Transcript

1

The structure of attitudes toward illegal immigration:the development of cross­national cumulative scales *

running head: Attitudes toward illegal immigration

Kees van der Veer aVrije University Amsterdam

Reidar OmmundsenUniversity of Oslo

Knud S. LarsenOregon State University

Hao Van LeNational Institute of Psychology, Hanoi

Krum KrumovUniversity of Sofia

Regina E. PerniceMassey University

Gerardo Pastor RomansUniversidad Pontificia de Salamanca

a Department of Social Research Methodology. VUA, De Boelelaan 1081c, NL­1081 HV Amsterdam, the Netherlands. Tel: +31 20 4446866, E­mail: [email protected]

2

Summary

This research examined the possibility of developing Mokken cumulative scales

measuring attitudes toward illegal immigrants in a nine­nation sample. A total of

1,407 respondents primarily from national and regional universities participated in the

surveys including the 20 item Illegal Immigration Scale. The scales displayed

acceptable reliability with alpha coefficients ranging from .79 to .93. A Procrustes

analysis yielded coefficients of congruence to the previously established three­factor

solution. The amount of variance accounted for varied between 33.1 to 54.7 percent,

supporting the presence of other factors in attitudes toward illegal immigrants.

Mokken scale analysis yielded robust and economical scales in two clusters of

national samples.

3

The current program of research evaluated to what degree it is possible to

develop cumulative scales with cross­national reliability. The research proceeded

from the idea that there are some communalities in the human experience with

common meanings in at least related cultures and nations. At the same time, it was

also expected that divergent cultural sensitivities and experience may affect both the

meaning of concepts and collateral attitudes.

In the post Cold war world, illegal immigration became a major issue in many

countries. The National Intelligence Council Report: Global Trends 2015 (2000)

asserts that illegal immigration is likely to increase in the future. In the case of some

high income countries, increased life expectancy and falling fertility rates contribute

to an aging population, which is drawing illegal as well as legal immigrants. Such

divergent demographic trends along with globalization of the market place and

despair from political instability and increased conflict, will continue to fuel dramatic

increases in migration. Illegal as well as legal immigrants now account for more than

15 percent of the population in more than 50 countries (Global Trends 2015, 2000, p.

23). Further, pressures for migration will significantly increase in the next 15 years,

the report concludes.

This migration pattern has produced significant changes in attitudes toward

immigrants, reflected in voting behavior, political changes, and government policy.

For example, by the 1980s, negative stereotypes and attitudes toward illegal

immigrants were common in California (Cowan, Martinez, & Mendiola, 1997). Illegal

immigrants were seen as lazy, dependent on handouts, and costly to society. By 1994,

proposition 187 was passed in California which restricted illegal immigrants from

receiving social services, including public health care, education, and welfare

benefits. The rationale of proponents was to reduce the cost of illegal immigration,

4

whereas the opposition felt that proposition 187 was simply an expression of racial

discrimination. Quinton, Cowan, and Watson (1996) found that the more negatively

illegal immigrants were valued compared to legal immigrants the greater the support

for Proposition 187. In other studies (Liu, Widjajawiguna, Shiau, & Dunbar, 1996;

Rodriquez, Lloyd, & Pollitt, 1996), favoring Proposition 187 was related to

authoritarianism and prejudice.

Attitudes toward illegal immigration have also been studied in other countries.

In Australia (Crock & Saul, 2002) researchers observed that polarity of opinion was

related to broader conservative­liberal world views. Overall, the reaction to asylum

seekers is hostile at the level of public opinion and government, reflecting a fortress

mentality. Some (McMaster, 2002) see a historical continuity of underlying fear in

existing populations as asylum seekers are seen as a threat to both national security

and identity. In other receiving countries illegal immigration is also a serious political

issue. Although less harsh than some countries, in New Zealand, immigration laws

after September 11 are primarily designed to ensure security (Haines, 2002,

Immigration and refugee law: update 2001­2002. Seminar paper presented at the

Auckland district law society on 25 November, Auckland.

