI values were higher than .95 (Table 2), suggesting that the two-factor model (Figure 2) had a better fit to the current data involving IPE students in Hong Kong. Together, these results provide evidence for the model’s psychometric validity in the IPE setting.

Figure 2. The measurement model of SIAS-6 and SPA-6.

Table 2. Goodness-of-fit statistics.

	Chi square	df	p	CFI	NFI	GFI	IFI	TLI	RMSEA	SRMR
Two-factor-12 items	396.3	52	.000	.965	.959	.931	.965	.955	.080	.029
One-factor-12 items	790.918	54	.000	.924	.919	.852	.924	.907	.122	.040
Criteria for good fit [55–57]				>.90	>.90	.90	>.90	>.90	.08 or lower	.08 or lower

Notes: CFI: comparative fit index; NFI: normed fit index; GFI: the goodness of fit index; IFI: incremental fit index; TLI: Tucker Lewis Index; RMSEA: root mean square error of approximation, SRMR: standardized root mean square residual.

Across genders, year levels and disciplines, the two-factor solution was examined for invariance. We put more restrictions progressively in conducting multi-group CFA. We calculated the reduction in CFI among the constrained models (i.e. metric model and scalar model) and the configural model. Cheung and Rensvold [60] claim that an increase in CFI of less than .01 indicates that the constrained parameters are invariant across groups. Results suggest that the scales were invariant across genders, between year levels and across disciplines (Tables 3, 4 and 5).

Table 3. Invariance test of the two-factor solution across genders (n = 925).

Model	Chi square	df	RMSEA	SRMR	TLI	CFI	ΔCFI
Configural model	554.173	104	.068	.0333	.942	.954	–
Metric model	595.528	114	.068	.0344	.943	.951	.003
Scalar model	598.416	117	.067	.0361	.945	.951	.000

Notes: df: degrees of freedom; RMSEA: root mean square error of approximation; SRMR: standardized root mean square residual; TLI: Tucker Lewis Index; CFI: comparative fit index; ΔCFI: difference between the two model CFIs.

Table 4. Invariance test of the two-factor solution across year levels (n = 821).

Model	Chi square	df	RMSEA	SRMR	TLI	CFI	ΔCFI
Configural model	914.067	234	.60	.0625	.914	.924	–
Metric model	941.148	254	.58	.0594	.920	.923	.001
Scalar model	966.202	260	.58	.0522	.919	.921	.002

Notes: df: degrees of freedom; RMSEA: root mean square error of approximation; SRMR: standardized root mean square residual; TLI: Tucker Lewis Index; CFI: comparative fit index; ΔCFI: difference between the two model CFIs.

Table 5. Invariance test of the two-factor solution across disciplines (n = 882).

Model	Chi square	df	RMSEA	SRMR	TLI	CFI	ΔCFI
Configural model	707.550	182	.071	.0610	.905	.912	–
Metric model	719.037	192	.069	.0609	.909	.912	.000
Scalar model	730.243	195	.069	.0567	.909	.911	.001

Notes: df: degrees of freedom; RMSEA: root mean square error of approximation; SRMR: standardized root mean square residual; TLI: Tucker Lewis Index; CFI: comparative fit index; ΔCFI: difference between the two model CFIs.

We also conducted a comparison of means to better understand the performance of scales. We specifically determined if the SIAS-6 and SPS-6 could distinguish between participants’ scores when they were grouped by gender, year level and discipline. There was no significant gender difference (Table 6) in social interaction anxiety (t(923)=3.64, p=.06), while there was a significant gender-related difference in social phobia (t(923)=4.21, p=.04, d=.11, 95% CI [4.90, 6.20]). The effect size was small, with a Cohen’s d of 0.11. There was no significant year-level difference (Table 7) in social interaction anxiety (F(3,820)=2.31, p=.07) and social phobia (F(3,820)=2.20, p=.09). There were discipline-level differences in social interaction anxiety (F(8,881)=3.53, p=.000, η²=.03, 95% CI [4.98, 5.68]) and social phobia (F(8,881)=3.18, p=.001, η²=.03, 95% CI [4.81, 5.56]; Figure 3). Using Scheffé’s post hoc comparison, the results showed that the social interaction anxiety of social work students was significantly lower than that of nursing students. Chinese Medicine students (n = 47) reported the highest mean score in social interaction anxiety (M = 7.02) and social phobia (M = 7.34), followed by Nursing students (n = 289) with an average of 6.17 in social interaction anxiety and 5.95 in social phobia, respectively (Table 8).

