Construction and validation of an instrument measuring attitudes towards human rights of Liberal Studies teachers in Hong Kong

Abstract

In 2009, Liberal Studies, which can be used as a platform for human rights education, was newly introduced as a compulsory subject for the senior secondary students in Hong Kong. As teacher’s attitudes impact students’ learning largely, a survey was conducted in 2010 to measure Liberal Studies teachers’ attitudes towards human rights. This article focuses on the development and validation of the questionnaire instrument measuring Liberal Studies teachers’ human rights attitudes, which is one of the pioneers in Asian societies. The dimensions of the instrument were first explored by exploratory factor analysis, and were confirmed to include Social Welfare, Civilian Constraints, Personal Liberties, Equality, Privacy – School Management and Privacy –Others. Then, the six-dimensional instrument was validated by confirmatory factor analysis and Rasch analysis. It is expected that it will serve as a base to build a potential tool for comparative study of human rights attitudes in Asian societies.

Keywords

attitudes towards human rights human rights education Liberal Studies teachers Hong Kong Rasch

Introduction

Liberal Studies (LS), a subject which can be used as a platform for human rights education (HRE), was newly introduced in 2009 for the senior secondary school students in Hong Kong. As teachers’ attitudes and knowledge could largely influence students’ attitudes and learning, a survey was conducted to measure the LS teachers’ attitudes towards human rights (HR) and the rule of law and knowledge of the Basic Law, HR and the rule of law, with the funding from the Quality Education Fund. Apart from simple descriptive analysis which had been discussed in detail by Leung and Lo (2012), the data collected had also been used to construct and validate an instrument for measuring attitudes towards HR. This article focuses on discussing the construction and validation of the instrument measuring LS teachers’ attitudes towards HR.

Human rights and human rights education

HR are considered as a set of values that are universal and indivisible. All human beings are born with freedom and equality in dignity and rights without distinction of any kind such as race, language, religion, national or social origin, sex and property. The founding documents of HR include the Universal Declaration of Human Rights (UDHR), International Covenant on Civil and Political Rights (ICCPR) and International Covenant on Economic, Social and Cultural Rights (ICESCR), which together formed the International Bill of Human Rights. Recognising the importance of education in the implementation of HR, the United Nations actively promotes human rights education (HRE), aiming at the cultivation of a universal culture that respects and protects HR and fundamental freedoms. However, it has been argued that successful implementation of HRE in schools depends much on favourable teachers’ attitudes and appropriate teacher education. Favourable attitudes and commitment of the schools’ leaders are also crucial for any success, particularly in matters relating to students’ empowerment (Yuen and Leung, 2010). On the other hand, teachers’ lack of knowledge and unfavourable attitudes towards child rights and their worry about threat to their authority can result in difficulties, resistance and failure (Covell et al., 2010). UNICEF UK (2008) also emphasises the promotion of HR respecting atmosphere through the Rights Respecting Schools Award.

HRE in Hong Kong

Hong Kong has been a British colony, which stringently upholds HR and the rule of law, for 150 years. Though the HR record in Hong Kong is generally satisfactory when compared to most of the Asian countries, HRE has never been high in the educational agenda. From 2009 onwards, LS was newly introduced as a mandatory subject for Secondary four to six students which aims to deepen their awareness of contemporary issues (Curriculum Development Council, 2007). Among the 18 sections, there is one section specifically dedicated to the rule of law and socio-political participation which discusses the concepts and relationship between HR, the rule of law and socio-political participation. Though HR are not explicitly mentioned, many contents in the other sections are closely related to various areas of HR. LS can be used as a platform for HRE (Leung, 2008). Nevertheless, as LS is an interdisciplinary subject, most of the LS teachers are recruited from different areas and they are not adequately trained (Leung and Lo, 2012). As teachers’ attitudes are critical for the success of HRE, it is necessary to investigate whether the LS teachers possess positive attitudes towards HR themselves.

Instruments measuring attitudes towards human rights

Scholars had attempted to construct a reliable instrument for measuring attitudes towards HR. Dimensionality of the instrument was the main focus. Getz (1985) first developed the Attitudes towards Human Rights Inventory (ATHRI) in 1985. The initial version of ATHRI contained 112 controversial items which were keyed on civil libertarian issues stemming from the US Bill of Rights. After performing exploratory factor analysis (EFA) on the first sample which contained 101 respondents from a pro-rights group and a selective-about-rights group, 30 items which showed the strongest divergence between groups loaded highly on the first factor. Together with these 30 items, 10 items which expressed platitudes which everyone agreed were chosen for the final version of the questionnaire, which is unidimensional (Narvaez et al., 2006). Narvaez et al. (1999) administered the 40-item ATHRI to 96 respondents from a conservative religious group and a liberal church group, and 62 undergraduates from a public liberal arts university. The students scored highest, which was followed by the more liberal church group and then the more conservative religious group. The analysis result confirmed the criterion group validity of this unidimensional instrument. However, the dimensionality was unquestioned, and no EFA was conducted.

