Portal de Periódicos da CAPES

Gapminder: Using a World of Human Ecology Data to Teach Students Critical Thinking Skills

2012; Ecological Society of America; Volume: 93; Issue: 4 Linguagem: Inglês

10.1890/0012-9623-93.4.358

2327-6096

David C. LeBlanc,

Education and Critical Thinking Development

Note: Charlene D'Avanzo is the editor of Ecology 101. Anyone wishing to contribute articles or reviews to this section should contact her at the School of Natural Sciences, Hampshire College, 893 West Street, Amherst, MA 01002. E-mail: cdavanzo@hampshire.edu Employers, politicians, and society at large are calling for university graduates to be able to think critically, solve problems and communicate effectively. The public emphasis on critical thinking and problem solving has increased recently with calls for greater accountability and student learning outcomes that are relevant to real world contexts (e.g. Arum and Roksa 2011). Development of student critical thinking skills has long been an objective of university science courses as these skills are essential to the practice of science. Some examples of student behaviors associated with critical thinking as applied in this context might include: (1) Identifying and accessing relevant information, (2) Analyzing and evaluating data as evidence to develop well-justified conclusions, and (3) Integrating multiple sources of information, including relevant literature and data analyses to formulate and defend recommendations for policy or action. One topic area that can provide a rich context for the learning and application of critical thinking skills is human population growth and its impacts on natural systems. The consequences of activities by a large and growing global human population are important factors influencing ecological systems and cause many environmental problems. Hence, many instructors of ecology and environmental science include a component dealing with human ecology in their courses. Case studies that investigate regulation of human population growth through changes in birth rate and(or) death rate and how these changes might come about (purposefully via well-considered policies or chaotically due to inaction or inappropriate policies) offer a rich context for students to apply general population ecology concepts to real-world problems. Much has been written on the general topic of human population growth and policy recommendations for influencing this growth rate. This literature has been an important resource for critical thinking exercises pertaining to the human population growth. Evaluating, integrating and summarizing the analyses of multiple literature sources are important critical thinking skills that students should learn. However, students should also learn how to independently analyze raw data and draw appropriate, well-founded conclusions. Student-accessible raw data related to human population dynamics have been hard to find, existing in many different databases, sometimes with limited access and often incompatible formats. The objective of this paper is to describe a single unified database of human population ecology data and an associated data analysis tool that are easily accessed, easily used and highly effective. The Gapminder Foundation has unified many databases from the United Nations, universities and non-governmental organizations into a single database with a consistent format (Rosling 2006, Link to TEDTalks video). The Gapminder World database includes over 600 variables (Table 1), with data from 258 current and former countries and territories. Time series for most variables and countries extend back to 1960; data for some extend back to the early 1800's. The database is still under active development; by the time this is published there will likely be more variables over longer time periods. These data are arranged in spreadsheet format. This consolidation and organization of human population data in a single database facilitates investigations by students who have limited experience in data management and analysis. The Gapminder database is integrated with a simple, yet powerful, data analysis program (Trendalyzer) that facilitates identification of associations among human population variables, spatial patterns and time trends. The default Trendalyzer graphic is a scatterplot that displays five variables, X, Y, dot size, dot color, and Year (see Fig. 1 for an annotated example). Each dot in the scatterplot represents a country and the array of dots in the default graph shows the pattern of association between the two population variables with countries serving as replicate observations. The default graph displays data for the most recent year that data are available. The scatterplot can be animated by clicking on the Play button to show how the pattern of points changes over time; this feature of the Trendalyzer program uses Adobe Flash Player. Detailed instructions for how to use this program are provided later in this paper. Trendalyzer scatterplot of spatial variation among countries, which indicates a positive association between the X-variable Child mortality (0 – 5 yr, per 1000 born), and Y-variable Children per woman (total fertility). Dot size (diameter) in the default graphic reflects population size (N). For example, the large red dot in this graph is for China and large light-blue dot is for India. Dot color indicates geographic region (see map color key in upper right corner of the graph). X- and Y-axis scales can be toggled between linear and log scales by clicking on the lin / log boxes at the end of each axis title. Year of data collection for dots in the scatterplot is displayed in the plot area background. Placing the mouse cursor over a dot causes the name of the associated country to appear, along with the specific X and Y-values. This graphic was produced using the online version of Trendalyzer. Note: