The Development of Generalized Trust among Young People in England

Round 1

Reviewer 1 Report

Generally I find the paper novel and interesting. However I have some comments how it could be improved.

It is not clear enough what is the main focus of the study, and therefore what is its main message and contribution. If it is about the prediction that precarious position of many young people may harm social trust (which seems up-to-date and interesting) it should be clearly stated, both in the Introduction and especially in the Abstract, which definitely needs correction in this area.

It is not convincing for me that the presumed impact of tenure and lack of stable job is a part of social learning theory. The possible main mechanism in action, as explained in your paper, is about higher risk connected with trust in strangers, and not change in trust due to interactions with others as it is suggested by social learning theory. It may be also a result of being, due to better economic situation, more self-confident and content with life and therefore better disposed to the world and people in general, which again has little to do with social influence and social learning. However, it is possible that low economic status affects the composition of social networks that an individual is a part of and the level of his/her civic participation, and therefore has a possible impact of generalized trust as well, but this way of argumentation is not given in the Introduction. Furthermore, what is lacking and could be helpful to be included, is a different expected influence connected with participation in different social groups. Some of social groups, especially those in-ward focused (and even out-ward hostile), foster only particularized trust and may even hinder generalized trust. This may be the case of young people with lower economic status. For reference, please consider among others:

Burt, R.S. Brokerage and Closure. An Introduction to Social Capital; Oxford University Press: Oxford, UK, 2005; 4–5, 10–28, 93–223.

Łopaciuk-Gonczaryk, B. Does Participation in Social Networks Foster Trust and Respect for Other People—Evidence from Poland. Sustainability 2019, 11, 1733.

Paxton, P. Association Memberships and Generalized Trust: A Multilevel Model across 31 Countries. Soc. Forces 2007, 86, 47–76.

Uslaner, E. The Moral Foundations of Trust; Cambridge University Press: Cambridge, UK, 2002; pp. 1–50, 115–159

I suggest to develop the discussion in the Introduction, focusing more on potential mechanisms in action and with aiming to underline a straight, central message of the text.

Due to the nature of the depended variables (trust, active civic participation), it is proper to use ordinal logistic regression and not linear regression. The given reason that “We did so because the coefficients of linear regression are easier to interpret” does not sound for me as a sound justification for this choice. As minimum, ordinal logistic models should be estimated and discussed as a robustness check, but I would recommend to perform ordinal logistic regressions as the final analyses.

Although longitudinal set of data was utilized, the approach taken (analysis of regression) is eventually cross-sectional. Even with a control of an initial level of trust in early adolescence, still it cannot be claimed that it was possible to verify the influence, and not purely the correlations. It is only partly remedied by the reverse analysis of the initial trust trust as a potential determinant of later social/economic conditions. Furthermore, the presented model does not do to well to explain the generalized trust, as R square is low, and it is also a limitation to be admitted.

For the reader it is not clear in the beginning what are “social” and “private rent” categories regarding the tenure. It starts to be clear after reading 4.4 – but I would recommend to  explain it better in the part introducing the variables. In Table 7 “Renting from council” is not clear – does it also include private renting as suggested in the begging of the part 4.4?, does it refer to the category of people living only individually or also with parents (who may rent or be owners)?

Trust towards “People of your own age” has high loadings of both generalized and particularized trust and by intuition is associated with peer group, therefore it can be argued, if it is really a measure of generalized trust. I suggest to reconsider this claim. It could be used as a measure of social trust in general, with an adequate definition, but it is disputable to use it as an indicator of generalized trust as opposed to particularized trust.

For the language, I find it a bit too informal with too much rhetorical questions used.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

Review of sosci-592283 v1

 

This is a well written and interesting study on an important topic – generalized trust – with particular attention to a younger generation in Britain who grew up in a context where policy and economic changes had made their lives less secure – a condition that undermines generalized trust – and a condition that, increasingly, will be a fact of life for others. The introduction and search of relevant literature was a pleasure to read. The two data sets provide different windows via the different samples, designs, and questions into the phenomenon of trust with the LLAKES Youth Survey exploring different forms of trust at one point in time for a large, nationally representative sample of 22 – 29 year olds in England whereas the CELS provides longitudinal data to examine the question of changes in trust between adolescence and young adulthood.

 

The authors have been straightforward about the problems of attrition in the CELS data (and the attrition is quite significant). The fact that the data were refreshed at Wave 3 and that the authors are using waves 3 and 6 in the analyses reduce the more major concerns about attrition. (Here I have to add a caveat – when one reads the results, it sounds as if the authors are using data from wave 1 when respondents were 11, although they say on p. 5 that the data come from waves 3 and 6). The demographics on the parents in the sample suggest that there is no reason to worry that selection factors associated with attrition have biased the sample toward the well-educated or upper class – although the comparisons of trust levels for the whole sample vs. the longitudinal sample (Figure 1) points to some selection bias in terms of trust.  The authors also were cognizant of the fact that the original data were collected in classrooms and so they treat the data as nested.

 

I did have a concern with the main dependent variable in the CELS measuring generalized trust. Typically, generalized trust is tapped with a generalized belief about humanity – i.e., most people. Who ‘most people’ are in the minds of respondents is fuzzy but that’s kind of the point about generalized trust, i.e., you don’t know ‘most people’ but do know people you interact with regularly, can predict their actions, and know whether to trust them to follow through, to keep a secret, etc. As the authors point out in the Results section, in the LLAKES survey, generalized trust is tapped with the item “most people can be trusted”. This is a belief about humanity, not people you know. In their analyses of the LLAKES data, the authors also show that generalized trust (most people) is more highly correlated with items tapping trust in strangers and members of outgroups than it is with trust in one’s family.  The authors then use factor analyses to show that there is a generalized and particularized trust factor underlying the LLAKES data – all of these results themselves make a contribution to the literature.

 

That leaves us with the issue of how generalized trust was measured in the CELS. The wording – “How much do you trust the people around you – people of your own age?” sounds closer to some particularized group in two ways – people around you (some kind of familiar territory) and people your own age (who, even if you never spoke to them, would have similarities with you). I feel mixed about whether this single item measures generalized trust. However, since the same item is used at waves 3 and 6, the authors are measuring change over time in this indicator.

 

The factors that impact this indicator of trust are revealing and, for this reason, the paper does make a contribution to understanding how life conditions, especially in late adolescence/young adulthood affect trust. The positive impact of education is known in the literature but having longitudinal data and controlling for parental education and books at home (which should impact foundations of trust in childhood) are important contributions. Perhaps the most interesting were the effects of housing: Living in subsidized housing and renting (compared with living in a home that one has bought) reduced levels of trust. Furthermore, the presentation of results at different ages was very enlightening in terms of how what people are going through at particular stages in life impacts their levels of trust – property owners slump at age 20 but recover by 23; those with low levels of education start out low in trust and show the steepest decline. All of these results speak to the dynamics of the phenomenon of trust and how one’s beliefs about whether others are trustworthy is affected by uncertainties and insecurities in one’s own life and perhaps by the cumulative aggregation of the lack of opportunities, networks, and social connections that are brought about by having little education.

 

In the end, this was an interesting paper that contributed to the developmental and social capital literature on trust.

 

 

 

Author Response

Please see the attachment

Author Response File: Author Response.pdf