Diaz-Veizades et al. (1995), on the other hand, developed the Human Rights Questionnaire (HRQ) based on the text of the UDHR. The UDHR contains 31 articles which embrace civil, political, social, cultural and economic rights. At least three items for each human right provision in UDHR were composed for the initial version of HRQ, totalling 116. The HRQ was administered to the first sample of 365 college students for development and then to the second sample of 212 college students and 42 adults for validation. After performing EFA by principal components analysis (PFA) with oblique rotation on the first sample, four factors, Social Security, Civilian Constraint, Equality and Privacy, were extracted and the final 38-item instrument was formed. Social security referred to citizens’ access to an adequate standard of living, Civilian Constraint revealed individuals’ support for restricting civil and political rights, equality referred to equal access to opportunities and equal status before law, and privacy referred to intrusions by the government into the privacy of citizens. The four-dimensional instrument was validated through EFA in the second sample.

Crowson (2004), also suspecting the unidimensionality of the ATHRI and concurring with the multidimensionality of the HRQ, conducted a study to assess the dimensionality of the ATHRI. He administered the 40-item ATHRI to 236 students at a mid-sized Southeastern university. After performing EFA by PFA with Varimax rotation, three factors, Personal Liberties, Civilian Constraint and Social Security, were formed. Personal Liberties addressed attitudes pertaining to liberties; Civilian Constraint referred to limiting the rights of those holding unpopular beliefs or involving in un-American activities and so on; and Social Security referred to governmental provision of welfare services. Finally, a three-dimensional model with a total of 15 items was formed.

McFarlnad & Mathews adopted the 116-item HRQ constructed by Diaz-Veizades et al. (1995) and administered it on the first sample of 67 students. A total of 34 items remained after EFA was conducted by using Principal Axis Factoring with oblique rotation. Two factors, namely HR Endorsement and HR Restriction, were formed. The 34-item questionnaire was then administered to a second sample of 161 nonstudent adults and 74 upper-division college students. Three factors, HR Commitment, HR Endorsement and HR Restriction, were formed. CFA was conducted to compare one-, two- and three-factor solutions. Though the goodness-of-fit indices for the three-factor solution were only acceptable, they were the highest among the three solutions. Thus, the three-factor model was adopted. HR Commitment measured respondents’ preference for HR policies even in the face of financial expenditure to the nation. HR Endorsement contained platitude items which all people inclined to agree. HR Restrictions were the limitations on citizens’ HR under certain situations.

Cohrs et al. (2007), adopting items of Fetchenhauer and Bierhoff (2004), Spini and Doise (1998), Diaz-Veizades et al. (1995) and so on, on endorsement and importance of HR, HR restrictions and military enforcement of HR respectively, developed a three-factor model to measure attitudes towards HR on a sample of 479. After conducting confirmatory factor analysis (CFA) using maximum likelihood estimation of parameters with Amos, the number of items was reduced from 19 to 17. The three factors were named HR endorsement and importance (four items), HR restriction (seven items), and military enforcement of HR (six items). The three-factor model fitted the data well, with root mean square error of approximation (RMSEA) = .054, comparative fit index (CFI) = .944, standardized root mean residual (SRMR) = .050 and Akaike Information Criterion (AIC) = 435.67 and factor correlations ranged between .31 and .54. Table 1 below summarises the above-mentioned influential studies.

Table 1.

Summary of empirical studies on attitudes towards human rights.

Author	Theoretical ground	Sample	Methods	Factors
Getz (1985) Attitudes towards Human Rights Inventory (ATHRI)	Based on US Bill of Rights	101 respondents from a pro-rights group and a selective-about-rights group	EFA; 112 items reduced to 30 items, 10 platitude items were added	Unidimensional
Narvaez et al. (1999) Attitudes towards Human Rights Inventory (ATHRI)	Based on US Bill of Rights	96 respondents from a conservative religious group and a liberal church group, and 62 undergraduates from a public liberal arts university	Comparison of criterion group validity; 40 items	Unidimensional
Diaz-Veizades et al. (1995) Human Rights Questionnaire (HRQ)	Based on UDHR	First sample of 365 college students, second sample of 212 college students and 42 adults	EFA (PCA with oblique rotation); 116 items reduced to 38 items	4 factors: Social Security, Civilian Constraint, Equality and Privacy
Clemence et al. (2001)	Partly based on UDHR	1239 young people from Costa Rica, France, Italy, Romania and Switzerland aged 13–20	EFA (PCA with Varimax rotation); 20 items	5 factors: Violations of Liberties and Equality of Rights, Principle of Assistance and Protection, Issues Regarding Family Members and Spouses, two situations excluded from the realm of HR; Measures against Minorities or Deviants
Crowson (2004)	Adopted the 40-item ATHRI	236 university students	EFA (PCA with Varimax rotation); 40 items reduced to 15 items	3 factors: Personal Liberties, Civilian Constraint and Social Security
McFarland and Mathews (2005)	Adopted the 116-item HRQ	First sample of 67 students and second sample of 161 nonstudent adults and 74 upper-division college students	EFA (PFA with oblique rotation); 116 items reduced to 34 items; CFA	3 factors: HR Commitment, HR Endorsement and HR Restriction
Cohrs et al. (2007)	Partly on existing scales, partly devised by authors	479 respondents, most were students	CFA (Amos) Maximum likelihood estimation of parameters; 19 items reduced to 17 items	3 factors: Endorsement and Importance of HR, HR Restrictions and Military Enforcement of HR
Moghaddam and Vuksanovic (1990)	Interviews with sample population and legal experts, and survey of major themes in important publications on human rights	Derived through interviews and discussions with sample population and legal experts. Administered to 155 undergraduate students from Canada	21 items, No EFA or CFA conducted	No factors
Crowson (2008)	Human Rights-Civil Liberties Scale-Revised Restricting human rights and civil liberties as a necessary part of US government’s war on terror	272 university students from Southern US	52 items, No EFA or CFA conducted	No factors

EFA: exploratory factor analysis; UDHR: Universal Declaration of Human Rights; PCA: principal components analysis; CFA: confirmatory factor analysis.

