<?xml version="1.0" encoding="UTF-8"?>
<rss  xmlns:atom="http://www.w3.org/2005/Atom" 
      xmlns:media="http://search.yahoo.com/mrss/" 
      xmlns:content="http://purl.org/rss/1.0/modules/content/" 
      xmlns:dc="http://purl.org/dc/elements/1.1/" 
      version="2.0">
<channel>
<title>Greenhouse Sociology</title>
<link>https://matthias-br.github.io/</link>
<atom:link href="https://matthias-br.github.io/index.xml" rel="self" type="application/rss+xml"/>
<description>Researching Society in the Anthropocene</description>
<generator>quarto-1.9.37</generator>
<lastBuildDate>Mon, 06 Oct 2025 22:00:00 GMT</lastBuildDate>
<item>
  <title>Simple Models vs. Experts. The Case of GDP Growth Prediction</title>
  <dc:creator>Matthias </dc:creator>
  <link>https://matthias-br.github.io/posts/post4.html</link>
  <description><![CDATA[ 




<p>Each year in November, the German Council of Economic Experts publishes a report on the country’s economic situation, containing also a forecast of real GDP growth for the following year. Since its establishment in 1963, the council has had a considerable impact on economic policy, with its members being dubbed as “economic sages”. How do the sages’ forecasts hold up when compared to actual GDP growth rates, and can they beat simple prediction models?</p>
<p>I will compare the expert forecasts of annual GDP growth from 2010 to 2025 against predictions based on i) a simple historical mean, ii) a 3-year moving average, and iii) an OLS autoregression of each years GDP growth on 3 lags, using data from a rolling 10-year window. Unlike the expert forecasts that build on additional data and domain knowledge, these predictive models are exclusively based on time series data of the dependent variable alone and can reasonably be expected to perform worse.</p>
<p>The figure below shows actual GDP growth (in % of previous year) and forecasted growth according to each method. On a side note, R packages <code>flexdashboard</code>, <code>plotly</code>, and <code>crosstalk</code> allow you to create Shiny-like dashboards - without Shiny. However, as the R Markdown document is simply knit to a static HTML file, the dashboard lacks Shiny’s reactivity, meaning it is limited to displaying existing data and can’t generate any new data. For our purposes here, that’s perfectly sufficient.</p>
<iframe src="../images/dashboard-gdp-growth-prediction.html" height="520px" width="115%" style="border:none;">
</iframe>
<p>As we can see, none of the forecast methods are very accurate, although some do fare better than others. The Covid-19-induced downturn in 2020 was of course absolutely unpredictable for all models, but it exhibits a particular risk associated with using recent lags for prediction, such as with the 3-year moving average and autoregression: If an unpredicted downturn is immediately followed by growth, as is often the case, these models will still predict a downturn. Especially the autoregression also exhibits some dramatical positive overshoots, compensating for the preceding downturns (2008 financial crisis, Covid-19 pandemic) by reversing to the mean and subsequently overcorrecting.</p>
<p>Ranked by RMSE, the experts indeed performed best, followed very closely by the historical mean, with the moving average further behind and the autoregression coming in last.</p>
<table class="caption-top table">
<thead>
<tr class="header">
<th>Metric</th>
<th>Experts</th>
<th>Historical mean</th>
<th>3-year MA</th>
<th>AR(3)</th>
</tr>
</thead>
<tbody>
<tr class="odd">
<td>RMSE</td>
<td>1.8</td>
<td>2.0</td>
<td>2.6</td>
<td>5.4</td>
</tr>
<tr class="even">
<td>Hit rate</td>
<td>5/15</td>
<td>4/15</td>
<td>1/15</td>
<td>3/15</td>
</tr>
</tbody>
</table>
<p>In other words, the most complex model by far (data-driven human judgement) performed best, as might be expected, but among the time series-based prediction models, the simplest one did not only beat the (somewhat) more complex ones, it also comes very close to the winning model’s accuracy.</p>
<p>If we arbitrarily define a tolerance of +- 0.5 p.pt. to be an acceptable margin of error for a forecast, the experts still lead with five “hits”, still followed by the historical mean, while the autoregression now outperforms the moving average by two hits.</p>
<p>If we think about it from the perspective of some decision maker, be it a business owner or a politician, none of these odds look to good. But what really stands out is just how close the simplest method of forecasting gets to a (presumably costly) expert judgement. Experts’ domain knowledge does appear to give them a certain edge as a crisis unfolds, they accurately predicted the 2021 growth following the pandemic downturn. But it is still not enough to foresee a future crisis. In fact, the very same council, like most economists, could famously also not predict the crises of 2007 or 2011.</p>
<p>In summary: Forecasts of GDP growth on time series data of the dependent variable alone are generally inaccurate, but experts are barely any better. Considering the cost of experts, prediction by historical mean could be a solid option for simple one-year-ahead forecasts. But what then is the point of having an expert council, and should we listen to them as much as we do? Perhaps this shows the need for an advisory institution with a more critical view on economic affairs (and itself). It certainly illustrates how higher model complexity does not equal higher predictive accuracy.</p>