Figure 3. Social interaction anxiety and social phobia among different disciplines.

Table 6. Gender differences in social interaction anxiety and social phobia as measured by SIAS-6 and SPS-6 (n = 925).

	Female (n = 605)		Male (n = 320)
Variable	Mean (SD)	95% CI	Mean (SD)	95% CI	t (923)	p	Cohen’s d
Social interaction anxiety	5.12 (5.3)	4.70–5.54	5.60 (5.50)	5.00–6.20	3.64	.06	–
Social phobia	4.93 (5.54)	4.49–5.37	5.55 (5.97)	4.90–6.20	4.21	.04	0.11

Notes: Score range from 0 to 24; CI: confidence interval.

Table 7. Year level differences in social interaction anxiety and social phobia as measured by SIAS-6 and SPS-6 (n = 821).

Variable	Year level	Mean (SD)	F (3,820)	p
Social interaction anxiety	2 (n = 219)	5.72 (5.11)	2.34	.07
	3 (n = 156)	5.32 (5.89)
	4 (n = 321)	5.18 (5.50)
	5 (n = 125)	4.28 (3.90)
Social phobia	2 (n = 219)	5.60 (5.34)	2.20	.09
	3 (n = 156)	5.60 (6.44)
	4 (n = 321)	5.16 (5.95)
	5 (n = 125)	4.10 (4.32)

Notes: Years 2, 3 and 4: undergraduate students; year 5: postgraduate students.

Table 8. Discipline-level differences in social interaction anxiety and social phobia as measured by SIAS-6 and SPS-6 (n = 882).

Variable	Discipline	Mean (SD)	95% CI	F (8,881)	p	η^2	Scheffé post-hoc comparison
Social interaction anxiety	Medicine (n = 160)	5.41 (5.24)	4.59–6.23	3.53	<.001	.03	Nursing > Social Work
	Nursing (n = 289)	6.17 (5.57)	5.53–6.82
	Social work (n = 128)	3.54 (3.82)	3.09–4.43
	Chinese medicine (n = 47)	7.02 (7.64)	4.78–9.27
	Pharmacy (n = 89)	4.85 (4.47)	3.91–5.80
	Speech language pathology (n = 52)	4.86 (4.69)	3.56–6.17
	Clinical psychology (n = 12)	3.50 (3.47)	1.29–5.71
	Food and nutritional science (n = 51)	5.11 (4.87)	3.75–6.49
	Physiotherapy(n = 55)	4.66 (4.90)	3.33–6.01
	Total(n = 883)	5.33 (5.23)	4.98–5.68
Social phobia	Medicine (n = 160)	4.63 (5.67)	3.75–5.52	3.18	.001	.03	–
	Nursing (n = 289)	5.95 (5.73)	5.29–6.62
	Social work (n = 128)	3.98 (4.66)	3.16–4.81
	Chinese medicine (n = 47)	7.34 (7.89)	5.02–9.66
	Pharmacy (n = 89)	4.66 (5.11)	3.59–5.74
	Speech language pathology (n = 52)	4.30 (4.79)	2.97–5.64
	Clinical psychology (n = 12)	2.33 (2.10)	1.00–3.67
	Food and nutritional science (n = 51)	5.64 (5.70)	4.04–7.25
	Physiotherapy (n = 55)	5.51 (6.20)	3.93–7.21
	Total (n = 883)	5.18 (5.66)	4.81–5.56

Note: CI: confidence interval.

Between-network study

Table 9 shows the relationships between the research variables, including additional variables utilized to confirm the external validity of SIAS-6 and SPS-6. Social interaction anxiety and social phobia were both negatively correlated with behaviour engagement (r= −.26 and r= −.24, p < .001) and positively correlated with behaviour disaffection (r= .25 and r= .22, p < .001).

Table 9. Correlation among study variables (n = 696).

Variable	1	2	3	4
1. Social interaction anxiety	–	.86***	−.26***	.25***
2. Social phobia		–	−.24***	.22***
3. Behavioural engagement			–	−.31***
4. Behavioural disaffection				–
Mean	5.45	5.26	2.14	1.1
SD	5.44	5.82	.56	.65
Cronbach’s alpha	.87	.89	.90	.83
Score range	0–24	0–24	0–4	0–4

***p < .001.

To understand SIAS-6 and SPS-6, we conducted a multiple regression to see if learners’ social interaction anxiety and social phobia predict their behaviour engagement and disaffection in the IPE learning activities. Table 10 shows that two predictors displayed a significant amount of variance in behavioural engagement (F(2,695)=25.093, p < .001, R²=.065) and behavioural disaffection (F(2,695)=22.169, p < .001, R²=.057). It was found that social interaction anxiety was a negative predictor of behaviour engagement (β = −.208, p < .01) and a positive predictor of behavioural disaffection (β = .214, p < .01). However, the effect of social phobia on behavioural engagement and disaffection was not established.