(http://www.refugee.org.nz/ADLS3.htm).

The importance of developing scales measuring attitudes toward illegal

immigration is apparent. Ommundsen and Larsen (1997) reported results on the

reliability and validity of a 30­item Likert type scale which measured attitudes toward

illegal immigrants. The result of the studies yielded a well balanced scale of positively

and negatively keyed items with moderate to high part­whole correlations. Other

results indicated that anti­illegal immigrant attitudes were related to being male,

authoritarian, and displaying prejudice toward other minorities. Ommundsen and

5

Larsen (1999) using a 20­item version of the scale later compared illegal immigration

attitudes between Danish, Norwegian, and U.S. undergraduate students. Again, the

studies yielded very high alpha coefficients, and identified radicalism­conservatism as

the primary predictor. Is it possible to develop robust scales which reflect common

meanings in various national samples? Also, could such scales be more economical

compared to the 20­item scale while still reflecting the major factors in national

samples? The results (Ommundsen, Hak, Morch, Larsen, & Van der Veer, 2002)

yielded a five­item Mokken cumulative scale in the four national samples, reflecting

three underlying factors.

While these results were encouraging in the search for cross­nationally valid

cumulative scales, they were not definitive. Obviously, the four national samples also

shared significant cultural meaning, did not reflect the national divergence which may

exist in more culturally diverse international samples. Further, some countries are

primarily “receiving”, others are “sending” illegal immigrants. This difference may

affect the structure of attitudes i.e., whether responses of receiving and sending

nations’ citizens will be congruent with the three­factor solution of the Ommundsen,

et..al. (2002) study.

6

Method and results

Samples

A total of 1,407 persons from nine national samples participated in the study. The

samples were drawn from undergraduate students at national and regional universities

in the respective countries, but included a few non student participants in the case of

Spain and New Zealand. The translation of the Illegal Immigration scale was verified

independently for each national sample by participants fluent in both English and the

native languages The samples were 101 from the U.S. (29.4% male, M age= 20.7 yr.,

Sd = 5.11); 97 from Denmark (77. 3% male, M age = 26.5 yr., Sd = 6.93); 102 from

Netherlands (31.4% male, M age = 23.0 yr., Sd = 1.19); 334 from Norway (39.2%

male, M age=23.1 yr., Sd = 4.60); 120 from Spain (40.2% male, M age = 31.3 yr., Sd

= 11.01); 153 from Australia (21.6% male, M age = 21.5 yr., Sd 5.93); 102 from New

Zealand (26.5% male, M age = 25.5 yr., Sd = 10.46); 219 from Bulgaria (22.4% male,

M = 23.0 yr., Sd = 1.44); 179 from Vietnam (70.6% male, M age = 20.9 yr., Sd =

6.93).

Scale analysis

The 20­item Illegal Immigration scale was reported in Ommundsen

and Larsen (1999). Previous results (Ommundsen, et al., 2002) using principal

component analysis with varimax rotation yielded three factors, labeled Cost­Benefit,

Open Borders/Free Flow, and Human Rights in four national samples. The current

study evaluated the robustness of this solution in the aforementioned nine national

samples. The means and standard deviations for all nine national samples are shown

in table 1.

7

­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­

Insert table 1 about here

­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­

Again, significant support was found for the reliability of the scale with alpha

coefficients ranging from a low of .79 in Vietnam to .93 in Australia and Denmark,

with the overall mean alpha .88. The slightly lower alpha in Viet Nam may be

accounted for by less cognition about immigration in sending countries, and therefore

the greater role of chance in decisions to responses categories. Overall, the results

must be considered as significant support for scale reliability in so varied an

international sample.

Procrustes analysis

Since a meaningful three­factor solution was found previously using Procrustes

analysis with the U.S. solution as target, this solution was used in the current study as

target for estimating congruence. Essentially, the Procrustes analysis computes the

agreement of factor loadings from different groups, and Tucker’s phi measures the

degree of congruence. Phi varies from 0 to 1, with 1 signifying perfect congruence,

and allows for differences in eigen­values across cultural groups (Welkenhuysen­

Gybels, J., & Van de Vijver F. J. R. (2003) Methods for the evaluation of construct

equivalence in studies involving many groups.

http://www.kuleuven.ac.be/dvz/softwaredata/webversion.pdf).