 ]]></description>
  <guid>https://matthias-br.github.io/posts/post4.html</guid>
  <pubDate>Mon, 06 Oct 2025 22:00:00 GMT</pubDate>
  <media:content url="https://matthias-br.github.io/images/post4.png" medium="image" type="image/png" height="86" width="144"/>
</item>
<item>
  <title>Climate Change Attitudes - Causes of Climate Change</title>
  <dc:creator>Matthias </dc:creator>
  <link>https://matthias-br.github.io/posts/post3.html</link>
  <description><![CDATA[ 




<p>An interactive map showing the share of people in each country who think that climate change is caused by humans, using ISSP 2020 survey data.</p>
<p>This is mostly an attempt at embedding interactive graphs made with the plotly package for R. The result is not pretty (yet), but it works.</p>
<p>The data is still interesting, though. While climate change deniers are a relatively small group in most countries, not everyone who acknowledges climate change also thinks it is mostly caused by human activity. In fact, as shown below, it is mostly Northern and Western European nations where an absolute majority shares this belief. This illustrates the need to understand attitudes towards climate change as a spectrum, ranging from environmentalists to deniers, with degrees of belief and scepticism in between.</p>
<p>Mouseover for percentages. <iframe src="../images/climatemap.html" height="500px" width="100%" style="border:none;"></iframe></p>



 ]]></description>
  <guid>https://matthias-br.github.io/posts/post3.html</guid>
  <pubDate>Wed, 09 Jul 2025 22:00:00 GMT</pubDate>
  <media:content url="https://matthias-br.github.io/images/post3.png" medium="image" type="image/png" height="86" width="144"/>
</item>
<item>
  <title>What Can Sociology Contribute to Climate Change Research?</title>
  <dc:creator>Matthias </dc:creator>
  <link>https://matthias-br.github.io/posts/post2.html</link>
  <description><![CDATA[ 