Table 10. Multiple regression analyses with social interaction anxiety and social phobia as predictors of behavioural engagement and disaffection (n = 696).

Variable	Behavioural engagement			Behavioural disaffection
	B	SE	β	B	SE	β
Social interaction anxiety	−.022	.008	−.208**	.026	.009	.214**
Social phobia	−.006	.007	−.058	.004	.008	.035
F	25.093***			22.169***
R^2	.065			.057

Note: **<.01, *** <.001.

Discussion

This study aimed to examine the construct validity of Social Interaction Anxiety (SIAS-6) and Social Phobia Scales (SPS-6) in interprofessional education - a learning context that aims to tear down disciplinary silos by training students to be better collaborators to ensure patient safety. This study was in line with our goal of examining how psychological tools may be adapted in IPE to equip researchers in conceptualizing meaningful research agenda. Generally, our data from 925 healthcare students suggest that SIAS-6 and SPS-6 appear to be psychometrically sound and are invariant between gender, across year levels and disciplines.

We found that the two-factor solution had an adequate fit to the current data implying that the measurement model is plausible when applied to the current sample. Specifically, CFI, NFI, GFI, IFI and TLI have values exceeding .90, while RMSEA and SRMR have values less than the suggested threshold values of .08. Using a comparison model, we posited a unidimensional model where a single factor underpinned the 12 items. This action could be our response to the call of Peters et al. [15,p.74] to ‘determine whether the two short scales can be combined into a single score’. Our results indicated that while many of the fit indices were acceptable, the GFI (.852) and RMSEA (.123) were less than adequate (Table 2). These establish the within-network construct validity of the scales. The internal consistencies of SIAS-6 (α=.87) and SPS-6 (α=.89) were within the acceptable range. CFA provided evidence for the applicability of the scales involving healthcare students in Hong Kong from medicine, nursing, social work, Chinese medicine, pharmacy, speech language pathology, clinical psychology, food and nutritional science and physiotherapy. The results provided psychometric evidence for the two-factor SIAS-6 and SPS-6. We tested the scales for invariance across gender, year level and discipline. Following a standard procedure, we imposed increasing constraints on the multi-group CFA (Tables 3 and 4). Consistent with Cheung and Rensvold’s criteria of change in CFI less than .01 as evidence of invariance [47], our results indicate that the SIAS-6 and SPS-6 use in IPE is invariant for females and males, and for year 2 undergraduate to master’s level IPE students in Hong Kong, and across disciplines.

We used the scales to understand whether they could differentiate our non-clinical sample using gender, year level and discipline as grouping variables. In our current sample, students obtained an average of 5.29 in social interaction anxiety and 5.15 in social phobia, which is similar to Lowe and Harris’s study [62]. There was no significant gender difference in terms of social interaction anxiety. As regards social phobia, t-test results indicated that males had significantly higher levels of social phobia than females (Table 6). This result countered earlier reports that phobia is more prevalent in women [63]. We would like to acknowledge, however, the potential sampling size bias in our data set where the female sample (n = 605) was significantly larger than the male sample (n = 320). There was no year-level difference in our data set. However, there was a significant discipline level difference where Nursing students reported higher levels of social interaction anxiety than social work students (Table 8).

To establish the between-network construct validity of SIAS-6 and SPS-6 using the data from 925 non-clinical students, we used the theoretical link between social interaction anxiety and social phobia with behavioural engagement and behavioural disaffection to IPE [40]. To do this, we performed correlation and regression where results indicated a significant correlation between interaction anxiety/phobia and behavioural engagement and disaffection to IPE. Our hypotheses were supported by the finding that social interaction anxiety negatively predicted behavioural engagement and positively predicted behavioural disaffection to IPE. This finding suggests that social interaction anxiety was detrimental to students’ subsequent engagement, which aligns well with the earlier study of Wu et al. [43]. Previous research in education reported that goal [38] and motivation [40,41] were some antecedents of engagement and disaffection. In addition, our current study indicated that social interaction anxiety predicted engagement and disaffection in IPE activities.