Table 2 shows that the three­factor solution previously found for four national

samples provided a reasonable fit to the results for the nine sample survey.

8

­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­

Insert table 2 about here

­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­

While the coefficients of congruence were generally high, it is also obvious that the

three­factor solution does not account for all the variance in the data. The amount of

variance varies from a high of 54.7 percent in the U.S. sample to 33.1 percent for the

Vietnamese. Consequently, while the congruence to the expected solution was

relatively high; there were also other factors contributing to attitudes. The result for

the U.S. sample is partially an artifact, since the U.S. factor solution was the target in

the first Procrustes analysis (Ommundsen, et al., 2002). The U.S. data were used as

target for the Procrustes analysis because the Illegal Immigration scale was originally

developed on U.S. samples, and subsequent studies sought to examine differences in

attitude structure and relevance of scale items in other national samples. Factor 1

(Cost/Benefit) and Factor 2 (Open Borders/Free Flow) are more in line with the

expected solution, with Factor 3 (Human Rights) having coefficients ≤ .57 for four

samples.

For the Bulgarian and Vietnamese samples the three­factor solution accounted

for only 38.9 percent and 33.1 percent of the variance, respectively, and it should be

noted that the alpha coefficients are also lower. A principal component analysis for

these samples reveal additional factors in addition to the three in the theoretical

solution. One explanation is that attitudes toward illegal immigration are more

complex in the “sending” nations. The CIA World Fact book (2002, http.//www.cia.

9

gov/publication/fact book/fields/2112/html) on migration, shows that Bulgaria has a

negative rate (sending nation) of ­4.74, and Vietnam a negative rate of ­0.37 per 1,000

population. All other countries in the survey had positive (receiving) rates varying

from 4.48 in New Zealand to 0.87 per 1,000 in Spain. Another explanation is that in

sending nations, illegal migration is not a major issue socially, politically, or

emotionally. Hence, there is less cognition about these issues and therefore less

consistency in item responses. A lower level of consistency is reflected in both

reliability and congruence coefficients.

The development of Mokken cumulative scales

The Mokken scale procedure computes a measure of scalability (Loevinger’s

Hi) for each single item and for a set of items. In general, an item is considered part of

a cumulative scale if it reaches or surpasses a value of .30. The analysis may be

employed for both dichotomous (Mokken, 1991) and polychotemous items (Sijtsma

& Molenaar, 1996), and is essentially a probalistic version of Guttman scale analysis.

In this case the dichotomous Mokken procedure was employed as Hak, Van

der Veer, and Ommundsen (2002) showed that the respondents found it difficult to

differentiate between the strongly agree and agree categories, and likewise the

strongly disagree and disagree responses. The data were therefore transferred into a

pro­illegal immigrant category versus a not pro­illegal category, by collapsing the

aforemetioned responses.

Searching the total sample for a common scale proved nonproductive for the

nine national samples. Consequently, the Mokken analysis was employed in clusters

of samples in three separate sets. Cluster 1 consisted of western nations (U.S.

Norway, Denmark, Netherlands & Spain). In these countries, illegal immigration and

10

political asylum were issues of public debate and political consideration. Cluster 2

(Australia & New Zealand) were relative neighbors with a somewhat similar history

and culture (Crock & Saul (2002). Although treated differently by the respective

governments and societies, illegal immigration remains a salient issue in both

countries. Australia and New Zealand were placed in a different cluster because

Mokken scale analysis yielded the same results for both nations, but different

from the other samples. Likewise, Cluster 3 consisted of sending nations (Bulgaria

and Vietnam), for which illegal immigration was not a political issue. Here therefore

migration occurred primarily for economic reasons.