<p>A burning question in sociology in recent years, which has unfortunately not been asked much by researchers outside the discipline. The goal here is not to summarize concrete findings (that would require a separate post), but rather to give readers unfamiliar with sociology a better idea of what this discipline offers in very general terms.<!--more--> To start it off, it might be useful to talk a bit about what sociologists even do and what our research looks like.</p>
<p><em>(Caution: This has become a longer read than intended, but you can of course jump straight to the summary.)</em></p>
<section id="some-notes-on-sociology-for-the-interdisciplinary-audience" class="level3">
<h3 class="anchored" data-anchor-id="some-notes-on-sociology-for-the-interdisciplinary-audience">Some Notes on Sociology for the Interdisciplinary Audience</h3>
<p>For starters, sociology is a social science.</p>
<p>Social science is the study of human societies and their underlying structures, the way people act and interact as a part of them, the way they change over time. Besides sociology, it includes disciplines such as political science, anthropology, human geography, economics, and psychology (though economists sometimes seem to think of themselves as quasi-natural scientists, while neuropsychologists actually cross into that territory, the boundaries can be blurry).</p>
<p>Generally speaking, sociology is not primarily concerned with the individual and their behavior, in which it differs from psychology, for instance. Instead, we focus on phenomena on a larger scale. That could be an entire nation, but also smaller groups of people, such as organizations or families. At the center of all sociological inquiry is the relationship of the individual and society, agency and structure, free will and circumstance. Sociological theories place varying importance on one or the other in explaining social phenomena, and like all social sciences, the discipline has been shaped by a history of competing schools of thought. In fact, early sociology of the 19th century was a more philosophical endeavor and often dealt with metaphysical and very normative questions of human (co)existence, while the period roughly since World War II has seen a rise in data-driven empirical work, especially with the advent of modern computing.</p>
<p>Modern day sociological research resembles the natural sciences in that it aims to generate generalizable findings. However, such generalizations are difficult to make due to the inherent complexity of human behavior, interactions, and social structures. Thus, the reach of sociological knowledge is quite limited when compared to the natural sciences, and we always operate on the basis of probabilism rather than determinism. That is, we make statements such as “if x happens, then y becomes more likely” instead of “if x happens, y happens”.</p>
<p>This unfortunately means that sociological insights tend to be somewhat inconclusive, which makes them unpracticable for politicians, often leading to them being ignored. But there is another reason for that, too. Since sociology is a science, it does not stop at describing reality, it also strives to explain what is observed. And explaining something is to acknowledge it is not self-evident that things have to be the way they are. As a result, sociology is a critical social science and political by design. It does not take social arrangements for granted and tries to view them from an outside perspective, with as little status-quo bias as possible.</p>
<p>When facing an existential threat such as climate change, this critical stance is invaluable. It allows us not only to ask “How did we get here?” and “How do we get out of this?”, but also to find answers that go deeper than “By emitting greenhouse gases” and “By emitting less”.</p>
</section>
<section id="take-the-sociology-pill-and-you-might-begin-to-see-structures" class="level3">
<h3 class="anchored" data-anchor-id="take-the-sociology-pill-and-you-might-begin-to-see-structures">Take the Sociology Pill and You Might Begin to See Structures</h3>
<p>In order to better understand this unique perspective sociology offers, it is perhaps useful to introduce the term “sociological imagination”. It has been used to describe a certain way of analytic thinking that allows us to see our own personal experiences within the broader context of society.</p>
<p>Sociologist C. W. Mills (1959) understood this imagination as the ability to link what happens in our personal sphere to the overarching “issues” of the social structure we live in; i.e., realizing that buying a new phone every two years is not merely an individual preference but also heavily influenced by a growth-dependent economy, marketing, learned notions of necessity, peer pressure, and short product life-cycles. This ability is thought to be lacking in most people, who rarely question the origins of their everyday practices. Unfortunately, this includes many climate researchers and politicians.</p>
<p>Norgaard (2018) speaks of an “ecological imagination” that enables us to see how we physically impact our planet’s environment. This kind of imagination has so far been dominant in science, politics, and the general public, but it is now hitting a wall. We know what causes climate change, but we can not make sense of why we are seemingly incapable of stopping it, despite all the knowledge climate scientists have accumulated.</p>
<p>In the social sciences, the more influential economic and psychological literature on climate change is characterized by an individualist perspective on climate change and society. Focusing on consumer choices, sometimes accounting for individual-level context factors such as climate change knowledge and attitudes, it mostly fails to apply true sociological imagination to the issue.</p>
<p>From this individualist perspective, economists and psychologists typically employ the concept of behavior; that is, decisions driven by motives, interests, attitudes and so on. By manipulating these drivers, individual behavior could then be steered into a desired direction. This approach, dubbed the “ABC” model by sociologists (attitude, behavior, choice), assumes that pro-environmental decisions of citizens can be achieved through education, awareness campaigns, incentives, etc. An approach that, while providing some important insights, has obviously not been enough, as harmful consumption remains high. At the same time, this way of thinking tends to shift the blame on consumers and relieves governments and corporations of their responsibility (Shove 2010).</p>