Among all the studies mentioned above, only the ATHRI constructed by Getz (1985) revealed that attitudes towards HR are unidimensional. All other researchers discovered that attitudes towards HR are multidimensional.

The reasons for the major discrepancy may be threefold. First, Getz’s research was conducted in 1985. By then, education and promotion of HR might not be common and widespread. People might not be able to distinguish between different types of rights (Crowson, 2004). Moreover, according to Stevens (2002), the reliability of the result of factor analysis very much depends on the ratio of sample size to number of factors. Gorusch (1983), Hatcher (1994) and Bryant and Yarnold (1995) recommended a minimum subject-to-item ratio of 5:1 to ensure a reasonably reliable solution. However, the sample size of Getz’s study was just 101 while the number of items reached 112. The extremely low subject-to-item ratio also posed a suspect to the reliability of the study. Furthermore, the study conducted by Narvaez et al. (1999) only validated the instrument through criterion group validity. Neither EFA nor CFA was conducted.

For all other researches which found that attitudes towards HR are multidimensional, the factors constructed share certain similarities. The factor Principle of Assistance and Protection in Clemence et al.’s (2001) study is similar to Diaz-Veizades et al.’s (1995) and Crowson’s (2004) Social Security, while the factors Violations of Liberties and Equality of Rights and Measures against Minorities or Deviants resemble Diaz-Veizades et al.’s (1995) Civilian Constraint and Equality and Crowson’s (2004) Personal Liberties and Civilian Constraint. The factors HR Commitment and HR Endorsement in McFarland and Mathews’ (2005) research are similar to Cohrs et al. (2007)’s Endorsement and Importance of HR, while the factor HR Restriction in McFarland and Mathews’ (2005) research basically resembles HR Restrictions in Cohrs et al. (2007).

Though the multidimensionality of attitudes towards HR was confirmed unanimously by Clemence et al. (2001), Cohrs et al. (2007), Crowson (2004), Diaz-Veizades et al. (1995) and McFarland and Mathews (2005), except McFarland and Mathews (2005) and Cohrs et al. (2007), CFA was not conducted in all other studies. Moreover, for those researches which had conducted EFA, the subject-to-item ratio of some of them was lower than 5:1 (Diaz-Veizades et al., 1995; McFarland and Mathews, 2005). In addition, most of the studies were conducted in the Western societies instead of Asia. It is inquisitive whether the same dimensions would form in Asian societies. Furthermore, the sample populations of most of the above-mentioned studies (Clemence et al., 2001; Cohrs et al., 2007; Crowson, 2004, 2008; Diaz-Veizades et al., 1995; McFarland and Mathews, 2005; Moghaddam and Vuksanovic, 1990; Narvaez et al., 1999) were students; none of them were tested on teachers. Thus, this research, using a subject-to-item ratio higher than 5:1, tested the attitudes towards HR of the LS teachers in Hong Kong.

Methodology

Research method

The study adopted a quantitative research method. A survey was conducted by using a self-administered questionnaire which explored LS teachers’ attitudes towards and knowledge of the Basic Law, HR and the rule of law.

Questionnaire design

The questionnaire was divided into three parts. The first part measured attitudes towards HR, and the rule of law, the second part assessed knowledge of the Basic Law, HR and the rule of law while the last section obtained information on demographic factors. This article focuses on the development and validation of the first 44 items of part one, which measured attitudes towards HR. The instrument adopted four-point Likert scales, with strongly disagree designated as 1 to strongly agree as 4.

Construction of instrument

Based on HRQ, ATHRI and other relevant instruments discussed above, five factors, which commonly emerged in the above-mentioned studies, namely Social Welfare, Civilian Constraints, Personal Liberties, Equality and Privacy, were constructed and formed the instrument measuring attitudes towards HR in this study. The items of the five factors were composed based on relevant literature, including McFarland and Mathews (2005), Cohrs et al. (2007), Moghaddam and Vuksanovic (1990), Trendgo Research Co. Ltd. (2003), Clemence et al. (2001), Crowson (2004, 2008), Committee on the Promotion of Civic Education (2002), Dunbar et al. (2007), The Hong Kong Federation of Youth Groups and Social Sciences Research Centre of the University of Hong Kong (1993) and Oxfam and Amnesty International Hong Kong (1996). The items were selected according to expert advice, and they were re-written to suit the Hong Kong context. Both platitude items (e.g. the government should provide adequate standard of living to citizens) and case-specific items (e.g. the government should provide Comprehensive Social Security Assistance (CSSA) to new arrivals who reside at Hong Kong for less than 7 years) were included in the questionnaire.

Sampling and data collection

There were a total of 460 secondary schools which offered LS in Hong Kong. As the number of LS teachers was different for every school, it was assumed that on average there were five LS teachers in each school. Therefore, the total target population amounted to 2300 (460 schools × 5 teachers per school).

As the size of target population was small, only six pilot tests were conducted on around 80 student teachers and LS teachers between October 2009 and February 2010. They were asked to fill in the questionnaire, classified the first 44 items into the five given factors, Social Welfare, Civilian Constraints, Personal Liberties, Equality and Privacy, (i.e. doing factor analysis manually, as the sample size was not large enough for conducting factor analysis in Statistical Package for the Social Sciences (SPSS)) and give feedbacks. The questionnaire was revised according to the feedbacks from the participants. The overall structure of the final questionnaire is presented in Table 2. The first 44 items of the questionnaire are presented in Table 3.