The Normal view button at the top-right of the graph toggles the display between full-screen and normal-view graph displays. Before you copy graphs into documents or PowerPoint presentations using a screen capture procedure, you should make the graph Full screen to enhance resolution of the copied graph. The following hyperlink was created using the Share graph button at the top of the graph. To access an online copy of this graph that can be animated click on the following: Hyperlink to Animated Scatterplot. The Gapminder database and integrated data analysis program are freely available via the Internet at http://www.gapminder.org. Data can be analyzed online (called Gapminder World) or the database and Trendalyzer program can be downloaded in its entirety and run on a personal computer. The downloaded program is named Gapminder Desktop and differs in several regards from the online version: (1) The online database in Gapminder World is updated more frequently (Karin Brunn Lundgren, pers. comm.). Hence, identical analyses will sometimes produce different results between the online and Desktop versions. (2) The online Trendalyzer program has a link to a short How to use video clip that very effectively describes the various features of the program. This button in the downloaded Trendalyzer links to a much less effective static graphic. (3) Trendalyzer online has a Share graph feature that allows for easy sharing of graphs and saving graphs online where they can be accessed via hyperlinks (as I have done later in this paper); the downloaded program does not. To access Gapminder online, go to the home page at www.gapminder.org. This home page provides a variety of resources, including: (1) links to a wide variety of video clips that display the use of the Trendalyzer program to analyze various human ecology questions, (2) links other databases created by the Gapminder Foundation, such as Gapminder Agriculture that includes data from the United Nations Food and Agriculture Organization, (3) links to example graphs created using the Gapminder database and Trendalyzer graphics program, and (4) the link to a web page where you can download the Gapminder Desktop database and program to a personal computer. To run Trendalyzer online, click on the GAPMINDER WORLD tab at the top of the home page. This will bring-up the default Trendalyzer graph of Income per person (X) vs. Life expectancy (Y). If you are using Trendalyzer for the first time, you should click on the How to use button at the top center of the default graph to view a 2.5 minute video clip that demonstrates how to use all of the program features. Patterns of spatial co-variation in population variables provide evidence for the strength of association among variables and a preliminary basis for identifying potential cause-effect relationships. The default Trendalyzer scatterplot graphic displays such patterns of spatial co-variation along with additional meta-data about geographic location and time. Select variables: Click on the X-axis title line Income per person. A menu of available variables will appear. Items in the menu ending with a ▸ are sub-menus that contain multiple variables; items without this symbol are individual variables. Click on the variable Child mortality (4th item in the menu). Click on the Y-axis title line Life expectancy and select Children per woman (1st item in the menu). Select linear or log scale: By default, Child mortality is presented on a log scale while the variable Children per woman is presented on a linear scale. The initial scatterplot has a distinct curvilinear pattern. At the end of the X-axis title line is a small box with the text log ▾. Click on this box and select the lin option to plot Child mortality on a linear scale. The scatterplot now has a more linear pattern. Use Zoom tool: To best see the pattern in a scatterplot with the least bias the cloud of points should fully occupy the graph space. After switching the X-axis scale to be linear, the scatterplot points will be crowded in the lower left corner and there will be much blank space in the graphic. To adjust this, click on the small “arrow-in-a-box” icon in the lower right corner of the white plot area (bounded within the X- and Y-axes). The Zoom tool will appear. Click on the magnifying glass with + icon; the cursor should change to a similar icon and the plot area should become darkened. Drag the cursor across the range of the data points from lower left to upper right to highlight the cloud of points but excluding the blank area; the selected area should become white. When you release the mouse button the cloud of points will expand to fill the plot area. If the zoom-in procedure goes too far and excludes some data points, you can zoom-out and begin again. Click on the magnifying class with a – symbol (cursor should change to mimic this) and then click anywhere in the plot area to zoom-out. To remove the Zoom tool from the plot area, click on the small arrow-in-a-box icon in the lower right corner again. Linking dots to countries and data: If you place the point of the cursor on a dot in the scatterplot, the name of the country will appear next to the dot and the X- and Y-values for that dot will appear on the X- and Y-axes. Spatial associations such as that displayed in Fig. 1 provide some evidence to evaluate the strength of association among population variables and provide preliminary evidence for potential cause-effect relationships. The pattern of country dots in Fig. 1 indicates that countries with lower child mortality rates have lower total fertility rates and vice versa. This suggests that lowering child mortality rate might encourage women to have fewer children. However, additional evidence would be required to make a stronger case that this is a true relationship and not merely a statistical association. This additional evidence