<p>From the sociological perspective, what prevents us from achieving a true understanding of the social side of climate change is, in very broad terms, a widespread blindness to how circumstances beyond our control and perception guide our decisions. In other words, it is important to realize that social structure (loosely defined as established, organized patterns of social action, the “framework” of society) shapes individual action. But, also, that repeated and accumulated actions create these structures in the first place, keep them alive, or change them – the duality of structure (Giddens 1979).</p>
<p>Simply put, sociology is more inclined to consider the wider context of individual action. It focuses less on each person’s conscious decisions and more on the work of largely invisible (but very powerful) social forces.</p>
</section>
<section id="where-is-sociology-still-lacking" class="level3">
<h3 class="anchored" data-anchor-id="where-is-sociology-still-lacking">Where is Sociology Still Lacking?</h3>
<p>Naturally, sociology does not have all the answers, various weak points exist within the discipline. The most significant ones with regards to climate change are, in my view, the following:</p>
<p>We still do not sufficiently engage with the <strong>physical/biological limits of human societies</strong>, and what coming close to or exceeding these limits means for societies. Already there are calls to take the more abrupt types of change into account, induced by singular high-impact events or social tipping points. In more drastic terms: Sociology should consider not only gradual change, but also the collapse of societies (Adloff 2022).</p>
<p>This goes hand in hand with a growing need for <strong>prediction</strong>, which sociologists have long been wary of for various reasons, despite our general aim to generalize findings beyond a specific sample (this deserves a separate post, though). On a practical level, predictions are currently hampered by sociology’s lack of established ways to model complex social systems, especially using simulation. The earth sciences provide us with models that show possible climatic futures. It is up to sociologists to add models of social futures. Spelling out in more concrete terms what changes in the environment might mean for societies would at the very least help us be better prepared for the future, perhaps even accelerate mitigation efforts. While exact forecasts are impossible, these don’t have to be the goal of prediction. First, prediction can spur debates and direct attention to potentially dangerous developments (see Limits to Growth, Meadows et al.&nbsp;1972). Second, using available data and knowledge of social and environmental systems, it is possible to present uncertain yet plausible paths of social change under a changing climate, which can be refined as more data becomes available and knowledge grows. In fact, some sub-disciplines have already developed expertise in this, such as migration research or population projection. Third, in an analytic sense, prediction can also serve to more rigorously test our theories against real-world data (Verhagen 2022).</p>
<p>All of this requires engaging more with other disciplines. Certain social science disciplines already <strong>bridge the theoretical and empirical gap between society and environment</strong>, such as human geography or social ecology. Mainstream sociology, however, continues to be at odds with the idea of incorporating environmental changes into our theories and empirical research. On the flip side, this of course also requires the natural sciences to be open to cooperation. Realizing that accumulating truly impressive amounts of knowledge on climate change has not been enough to stop it, is the first step.</p>
</section>
<section id="in-summary" class="level3">
<h3 class="anchored" data-anchor-id="in-summary">In summary…</h3>
<ul>
<li>Sociology considers the wider context of individual action and its relation to overarching social structures. Other approaches see the individual as the primary object of research and the target of climate policies, failing to appreciate the complexity of society, unfairly shifting responsibility to the people, and hindering true progress in the fight against climate change.</li>
<li>Sociology is comfortable with the idea that societal arrangements may (have to) change, but politicians often prefer simpler and more practicable answers.</li>
<li>Sociology is equipped to find out why we are not doing enough, and how we can enable the change that is necessary to mitigate/adapt to climate change.</li>
<li>Sociology can’t present definitive solutions to the climate crisis or exact forecasts of social change, but can contribute to preparing ourselves for a turbulent future.</li>
<li>Sociology needs help in linking the social to the natural world.</li>
</ul>
</section>
<section id="references" class="level3">
<h3 class="anchored" data-anchor-id="references">References</h3>
<p>Adloff, F., 2022. Gesellschaftlicher Kollaps und Kollapsologie, in: Ibrahim, Y., Roedder, S. (Eds.), Schluesselwerke Der Sozialwissenschaftlichen Klimaforschung. transcript Verlag, Bielefeld, pp.&nbsp;339–348.</p>
<p>Giddens, A., 1979. Central Problems in Social Theory. Macmillan Press, London.</p>
<p>Meadows, D.H., Meadows, D.L., Randers, J., Behrens III, W.W., 1972. The Limits to Growth: A Report for the Club of Rome’s Project on the Predicament of Mankind. Universe Books, New York.</p>
<p>Mills, C.W., 1959. The Sociological Imagination. Oxford University Press, New York.</p>
<p>Norgaard, K.M., 2018. The Sociological Imagination in a Time of Climate Change. Global and Planetary Change 163, 171–176. https://doi.org/10.1016/j.gloplacha.2017.09.018.</p>
<p>Shove, E., 2010. Beyond the ABC: Climate Change Policy and Theories of Social Change. Environment and Planning A: Economy and Space 42, 1273–1285. https://doi.org/10.1068/a42282.</p>
<p>Verhagen, M.D., 2022. A Pragmatist’s Guide to Using Prediction in the Social Sciences. Socius 8. https://doi.org/10.1177/23780231221081702.</p>


</section>

 ]]></description>
  <guid>https://matthias-br.github.io/posts/post2.html</guid>
  <pubDate>Sun, 31 Mar 2024 22:00:00 GMT</pubDate>
  <media:content url="https://matthias-br.github.io/images/post2.png" medium="image" type="image/png" height="86" width="144"/>
</item>
<item>
  <title>Hello World!</title>
  <dc:creator>Matthias </dc:creator>
  <link>https://matthias-br.github.io/posts/post1.html</link>
  <description><![CDATA[ 




<p>Content coming very soon!</p>



 ]]></description>
  <guid>https://matthias-br.github.io/posts/post1.html</guid>
  <pubDate>Wed, 08 Nov 2023 23:00:00 GMT</pubDate>
  <media:content url="https://matthias-br.github.io/images/post1.png" medium="image" type="image/png" height="86" width="144"/>
</item>
</channel>
</rss>