Table 2.

Structure of the questionnaire.

Section	Variables	Subscales	Item number
I	Attitudes towards Human Rights	Social Welfare (A)	1–8
		Civilian Constraints (B)	9–17
		Personal Liberties (C)	18–25
		Equality (D)	26–35
		Privacy (E)	36–44
		Human Rights Education and Other Relevant Concepts (F)	45–53
	Attitudes towards Rule of Law	(G)	54–66
II	Knowledge of Basic Law, Human Rights and Rule of Law		67–83
III	Demographic factors		84–92

Item no. 10, 11, 12, 13, 14, 15, 16, 17, 26, 28, 30, 33, 36, 37, 38, 39, 40, 41, 42, 43, 44, 46, 49, 50, 51, 52, 55, 58, 59, 61 and 62 are reversed questions.

Table 3.

First 44 items of the questionnaire.

A. Social Welfare

1. The government should provide social protection to all children.

2. The government should provide adequate standard of living to citizens.

3. The government should provide security to people who lose their means for making a living due to circumstances beyond their control.

4. The government has the responsibility to provide all citizens with adequate medical care; therefore extra fees for expensive medications should NOT be charged.

5. The government should include Internet charges in Comprehensive Social Security Assistance.

6. The government should provide Comprehensive Social Security Assistance to new immigrants who reside at Hong Kong for less than 7 years.

7. The government should provide retirement security for all people.

8. The government should exempt all the elderly applying for Old Age Allowance (fruits grant) from income and assets tests.

B. Civilian Constraints

9. Children should be free from corporal punishment by their parents.

10. In times of crisis, the government should have the right to arrest those whose political views would affect social stability.

11. The government should have the right to deny those who would affect public security the right to vote.

12. Schools with religious background should have the right to oblige students to participate in religious activities.

13. The government should have the right to ban Falun Gong-related activities.

14. For political considerations, the government should have the right to restrain citizens’freedom to enter or leave Hong Kong.

15. The government should have the right to stop a peaceful assembly due to mild physical conflicts.

16. The government should have the right to ban the publication of books and other materials which would be a threat to public interest.

17. The government should have the right to refuse to provide costly legal assistance to people who are 100% sure guilty.

C. Personal Liberties

18. A terminally ill patient should have the right to euthanasia.

19. Asylum seekers who would be tortured after returning to home country should have the right to stay in the present country.

20. Women should have the right to abortion.

21. Citizens should have the right to manifest his religion or belief in public.

22. Professors should have the right to teach any theory or belief in university.

23. Political organisations or bodies in Hong Kong should have the right to establish ties with foreign political organisations or bodies, if it would not harm national security or public order.

24. Citizens should have the right to publish articles criticising the Chinese Government in the media.

25. Illegal immigrants who commit crimes in Hong Kong should have the right to legal assistance.

D. Equality

26. Secondary schools should take the balancing of the number of male and female students as a principle when admitting students.

27. Society should encourage women to consider the traditionally male-dominated occupations such as pilots and firemen.

28. The government should legislate to prevent homosexual couples from adopting children.

29. Homosexual teachers can also be competent teachers just like other teachers.

30. Proprietors should have the right to refuse to rent their flats to ethnic minorities.

31. The government should provide non-Cantonese speakers with interpretation when using public services.

32. The government should provide illegal immigrant children with the same right to study as local children.

33. The government should give more voting rights to people with higher economic and social status.

34. I agree with the establishment of a social rehabilitation centre for ex-mentally ill persons near my home.

35. The government should subsidise children with intellectual disabilities to continue their study after 18.

E. Privacy

36. For preventing children from going astray, parents should have the right to monitor their emails.

37. For crime investigation, the law enforcement departments should have the right to intercept our correspondence.

38. Schools should have the right to force students under 18 to take drug test if their parents consent.

39. For combating with drug abuse, the government should have the right to require schools to hand over drug testing results to the law enforcement departments.

40. For enhancing management, schools should have the right to ask for finger prints as identification when teachers enter and leave schools.

41. For enhancing management, schools should have the right to ask for finger prints as identification when students enter and leave schools.

42. Schools should have the right to install security cameras in classrooms.

43. For enhancing public order, the government should have the right to install security cameras in public areas.

44. If an employee applies sick leave for more than five days per month, the employer should have the right to ask for all his/her medical records.

After the pilot tests, as the target population was small, the self-administered questionnaires were mailed to all secondary schools offering LS with return envelopes for the first time in May 2010. All LS teachers were invited to participate. No sampling was required. To increase the response rate, the questionnaires were sent to the schools which had not returned any completed questionnaires for three more times followed by phone calls to encourage teachers to participate. By 31 October 2010, 791 completed questionnaires were returned. Participant response rate was 34.40% (791/2300). A total of 252 schools took part in the survey. The school response rate was 54.8% (252/460).

Data analysis

EFA was first conducted to test whether attitudes towards HR could be measured by the five factors namely, Social Welfare (items 1–8), Civilian Constraints (items 9–17), Personal Liberties (items 18–25), Equality (items 26–35) and Privacy (items 36–44) by using SPSS version 19. An examination of the Kaiser-Meyer Olkin (KMO) measure of sampling adequacy suggested that the sample was factorable (KMO = .856) and Bartlett’s test of sphericity was significant (χ²(946) = 9759.958, p < .0001). Then, CFA was conducted to confirm the five dimensions using MPlus version 6 (Muthen & Muthen). Finally, Rasch analysis was conducted by ConQuest version 1 to validate the model.