Limitations of the study

In spite of the promising results to support the applicability of SIAS-6 and SPS-6 in IPE, we also noted several limitations of this study. First, even if we had a large number of participants from three HEIs in Hong Kong, all of these students volunteered to participate, which might introduce bias to their self-report responses. Second, there were big sample size differences between gender and disciplines, which might have affected the statistical results. Third, the variables identified to establish the between-network construct validity were limited. Fourth, while social interaction anxiety and social phobia predicted behavioural engagement and behavioural disaffection, the proportion of variance explained was low. Fifth, there was a wide variation in students’ number from nine disciplines. Finally, similar to the findings of Peters et al. [15], there was a high correlation between SIAS-6 and SPS-6. We encourage future examination of these limitations.

To draw traction to the topic of social interactions in IPE, we identified several research topics that may be initiated using SIAS-6 and SPS-6 in IPE (Table 11). We hope that these topics will inspire all researchers to contribute to the advancement of IPE research scholarship in healthcare.

Table 11. List of potential research topics using SIAS-6 and SPA-6 in IPE.

Research topic	Research approach
1. Examination of effects of intervention e.g. sentence-completion priming experiment, to help students who are high in social interaction anxiety	Quantitative
2. Identification of approaches or strategies in the management of students with social interaction anxiety e.g. team composition with the strategic assignment of students high in social interaction anxiety	Qualitative
3. Identification of factors which trigger interaction anxiety in IPE	Quantitative
4. Distinguishing the interactional patterns of students high in interactional anxiety	Qualitative
5. Establishing the nomological network of social interaction anxiety in the large classroom e.g. relationship of self-esteem and interaction anxiety, collaboration outcomes	Quantitative
6. Identifying the cultural dimension of social interactional anxiety in IPE	Mixed methods
7. Examining the effects of team-based or cooperative online learning IPE in reducing social anxiety	Quantitative
8. Distinguishing the effects of promoting interdependence and teamwork to reduce social anxiety and its effects on collaboration outcomes	Quantitative

Conclusion

We end by reiterating our motivation to contribute to developing meaningful studies in IPE through the validation of scales. Our study showed evidence for the construct validity of SIAS-6 and SPS-6 in the context of IPE in Hong Kong. The two-factor solution had an adequate fit for our data obtained from a large number of healthcare students in Hong Kong. This factor structure was invariant across gender, year level and across disciplines. We therefore propose using the scales to develop meaningful research agenda in IPE to better represent Chinese students in IPE literature.

Ethics approval

Data collection was done with the approval of the Human Research Ethics Committee for Non-Clinical Faculties (HRECNCF, reference number EA210433) of the University of Hong Kong.

Author contributions

Fraide A. Ganotice, Jr.: Drafting the paper, Literature Review, Visualization, Conceptualization, Methodology and Writing. Xiaoai Shen: Programme Implementation, Visualization, Data Curation, Data analysis and Writing. Jacqueline Kwan Yuk Yuen, Yin Man Amy Chow, Anita M.Y. Wong, Karen M.K. Chan, Binbin Zheng, Linda Chan, Pauline Yeung Ng, Siu Chung Leung, Elizabeth Barrett, Hoi Yan Celia Chan, Wing Nga Chan, Kit Wa Sherry Chan, Siu Ling Polly Chan, So Ching Sarah Chan, Esther W.Y. Chan, Yuet Ying Jessica Cheuk, Jacky Choy, Qing He, Julienne Jen, Jingwen Jin, Ui Soon Khoo, Ho Yan Angie Lam, May P.S. Lam, Yik Wa Law, Jetty Chung Yung Lee, Feona Chung Yin Leung, Ann Leung, Rebecca K. W. Liu, Vivian Wei Qun Lou, Pauline Luk, Zoe Lai Han Ng, Alina Yee Man Ng, Maggie Wai Ming Pun, Mary Lok Man See, Jiangang Shen, Grace Pui Yuk Szeto, Eliza Y.T. Tam, Winnie Wan Yee Tso, Ning Wang, Runjia Wang, Janet Kit Ting Wong, Janet Yuen Ha Wong, Grace Wai Yee Yuen: Implementation, Data Curation, Writing- Reviewing and Editing. George Lim Tipoe: Supervision, Conceptualization, Writing, Editing and Final approval of the version. All authors above agree to be accountable for all aspects of the work.

Acknowledgement

We would like to acknowledge the near-peer-teachers who facilitated the team discussion and the IT officers who managed the online operation of the programme.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The authors confirm that the data supporting the findings of this study are available from the corresponding author, GLT, upon reasonable request.

Additional information

Funding

This research was funded by Seed Fund for Basic Research for New Staff granted to the first author (Ref no. 202009185081). The source of the fund has no participation in preparing the manuscript.