markus freitag & paul c. bauer university of konstanz department of politics & management...

13
Markus Freitag & Paul C. Bauer University of Konstanz Department of Politics & Management University of Zürich, 16th of July 2011 Dimensions of trust

Post on 19-Dec-2015

215 views

Category:

Documents


0 download

TRANSCRIPT

Markus Freitag & Paul C. BauerUniversity of KonstanzDepartment of Politics & Management

University of Zürich, 16th of July 2011

Dimensions of trust

• Research question and relevance

• Dimensions of social trust

• Data and operationalization

• Methodological approach

• Empirical results

• Conclusions

Outline

• Which distinct dimensions of social trust can we identify, both theoretically as well as empirically and to what extent are these dimensions equivalent cross-culturally?

Research question

Relevance of research question

o Increasing popularity of “trust” was not paralleled by an increase in conceptual clarityo Trust is the “belief that others, through their action or inaction,

will contribute to my/our well-being and refrain from inflicting damage upon me/us.” (Offe 1999: 47)

o Disagreement among scholars with regard to the number of dimensions of social trust

o Do respondents understand trust questions differently in different cultural contexts?

• Trust as a one-dimensional construct Trust in people we know and trust in people we do not know

form a single dimension due to the fact that the latter arises as an externality from the former

• Trust as a two-dimensional construct Particularized trust

• Trust toward personally known people Generalized trust

• Trust toward people beyond immediate familiarity (unknown people, strangers, random people one meets on the street)

Dimensions of social trust

• Trust as a three-dimensional construct Particularized trust Generalized trust Identity-based trust

• Based on identification and categorization• Identities/categories may refer to behavioral similarities, ethnicity, or

traditions (e.g. nationality, religion)

Dimensions of social trust

Item Question wordingTrust in most people Generally speaking, would you say that most people can be trusted, or

that you can't be too careful in dealing with people? If we take a scale on which 0 means that you can't be too careful in dealing with people and 10 means that most people can be trusted, where would you locate yourself on this scale?

Trust in people one meets for the first time

And what does it look like for certain groups of persons. If you take again the scale from 0 to 10, on which 0 means „no trust at all“ and 10 „a lot of trust“, how great is your trust in persons that you meet for the first time?

Trust in friends …in your friends?Trust in neighbors …in your neighbors?Trust in people of other religion

…in persons of another religion?

Trust in people of other nationality

…in persons of another nationality?

• Data: Survey “Volunteering in Swiss Municipalities” (2010); 4955 respondents in Switzerland

Data and operationalization

Model 1

Model 2a

Model 2b

Model 3

Empirical analysis: Part I – CFA x 4

• Multiple-group confirmatory factor analysis (MGCFA) Test the 3-factor model in population subgroups Subgroups: Individuals belonging to German-, French- and

Italian-speaking regions Levels of invariance

• Configural invariance• Metric invariance• Scalar invariance

Empirical analysis: Part II - MGCFA

Empirical results: Dimensionality of social trust

• Cut-off values indicating good model fit (Brown 2006): • RMSEA < 0.08; SRMR < 0.06; TLI > 0.95; CFI > 0.95

• Model 1, 2a and 2b display rather poor fit; Model 3 fares comparably better• Particularized, identity-based, and generalized trust emerge as three distinct

constructs

Model (N=4289) Chi-Squared Df SRMR RMSEA TLI CFI AIC

1: One dimension 271.46 9 0.041 0.082 0.884 0.930 99542

2a: Two dimensions 185.58 8 0.033 0.072 0.911 0.953 99381

2b: Two dimensions 225.56 8 0.040 0.080 0.891 0.942 99456

3: Three dimensions 24.71 6 0.012 0.027 0.988 0.995 99096

Empirical results: MGCFA – Model 3

  χ2 Df P-value RMSEA ∆RMSEA CFI ∆CFI TLI SRMR

Single Group SolutionsGerman-speaking Switzerland (n=3307)

18.83 6 0.004 0.025 0.988 0.013

French-speaking Switzerland (n=835)

9.11 6 0.168 0.025 0.997 0.992 0.013

Italian-speaking Switzerland (n=147)

2.30 6 0.890 0.000 1.000 1.081 0.020

 Measurement InvarianceConfigural Invariance 29.43 18 0.043 0.021 0.997 0.992 0.013Metric Invariance 36.09 24 0.054 0.019 -0.002 0.997 0.000 0.994 0.016Scalar Invariance 62.72 30 0.000 0.028 0.007 0.991 -0.006 0.987 0.020

• Cut-off values indicating good model fit (Chen 2007): • Change of CFI lower than ≤.005 • Change of RMSEA lower than ≤ .010

Conclusions

1. Particularized, identity-based, and generalized trust emerge as three distinct constructs in our analysis and the theoretically assumed three-dimensionality of social trust is reflected by empirical data

2. It is possible to compare means of the latent constructs particularized, identity-based, and generalized trust between the three Suisse cultural regions investigated here

3. Since our analysis was conducted with the “special case” Switzerland we have reason to believe that the results yielded in our analysis might be true for other European countries as well

Thank you very much!