­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­

Insert table 3 about here

­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­­

Table 3 shows the results for Clusters 1 and 2 searching for common items

with H larger then .30. Cluster 1 Mokken analysis yielded both a robust and

economical scale with H coefficients larger than .45. Also, all factors of the three­

factor solution are represented in the scale, suggesting the utility of this scale in larger

surveys in western Europe and the U.S. Mokken analysis for Cluster 2 yielded H

coefficients larger than .37. Again, for the Australian and New Zealand samples the

Mokken analysis yielded a cumulative scale both representative of the underlying

three factors and economical. Only item 8 overlapped in the two clusters, suggesting

the salience of different history and experience with illegal immigration. Thus it is

clear that although evidence is found for considerable cross­national utility of a

Mokken scale, that oversimplification is not warranted by the data. In cross­national

11

research, Mokken analysis provides a useful tool for both selecting common items for

a scale but also by pointing to the importance of what is unique in cultural

experiences and the absence of experience.

This is especially demonstrated in the case of the Bulgarian and Vietnamese

samples. The Mokken analysis for these samples yielded no cumulative scale. These

samples were drawn from “sending” nations where the experience with illegal

immigration is different, and where it is not a salient domestic issue. The lack of

responses may be reflected in cognition by lower reliability, and hence the difficulty

in building cumulative scales. It is likely that a different set of items may have tapped

whatever attitudes exist, but also likely that there is less cognition about the issue of

illegal immigration.

Discussion

Recent studies (Tucker and Harman 2002) have pointed to the general dearth

of research and theories sensitive to cultural, ethnic, and national differences. They

challenged the implicit assumptions of comparative studies; i.e., there is little in the

human experience which transcends social categories. Helms and Cook (2002), on the

other hand, argued that the communalities of human existence do transcend social

boundaries and categories, and principles derived are applicable to all human beings.

Certainly most of the existing international research has tended to be comparative,

including studies on ethnic minorities (Coll, Akerman, & Ciechetti, 2000). The

current studies on illegal immigration lend support to communalities in experience

12

and attitudes. These results suggest the possibility of building cumulative scales in at

least related societies, and therefore permit cross­national surveys.

Many factors will contribute to rising illegal immigration in the coming years.

Political instability produces terror in the lives of many people in varying parts of the

world. Practically everywhere in the world, outside of a few favored countries, the

world’s population lives without having minimum needs met. The results are personal

insecurity and poverty propelling millions to look for new beginnings and

a better life. Along with family ties which act as a magnet for many, the existence of

smuggling rings creates new criminal and economic realities in receiving countries.

Some illegal immigrants live under conditions barely that of a slave, indebted to

criminal gangs.

For the receiving countries, there is already an affect on voting behavior,

vigilantism, and laws. In Europe, new political parties have emerged which address

the fears of the existing population of being overwhelmed and experiencing the loss

of national identity. Some of these political parties become popular parties overnight

and have affected political changes toward the right of the political spectrum.

Consequently, illegal immigration will likely become an even more contentious issue

in the coming years and decades.

The development of cumulative scales measuring shifting attitudes is of

transparent importance. As indicated in the aforementioned discussion, illegal

immigration is an issue in many parts of the world, particularly in developed

countries. Previous research employed survey questions of unknown reliability and

validity, or anecdotal studies. This report and the preceding studies have shown the

possibility of developing robust, cumulative, economical scales while reflecting the

underlying attitudinal structure. Further, this study has also shown that cumulative

13

scales can be developed not only within national samples, but also across related

national samples. On the other hand, it also suggests the importance of not

oversimplifying cross­national research. Although there is evidence for cumulative

scales within clusters of nations, Mokken analysis produced no evidence for such

scales in the total nine­ nation sample. In the culturally most divergent samples,

Bulgaria and Vietnam, it was not possible to build Mokken scales. Hence, the

importance of starting with the initial sampling of attitude items.