Results and discussion

Exploratory factor analysis

Maximum likelihood was adopted as the extraction method for conducting EFA as it allows researchers to compute various goodness-of-fit indices, test the statistical significance of factor loadings and calculate correlations among factors and confidence intervals (Cudeck and O’Dell, 1994; Fabrigar et al., 1999). The output was then rotated by Direct Oblimin, t, as the factors were assumed to be correlated (Gorsuch, 1983). Ten factors were formed after performing EFA on all the 44 items. Factors which had no face validity and those which contained less than three items were deleted, as they were normally unstable and unreliable (Tabachnick and Fidell, 2007). After deletion, there were six factors left. Items with factor loadings lower than .4 (Stevens, 2002) and items which did not have face validity were deleted (Costello and Osborne, 2005). Table 4 presents the result of EFA after items were deleted. All factors were similar to our original categorisation, with Factor A focussed on social security (items 1–4), Factor B focussed on civilian constraints (items 10, 11, 13–17) and Factor C focussed on personal liberties (items 21–24). Only some items with low factor loading were deleted. Factor D focussed on equality (items 5, 6, 25, 32 and 35), with items which were originally put under Social Welfare (item 5: people receiving CSSA; item 6: new immigrants) and Equality (item 25: illegal immigrants; item 32: illegal immigrant children; item 35: children with intellectual disabilities) when the questionnaire was developed. The original factor of Privacy (items 36–44) split into two, which were Factor E: Privacy – School Management (items 40–42) and Factor F: Privacy – Others (items 38, 39, 43, 44). Finally, there were 26 items left, as shown in Table 4 below.

Table 4.

Factors left after deletion.

Factors	Original items	Items left after EFA
Factor A: Social Welfare	1–8	1–4
Factor B: Civilian Constraints	9–17	10, 11, 13–17
Factor C: Personal Liberties	18–25	21–24
Factor D: Equality	26–35	5, 6, 25, 32
Factor E: Privacy – School Management	36–44	40–42
Factor F: Privacy – Others	36–44	38, 39, 43, 44

The percentage of variance explained by each factor, the Cronbach’s alpha coefficient and scale mean of each factor, and the factor loadings of each item are presented in Table 5 below. The percentages of variance explained for the factors, ranging from 32.021% to 42.483%, were not acceptable except Factor B: Civilian Constraints (51.763%) and Factor E: Privacy – School Management (70.548%), indicating that most of the factors could only account for a small ratio to the total variance in all the variables. The Cronbach’s alpha coefficients were acceptable to adequate, ranging from .694 to .880.

Table 5.

Percentage of variance explained, Cronbach’s Alpha coefficient and factor loadings of the factors.

Factor	% of variance explained	Cronbach’s alpha	Mean	SD	Max	Min	Items	Factor loadings
Factor A: Social Welfare	38.957	.699	3.197	.4445	4.00	1.00
							1	.603
							2	.767
							3	.630
							4	.457
Factor B: Civilian Constraints	51.763	.880	2.911	.6011	4.00	1.00
							10		.645
							11		.681
							13		.705
							14		.800
							15		.791
							16		.772
							17		.620
Factor C: Personal Liberties	42.483	.738	3.323	.4638	4.00	1.00
							21			.663
							22			.589
							23			.708
							24			.641
Factor D: Equality	36.008	.683	2.583	.5529	4.00	1.00
							5				.564
							6				.756
							25				.507
							32				.542
Factor E: Privacy –School Management	70.548	.832	2.848	.6829	4.00	1.00
							40					.958
							41					.984
							42					.480
Factor F: Privacy –Others	39.175	.700	2.230	.5382	4.00	1.00
							38						.779
							39						.728
							43						.480
							44						.448

SD: standard deviation.

Confirmatory factor analysis

CFA was then conducted to examine the model fit indices of the Factors A, B, C, D, E and F. For Factor E, as there were only three items, the degree of freedom is 0. The model was just identified. The recommended standards for tests of model fit as suggested by Hu and Bentler (1999) were >.900 for CFI and TFI (Tucker-Lewis fit index), < .060 for RMSEA and <.080 for SRMR, respectively. The model fit indices for this research were generally adequate or good. The CFI for Factors A, B, C, D and F were all higher than .9, which indicated good model fits. Factors A and C had an RMSEA estimate lower than .06, which indicated good model fits, while the model fits for Factors B, D and F were poor (RMSEA estimate > .06). The values of SRMR for Factors A, B, C, D and F were all lower than .08, which indicated good model fits. Though the values of chi-square were generally large and the probabilities were generally smaller than .05, it may be caused by the large sample size instead of poor model fit (McDonald and Marsh, 1990). The model fit indices for attitudes towards HR were also good, with CFI and TFI both larger than .9 and RMSEA smaller than .05. Table 6 below summarises the model fit indices for all the factors.

Table 6.

Fit indices for attitudes towards human rights.

