How do social networks influence waste separation?

The model is based on the principal of environmental sociology studying the influence of social factors on the environment. Model works with the empirical-theoretical concept New ecological paradigm (NEP) measuring the values and opinions on the environment. The origin of the project was motivated by the absence of the projects combining the method of empirically calibrated agent-based modeling and sociological grounds, particularly in the Czech context, but also abroad. Based on the environmental module of Czech data ISSP 2010 and research question “How parameters of social network influence the willingness to sort waste?” model was created and analyzed. Relationship between both types of agents (sorting and not sorting waste during the whole simulation) and their neighborhood was found. The higher the number of neighbors, the more agents follow the major behavior - sorting of waste in our case. But homophily blocks this - in case agents establish their neighborhood according their ecological attitude, the rise of major behavior stops with 8 neighbors and with more neighbors is constant. The number of close links amplifies the major behavior, but homophily helps survive the minor behavior.

Model url:

1 Like

Hi Kalvas, nice to see the application of the consumat approach to waste separation. Perhaps you can post the slides since it was not all clear from your class room presentation. I also poke @wander.jager who might be interested to check this out.

Dear Marco, thank you very much for visiting our post! I am uploading the slides right now. I just have to find how to do it :slight_smile:

With deference,

For the presentation asked by Marco Janssen click here:

Our complete project on GitHub:

Thanks for the presentation & sharing the slides. Such models can inform policy decisions and I think that is (or at least should be) an important aspect of computational social science field.

I have a few general comments which are rather related to further directions where this research could be extended.

From a real-world perspective, we know that homophily plays a significant role in formation of friendship ties. Yet, environmental attitude itself is probably not a major determinant of similarity. Thus, it might be tempting to ignore or minimize the effects of environmental attitude in homophily-based tie formation mechanism. However, environmental attitude could be quite correlated with actual determinants of homophily for the purposes of friendship tie formation. Therefore, the extent in which environmental attitudes are correlated with factors that people look for in their friends could be estimated & used to parameterize homophily-based tie formation.

In the model, you utilize a Watts-Strogatz small-world network model with or without homophily in their random link formations. On top of these friendship links, physical neighborhood structure could also be included in the model. In this way, agents’ behavior are not only influenced by their voluntary friends but also by whomever resides in the adjacent houses. This becomes more important in the case where people can easily observe if their neighbors separate waste or not - for instance via the visible separate waste bins or at the time of waste collection. This also might help break the close-knit groups of friends who have negative environmental attitude and do not separate waste.

Lastly, the attitude and behavior are not necessarily equivalent. I am not familiar with the ISSP data so I cannot comment whether it accounts for such difference. Distinguishing between attitude and behavior & reflecting the underlying theories might improve the model. It, of course, might be shown or argued that such distinction provides no additional value & serves no practical purpose (particularly for the purposes of a certain model).

Hoping to see the final paper!

Best regards,

1 Like

Dear Francesko,

Thank you for sharing this interesting work. I am interested in the applications of the consumat approach and diffusion in social networks in general, so I am enjoying reading about your work. But as I am relatively new to ABM, I have many questions. Let me ask you one - I wonder how you would treat/model behavior of an agent with competing factors or sets of information - (perhaps this is close to what Furkan also raised) - for example, how would you model an agent’s behavior when she sees a friend/neighbor reporting separation but not actually separating waste?

Thanks again and all the best,