The fluidity of the immigration situation encourages the use of the Illegal

Immigration scale to assess changes in what may be a worsening crisis. It also allows

the possibility to better understand political reaction in countries which are in a

worsening economic crisis, thus making direct contribution to development of conflict

theory. Unlike culturally dependent scales, cross­national cumulative scales offer an

opportunity to study similarities and differences within clusters of nations. For

example, Australia and New Zealand share to some degree a cultural history, but have

experienced very different government policy in the treatment of illegal

immigrants (Crock & Saul, 2002, & McMaster, 2002). Further studies could examine

the consequences over time resulting in differential attitudes and the degree of policy

consensus in the respective populations.

14

References

Coll, C. G., Akerman, A., & Ciechetti, D. (2000) Cultural influences on

developmental processes and outcomes: implications for the study of

development and psychopathology. In Development and Psychopathology,

12, 333­356.

Cowan, G., Martinez, L, & Mendiola, S. (1997) Predictors of attitudes toward illegal

Latino immigrants. In Hispanic Journal of Behavioral Sciences, 19 (4), 403­

415.

Crock, M., & Saul, B. (2002) Future seekers: refugees and law in Australia. Sydney:

Federation Press.

Global Trends, 2015 (2000) A dialogue about the future with non government

experts.(2000). Washington, DC: National Foreign Intelligence Board under

the authority of the Director of Central Intelligence.

Hak, T., Van der Veer, K., & Ommundsen, R. (2002) An application of the three­step

test­interview (TSTI): a validation study of Dutch and Norwegian versions of

the Illegal Aliens’s Scale. Paper presented at the International Conference on

questions and development, evaluation, and testing methods. Charleston,

South Carolina.

Helms, J. E., & Cook, D .A. (2000) Using race and culture in counseling and

psychotherapy: theory and process. Boston, MA: Allyn & Bacon.

Liu, J. F., Widjajawiguna, S., Shiau, E. C., & Dunbar, E. (1996 ) Immigration

attitudes in a multicultural community: the relationship of subject

demography, cultural contact, and prejudice. Poster session presented at the

15

annual meeting of Western Psychological Association, San Jose, CA., April

1996.

McMaster, D. (2002) Asylum­seekers and the insecurity of a nation. In Australian

Journal of International Affairs, 56(2), 279­290.

Mokken, R. J. (1991) A theory and procedure of scale analysis. The Hague, the

Netherlands: Mouton.

Ommundsen, R., Hak, T., Morch, S., Larsen, K.S. & Van der Veer, K. (2002)

Attitudes toward illegal immigration: a cross­national methodological

comparison. In The Journal of Psychology, 13(1), 103­110.

Ommundsen, R., & Larsen, K. S. (1997) Attitudes toward illegal aliens: the reliability

and validity of a Likert­type scale. In The Journal of Social Psychology, 137,

665­667.

Ommundsen, R., & Larsen, K.S. (1999) Attitudes toward illegal immigration in

Scandinavia and United States. In Psychological Reports, 84, 1331­1338.

Quinton, W. J., Cowan, G., & Watson, B. D. 1996 Personality and attitudinal

predictors of support of proposition 187 ­ California’s anti­illegal immigrant

initiative. In Journal of Applied Social Psychology, 26, 2204­2223.

Rodriquesz A., Lloyd, K. L., & Pollitt, B. K. (1996) Psychological correlates and

attitudes toward proposition 187. Poster session presented at annual meeting

of Western Psychological Association, San Jose, CA, April 1996.

Sijtsma, K., & Molenaar, I. E. (1996) Mokken scale analysis for polychotomous

items: theory, a computer program and an empirical application. In Quality

and Quantity, 24, 173­188.

Tucker, C. M., & Harman, K. C. (2002) Using culturally sensitive theories and

research to meet the academic needs of low­income African American

children. In American Psychologist, October, 762­773.

16

Footnote

* Thanks are expressed to Gerhard van de Bunt for assistance with the Mokken scale

procedure, Marjorie Seaton for data collection, Ladd Wheeler for negotiating ethics

approval at Macquarie University, Dag­Erik Eilertsen for assistance with Procrustes

analyses, and Anna Crane for useful information from her honors thesis.