Factors	Items	CFI	TFI	RMSEA			SRMR	Chi-square
		CFI	TFI	Probability	Estimate	Upper bond	SRMR	Value	p
		>.9	>.9	Probability	<.06	Upper bond	<.08	Value	p
A: Social Welfare	1–4	.996	.989	.594	.036	.087	.012	4.042	.133
B: Civilian Constraints	10, 11, 13–17	.977	.966	.021	.070	.087	.026	68.400	.000
C: Personal Liberties	21–24	.998	.994	.681	.028	.081	.010	3.217	.200
D: Equality	5, 6, 25, 32	.931	.862	.001	.102	.130	.037	46.288	.000
E: Privacy – School Management	40–42	1.000	1.000	.000	.000	.000	.000	.000	.000
F: Privacy – Others	38, 39, 43, 44	.982	.945	.088	.082	.128	.025	12.420	.002
Attitudes towards Human Rights		.930	.921	.879	.047	.051	.055	854.851	.000

CFI: comparative fit index; RMSEA: root mean square error of approximation ; SRMR: standardized root mean residual; TFI: Tucker-Lewis fit index.

Rasch analysis

As confirmed in Mplus, the model of Attitudes towards HR was a multidimensional model, with six dimensions identified as Social Welfare, Civilian Constraints, Personal Liberties, Equality, and Privacy – School Management and Privacy – Others. After the model of Attitudes towards HR was identified by conducting EFA and CFA by SPSS and Mplus, respectively, to test how well the data fit the model, ConQuest, a specific programme for conducting multidimensional Rasch analysis, was used to validate the model. ConQuest is a computer programme for fitting item response and latent regression models. It can be used to fit a wide range of models, including Rasch’s simple logistic model, rating scale model, partial credit model, and multidimensional item response models and so on. It provides an integration of item response and regression analysis (Wu et al., 2007: 3). The within-items multidimensional model estimated 99 parameters and yielded a deviance of 39025.884.

Item and person reliabilities

Reliability indices measure the internal consistency of person reliability indices and item difficulty measures, while separation indices represent the number of performance levels in the test and heterogeneity of the respondents. High reliability (of persons or items) means that there is a high replicability that persons (or items) estimated with high measures actually do have higher measures than persons (or items) estimated with low measures (Bond and Fox, 2007).

The average item separation reliability was .997. It demonstrates that the sample size was large enough to establish a reproducible item difficulty hierarchy. However, as shown in Table 7 below, the person separation reliability indices for the dimensions only ranged from .600 to .827. The number of items might not be adequate to discriminate the sample into enough levels for analysis (Fisher, 2007).

Table 7.

Person-separation reliability.

Dimensions	Person-separation reliability (EAP/PV)
Social Welfare (Factor A)	.731
Civilian Constraints (Factor B)	.827
Personal Liberties (Factor C)	.793
Equality (Factor D)	.613
Privacy – School Management (Factor E)	.600
Privacy – Others (Factor F)	.741

Fit indices of items

The infit and outfit mean square (MNSQ) indices were then examined to assess the extent to which the data fitted the model well. Outfit MNSQ (unweighted mean-square) is the outlier-sensitive fit statistic. This is more sensitive to unexpected observations by persons on items that are relatively very easy or very hard for them. Infit index MNSQ (information-weighted mean-square) is the inlier-pattern-sensitive fit statistic. This is more sensitive to unexpected patterns of observations by persons on items that are roughly targeted on them (Linacre, 2003).

Measure is the respondents’ scores converted into logits while the standard errors (SE) indicate the imprecision of the item locations. The lower the standard error, the higher the confidence in the location of item difficulty measures (Bond and Fox, 2007).

As presented in Table 8 below, the infit MNSQ and outfit MNSQ statistics of all the items were within acceptable range (between .6–1.4 according to Bond and Fox (2007); between .5–1.5 according to Linacre (2003)). It shows that they were as predictable as the model expected.

Table 8.

Item parameter estimates and fit statistics.

	Item	Estimate	Error	Unweighted Fit		Weighted Fit
	Item	Estimate	Error	MNSQ	t	MNSQ	t
A: Social Welfare	Q1	−2.345	.070	.99	−.2	.99	−.3
	Q2	−2.223	.072	.88	−2.4	.88	−2.4
	Q3	−2.595	.076	.89	−2.2	.91	−1.7
	Q4	−1.463	.068	1.08	1.5	1.08	1.5
B: Civilian Constraints	Q10	−.108	.008	.94	−1.1	.93	−1.5
	Q11	−.157	.009	.88	−2.4	.87	−2.8
	Q13	−.119	.009	.85	−3.0	.85	−3.0
	Q14	−1.351	.060	.86	−2.9	.88	−2.6
	Q15	−1.431	.063	.85	−3.1	.89	−2.2
	Q16	−.131	.009	.78	−4.7	.79	−5.0
	Q17	−.767	.058	1.20	3.7	1.09	1.9
C: Civil Liberties	Q21	−2.445	.075	1.17	3.1	1.13	2.1
	Q22	−1.810	.064	1.23	4.2	1.21	4.0
	Q23	−2.202	.076	1.11	2.2	1.11	1.8
	Q24	−2.136	.076	.99	−.1	.99	−.1
D: Equality	Q5	−.276	.053	.98	−.3	.98	−.4
	Q6	.378	.051	1.00	.0	.99	−.1
	Q25	−.028	.009	1.08	1.6	1.07	1.3
	Q32	−.067	.009	1.25	4.6	1.23	4.5
E: Privacy – School Management	Q40	−1.214	.062	.79	−4.6	.80	−4.3
	Q41	−1.190	.062	.75	−5.3	.77	−5.1
	Q42	−.828	.061	1.16	3.0	1.15	2.9
F: Privacy – Others	Q38	.091	.009	1.02	.4	1.07	1.4
	Q39	.477	.062	1.24	4.4	1.28	5.0
	Q43	.595	.061	1.09	1.8	1.10	1.5
	Q44	.037	.009	1.15	2.8	1.17	3.2

MNSQ: mean square.