might include spatial replication of this association across multiple countries and a plausible mechanism for a cause-effect linkage. Spatial replication can be evaluated by analyzing temporal correlation between these two variables across many countries (described later in this paper). Certainly, women who are confident that most or all of their children will survive to adulthood might be expected to reduce the number of children they desire, which would be a plausible cause-effect mechanism. If you are not already in the Trendalyzer program, go to the home page at www.gapminder.org and click on the Gapminder World tab at the top of the screen. In the upper left corner of the default graph, click on the tab labeled Map. This will bring-up the default map of the world, with a dot in each country that reflects the default variable, population size. Dot diameter is determined by the value for the variable Population, total. Controls for this map are in the lower right corner of the Trendalyzer screen. To change the variable to Median age, click on the variable name Population, total, select the Indicators ▸ sub-menu, select the Population ▸ sub-menu, and select the variable Median age (years) from this sub-menu. The size of dots can be manipulated to reduce overlap and improve resolution of individual countries and contrast among countries. In the lower right corner is a slider control with two Δ icons and two numbers that reflect the minimum and maximum values for the selected variable. The lower Δ slider controls minimum dot diameter and the upper Δ slider controls maximum diameter. These two icons can be dragged to modify dot size. Increasing the distance between the two Δ sliders increases the contrast among country dots. However, large maximum dot size will cause dots to overlap in regions with many small countries and it will be difficult to resolve individual countries. Trendalyzer map that displays geographical variation in the value of a single Gapminder variable, Median age (years) in this example, for a specified year. If you place the mouse cursor on any dot in the map, the country name will appear next to the dot and the data value for that country will appear in the lower right corner just below the variable name. This map graphic can also be animated to display change over time (size of dots for each country changes over time) Hyperlink to Animated Map. In this particular map graphic the year scale extends to 2050, indicating that the Gapminder database includes projections of future values for some population variables. Trendalyzer scatterplot and map graphics can be animated to display change-over-time in the pattern of association between two variables or in the spatial variation of a single variable, respectively. Analysis of temporal co-variation between two variables for individual countries provides a basis for documenting spatial replication of the association. Finding the same pattern of temporal co-variation between two population variables for multiple countries that include a diversity of geography, economic development, culture, and religious traditions provides evidence that the observed correlation between variables is more than mere coincidence. Such evidence is part of the documentation needed to infer that statistical association reflects an actual cause-effect relationship among the population variables. For example, Fig. 3 displays temporal co-variation in Child mortality and Children per woman for several countries. All countries display a similar pattern that when Child mortality drops below some threshold (15 – 20% in this example), Children per woman begins to decrease rapidly in a positive association with further reductions in Child mortality. The similar pattern seen in all of these different countries is evidence of a real relationship and that reducing child mortality can cause a reduction in the number of children per woman. Trendalyzer scatterplot of temporal variation in Child mortality and Total fertility (children per woman) for selected countries. Specific countries can be selected by either clicking on their dot within the graphic or by selecting them from the list on the right side of the screen. Dots for the selected countries will remain dark while dots for unselected countries are de-emphasized. The triangle slider below the text Deselect all, just below the list of countries, can be dragged left–right to control the degree to which dots for unselected countries are de-emphasized (to the point of erasing them from this graph). The time trend lines are created when the graph is animated by clicking on the Play button in the lower left corner of the screen. While the scatterplot is animated, the year value in the background changes to display the year of the current display. The Country name / Begin year labels can be moved by dragging them to the desired position. Note: The length of time series varies among countries and among different variables. Hyperlink to Animated Scatterplot (with time trails). The Trendalyzer program uses animation to create graphics that display change over time in the association between two population variables. By default, the scatterplot will display data for the most recent year available for both X and Y variables. Below the X-axis title line is a Year scale with a pointer located at the same year as displayed in the plot area background. To view how the array of points in the scatterplot changes over time, drag this year pointer along the year scale to view data for earlier years. An alternate way to animate the scatterplot is to click on the Play button below the X-axis scale. This will display temporal variation in the pattern of scatterplot points beginning with the earliest year for which data were available. When exploring temporal variation alone, sliding the year pointer along the time scale provides the greatest control. When