17

Table 1Means and Standard Deviations of the Illegal Immigration Scale for Total 9­Nation Samples.Items Forming a Mokken Scale in Clusters 1 and 2 are Marked by I and II Respectively.

1. Illegal aliens should not benefit from my tax (name of currency)(M = 3.15; Sd = 1.32).

2. Our taxes should be used to help those residing illegally in the (name of country)(M = 3.18; Sd = 1.35); (I).

3. There is enough room in this country for everyone(M = 3.09; Sd = 1.28); (I).

4. Illegal aliens are not infringing on our country’s resources(M = 3.01; Sd = 1.21).

5. Illegal aliens are a nuisance to society(M = 3.19; Sd = 1.18).

6. There should be open international borders(M = 3.15; Sd = 1.32).

7. Access to this country is too easy(M = 2.77; Sd = 1.35).

8. Illegal aliens should be excluded from social welfare(M = 2.93; Sd = 1.24); (I )(II).

9. The (country) should accept all political refugees(M = 3.20; Sd = 1.26); (II)

10. Illegal aliens who give birth to children in the (country) should be made citizens(M = 3.06; Sd = .96); (II).

11. Illegal aliens cost the (country) millions of (currency) each year(M = 3.15; Sd = 1.32).

12. Illegal aliens should be eligible for welfare(M = 3.14; Sd = 1.20); (II).

13. Illegal aliens provide the (country) with a valuable human resource(M = 3.17; Sd = 1.09); (II).

14. The government should pay for care and education of illegal aliens(M = 2.77; Sd = 1.24).

15. Illegal aliens should not have same rights as (country’s) citizens(M = 2.98’ Sd = 1.29).

16. Illegal aliens have rights, too(M = 3.12; Sd = 1.41); (I).

17. Taking care of people from other nations is not the responsibility of the (country)

18

(M = 3.26); Sd = 1.33); (I)

18. All illegal aliens deserve the same rights as (country’s) citizens(M = 2.99; Sd = 1.28); (I).

19. Illegal aliens should be forced to go back to their own countries(M = 3.11; Sd = 1.32).

20. Illegal aliens should not be discriminated against(M = 3.16; Sd = 1.35); (I).

19

Table 2Coefficients of Congruence (Tucker’s phi) Between Three Orthogonally Rotated Factor Matrices of the Illegal Immigration Scale (USA, Denmark, The Netherlands, Norway, Spain, Australia, New Zealand, Bulgaria, and Vietnam) With a Theoretical Three­factor Matrix as Target. Percentage Cumulated Variance Explained by a Three­factor Solution for Each National Sample.

Orthogonal Factors % Variance explainedCost/Benefit Open Borders/ Human Rights

Free Flow

USA .83 .81 .79 54.7Denmark .70 .73 .54 58.7Netherlands .71 .77 .57 47.1Norway .76 .79 .57 55.9Spain .73 .75 .71 59.6Australia .73 .67 .72 55.8New Zealand .67 .58 .73 52.9Bulgaria .78 .76 .55 38.9Vietnam .78 .54 .60 33.1

20

Table 3Coefficients of Congruence for Two Clusters of National Samples Reporting Loevinger H values.

Cluster 1 (US, Norwegian, Danish, Dutch, and Spanish samples; N = 641;5 items; H ≥ .45)

Items* X H Factor 18 .21 .60 Human Rights 8 .38 .50 Cost/Benefit 3 .44 .45 Open Borders/Free Flow 20 .70 .51 Human Rights 16 .75 .53 Human Rights

Cluster 2 (Australia and New Zealand samples; N = 239; 7 Items; H ≥ .37)

Items* X H Factor 8 .38 .46 Cost/Benefit 13 .42 .43 Cost/Benefit 10 .44 .40 Human Rights 9 .47 .37 Open Borders/Free Flow 12 .50 .48 Human Rights 14 .54 .50 Cost/Benefit 2 .62 .44 Cost/Benefit

*Items are referred to by number in Table 1.