Person-item map

Then, the person-item map was examined to measure person ability against item difficulty. The respondents were arranged according to scores on the left side, while the items were laid out on the right side according to their difficulty measure. An ‘easier’ item, in Rasch terms, means that a larger proportion of respondents have positive endorsements and a smaller proportion of them have negative endorsements. A ‘difficult’ item, on the other hand, means that a proportion of respondents have positive endorsements and a larger proportion of respondents have negative endorsements (Bond and Fox, 2007: 45, 107–108).

As shown in Figure 1, the distribution of persons was generally consistent, with a curve-like shape which peaked around the mean. The items were generally too ‘easy’ for the respondents in all dimensions, that is, the LS teachers tended to agree with the statements. The difficulty levels of the items in the sub-scales were generally too concentrated and there were not sufficient items to further distinguish the respondents with high scores or low scores. In order to further distinguish the respondents, items which are more controversial (i.e. more respondents tend to disagree with) should be added to the scales.

Figure 1.

Person-item map.

Conclusion

This study concurred with Clemence et al. (2001), Cohrs et al. (2007), Crowson (2004), Diaz-Veizades et al. (1995) and McFarland and Mathews (2005) that attitudes towards HR were multidimensional. The present six-dimensional model was very similar to the four-factor HRQ constructed by Diaz-Veizades et al. (1995). The four factors, namely Social Welfare, Civilian Constraints, Equality and Privacy, were similar to each other in the two studies. The only differences were that the LS teachers distinguished issues concerning privacy in school from issues related to privacy in other occasions in this research and that the existence of a sixth dimension, Personal Liberties, which was adopted from Crowson’s (2004) research, was found in this study. Though the analysis of individual items revealed the impact of local values and Confucianism on the attitudes towards HR (Leung and Lo, 2012), it seems that the dimensions of attitudes towards HR shared certain similarities across different samples under different cultures. The similarities shared in the formation of the dimensions of attitudes towards HR between the previous Western studies (such as Crowson, 2004; Diaz-Veizades et al., 1995) and the existing research may be due to the colonial history of British rule and the influence of the West in Hong Kong peoples’ conception of HR. Further investigation would be required to explore the relative strength of impacts of particularity and universality on the attitudes towards HR of different people under different cultures, especially between the East and the West.

For the goodness-of-fit indices of the current model, though the model fit indices for some of the individual dimensions were not within the acceptable range (e.g. the RMSEA for Civilian Constraints, Equality and Privacy – Others were all larger than .06), all the fit indices for the whole model Attitudes towards HR were satisfactory. Moreover, the infit MNSQ and outfit MNSQ statistics of all the items were within acceptable range. The items were as predictable as the model expected.

However, there were still some areas of improvements. The percentages of the variance explained by five of the six dimensions were lower than 60%. The explanatory power of these six dimensions was not strong enough. It might be caused by the lack of opportunity to conduct a quantitative pilot test. As the population of the LS teachers was too small (approximately 2300 people), if pilot test was conducted using quantitative method, there would not be enough people for the formal test. Therefore, the only possible way was to conduct the pilot test using qualitative method, for which the process has been mentioned in detail above. Thus, after items with low factor loadings or no face validity were deleted during EFA and CFA, the chance of adding new items to the dimensions and re-conducting EFA and CFA for improving the indices were lost. As there were only four items for three of the dimensions and only three items for one of the dimensions, the number of items was not adequate to discriminate the sample into enough levels for analysis. It was the reason why the person separation reliability indices for the dimensions were low. It could also explain why there were not sufficient items to further distinguish the respondents with high scores or low scores. Moreover, qualitative research such as in-depth interviews could be conducted to explore the LS teachers’ conceptions of the dimensions of attitudes towards HR and the reasons behind.

To conclude, the overall fit for the instrument measuring Attitudes towards HR was generally satisfactory, which largely concurred with the theoretical framework established by Diaz-Veizades et al. (1995), Narvaez et al. (2006) and so on. However, it is strongly advised that in future research, new items should be added to the six dimensions and EFA, CFA and Rasch analysis should be conducted again to improve the percentages of variance explained by the dimensions, enhance the person separation reliability indices and increase the number of difficulty levels of the items.

Footnotes

Funding

This work was supported by the Quality Education Fund of Hong Kong (2007/0752).

References

Bond

Fox

(2007) Applying the Rasch Model: Fundamental Measurement in the Human Sciences. London: Lawrence Erlbaum Associates.

Bryant

Yarnold

(1995) Principal components analysis and exploratory and confirmatory factor analysis. In: Grimm

Yarnold

(eds) Reading and Understanding Multivariate Statistics. Washington, DC: American Psychological Association, pp.99–136.

Clemence

Devos

Doise

(2001) Social representations of human rights violations: Further evidence. Swiss Journal of Psychology 60(2): 89–98.

Cohrs

Maes

Moschner

. (2007) Determinants of human rights attitudes and behavior: A comparison and integration of psychological perspectives. Political Psychology 28(4): 441–470.

Committee on the Promotion of Civic Education (2002) Biennial Opinion Survey on Civic Education 1998 – Report and Executive Summary. Hong Kong: Committee on the Promotion of Civic Education.

Costello

Osborne

(2005) Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical Assessment, Research and Evaluation 10(7). Available at: http://pareonline.net/getvn.asp?v=10&n=7

Covell

Howe

McNeil

(2010) Implementing children’s human rights education in schools. Improving Schools 13(2): 117–132.

Crowson

(2004) Human rights attitudes: Dimensionality and psychological correlates. Ethics & Behavior 14(3): 235–253.

Crowson

(2008) Belief, motivational, and ideological correlates of human rights attitudes. The Journal of Social Psychology 148(3): 293–310.

10.

Cudeck

O’Dell

(1994) Applications of standard error estimates in unrestricted factor analysis: Significance tests for factor loadings and correlations. Psychological Bulletin 115: 475–487.

11.

Curriculum Development Council (2007) Liberal Studies Curriculum and Assessment Guide (Secondary 4–6). Hong Kong: Curriculum Development Council.

12.

Diaz-Veizades

Widaman

Little

. (1995) The measurement and structure of human rights attitudes. The Journal of Social Psychology 135(3): 313–328.

13.

Dunbar

Sullaway

Blanco

. (2007) Human rights attitudes and peer influence: The role of explicit bias, gender, and salience. International Journal of Intercultural Relations 31: 51–66.

14.

Fabrigar

Wegener

MacCallum

. (1999) Evaluating the use of exploratory factor analysis in psychological research. Psychological Methods 4(3): 272–299.

15.

Fetchenhauer

Bierhoff

(2004) Attitudes toward a military enforcement of human rights. Social Justice Research 17: 75–92.

16.

Fisher

Jr (2007) Rating scale instrument quality criteria. Rasch Measurement Transactions 21: 1095.

17.

Getz

(1985) Moral reasoning, religion, and attitudes toward human rights. Doctoral Dissertation, University of Minnesota, Minneapolis, MN.

18.

Gorsuch

(1983) Factor Analysis, 2nd edn. Hillsdale, NJ: Lawrence Erlbaum Associates.

19.

Hatcher

(1994) A Step-by-Step Approach to Using the SAS? System for Factor Analysis and Structural Equation Modeling. Cary, NC: SAS Institute, Inc.

20.

Bentler

(1999) Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling 6: 1–55.

21.

Leung

(2008) An ‘action-poor’ human rights education: A critical review of the development of human rights education in the context of civic education in Hong Kong. Intercultural Education 19(3): 231–242.

22.

Leung

(2012) Are Liberal Studies teachers ready to prepare human rights respecting students? A portrait of teachers’ attitudes towards human rights. Intercultural Education 23(4): 341–358.

23.

Linacre

(2003) Size vs. significance: Standardized chi-square fit statistic. Rasch Measurement Transactions 17(1): 918.

24.

McDonald

Marsh

(1990) Choosing a multivariate model: Noncentrality and goodness of fit. Psychological Bulletin 107: 247–255.

25.

McFarland

Mathews

(2005) Who cares about human rights? Political Psychology 26(3): 365–385.

26.

Moghaddam

Vuksanovic

(1990) Attitudes and behavior toward human rights across different contexts: The role of right-wing authoritarianism, political ideology, and religiosity. International Journal of Psychology 25: 455–474.

27.

Narvaez

Getz

Rest

. (1999) Individual moral judgment and cultural ideologies. Developmental Psychology 35: 478–488.

28.

Narvaez

Thoma

Getz

(2006) ATHRI GUIDE: Guide for Using the Attitudes toward Human Rights Inventory (ATHRI). Notre Dame, IN: Center for Ethical Education, University of Notre Dame.

29.

Oxfam and Amnesty International Hong Kong (1996) Survey Report on Civic Awareness and Civic Education of the Secondary Students and Teachers in Hong Kong. Hong Kong: Oxfam and Amnesty International Hong Kong.

30.

Spini

Doise

(1998) Organizing principles of involvement in human rights and their social anchoring in value priorities. European Journal of Social Psychology 28: 603–622.

31.

Stevens

(2002) Applied Multivariate Statistics for the Social Sciences, 4th edn. Mahwah, NJ: Lawrence Erlbaum Associates.

32.

Tabachnick

Fidell

(2007) Using Multivariate Statistics, 5th edn. Boston, MA: Allyn & Bacon.

33.

The Hong Kong Federation of Youth Groups and Social Sciences Research Centre of the University of Hong Kong (1993) Survey on Youth and Human Rights. Hong Kong: The Hong Kong Federation of Youth Groups and Social Sciences Research Centre of the University of Hong Kong.

34.

Trendgo Research Co. Ltd. (2003) Survey on Attitudes towards Human Rights of the Senior High School Teachers. Taipei, Taiwan: Trendgo Research Co. Ltd.

35.

UNICEF UK (2008) UNICEF UK rights respecting schools in England: Briefing paper 2008. Available at: http://www.renecassin.org/downloads/rrs-briefing.pdf

36.

Adams

Wilson

MR, et al.

(2007) ACER ConQuest Version 2.0: Generalised Item Response Modelling Software. Camberwell, VIC, Australia: ACER Press.

37.

Yuen

Leung

(2010) How an advocacy NGO can contribute to political socialization: The case of Hong Kong. Citizenship, Social and Economics Education 9(3): 212–224.