displaying and explaining temporal variation to an audience, the “hands-off” aspect of simply clicking on the Play button is easier. Click on the hyperlink in the caption for Fig. 1 to access an online version of this graphic that will allow you to experiment with these animation features. Note: I have adjusted the axis scales of this online graphic to include the wider range of variation included in data for earlier years, when child mortality rates were much higher than present. I also specified the begin year as 1947 because this is about the year when data time series for many countries begin. Click on the Play button to see the animation for 1947 to 2010; the dots in the scatterplot will change position and size, displaying year-to-year variation. If you click on the Play button again the entire animation will display variation beginning in the early 1800's when data are available for only a few countries. This animation clearly shows that for all countries in the database, when child mortality was high the number of children per woman was also high. When child mortality declined below some threshold, children per woman also began to decrease. This pattern of decreasing family size with declining child mortality was first apparent in European countries after 1930 and the same pattern is exhibited by even the poorest countries of sub-Saharan Africa in recent years. A similar procedure can be used to animate Map graphics to see temporal change in spatial patterns of variation for a single variable. Click on the hyperlink in the caption of Fig. 1 to access an online version of the scatterplot of Child mortality vs. Children per woman that can be animated. Selecting countries: On the right side of the Trendalyzer graphic, just below the color-coded map of the world, is a list of all countries in the Gapminder database. To select a country, find its name in the alphabetical list and click in the box in front of the name. The dot for that country will remain full-color while dots for all other countries will become semi-transparent and the dot for the selected country will be labeled with a name tag. To re-create the graphic in Fig. 3 select the same set of countries from the list. Controlling transparency of unselected country dots: Below the list of countries on the right side of the Trendalyzer graphic is the option Deselect all and below this is the transparency control. The degree of transparency of unselected dots can be controlled by dragging the Δ icon along the transparency spectrum. If this icon is dragged to the far left end of the spectrum dots of unselected countries will disappear from the graphic, similar to Fig. 3. Visualizing time trends: After selecting the countries, click on the Play button. During the animation a “time trail” of dots for each selected country will be created (e.g. Fig. 3). The dot for the earliest year in the time trail will be labeled with the country name and the year. As the animation proceeds dots for successive years will appear and be connected to dots for earlier years. The time trails display temporal co-variation between the two population variables for all selected countries. You may have to click on Play twice to see the full animation. Also, you may need to use the zoom-in tool to eliminate blank space above and below the time trails, expanding the dots to fully utilize the plot space. The country name tags can be dragged to positions where they do not obscure the time trails, as shown in Fig. 3. It is sometimes useful to graph temporal variation one variable at a time. For example, graphics that display the pattern of temporal change for two variables (e.g. Fig. 3) can produce erratic results if countries are impacted by conflict or immigration that cause rapid changes over short periods of time. Simple time plots for a single variable of interest on the Y-axis and Year on the X-axis can sometimes provide a better basis for evaluating temporal variation (e.g. Fig. 4). To create a time plot, select the variable of interest for the Y-axis and select the variable Time for the X-axis (located at the bottom of the main variable menu). Select the countries you want and adjust the transparency control to eliminate dots for unselected countries. Click on the Play button to create the time plot. Trendalyzer time plot of population growth rate for selected countries that display different levels of temporal variation (Time was selected from the Main menu as the X-axis variable). Dots for unselected countries have been de-emphasized. Smooth time trends (Bangladesh) suggest gradual changes in birth and death rates. Large, rapid, short-term fluctuations (Eritrea) indicate changes in population growth rate are more likely due to population displacements associated with warfare or other civil strife. Rapid threshold changes between two stable time trends (Iran) suggest major public policy shifts. The exact year when a break in the trend line occurred can be determined by sliding the Year pointer in the time scale below the plot back and forth. The Zoom-in(out) tool in the lower right corner of the plot area was used to adjust the range of the X- and Y-axis scales so the dots fully occupied the plot area. This graphic was created using Gapminder Desktop, the version of the database and Trendalyzer that is downloaded to a personal computer. This program does not have built-in internet linking capability but does allow you to run the program independent of the Internet. Gapminder Desktop also lacks the How to use and Share graph features of the online Gapminder World program. Accessing information about variables: If the mouse cursor is placed over a variable name in the axis title of a graph, a box will appear that provides an abbreviated variable definition. To access full definitions of variables and other meta- information associated with the data, click on the small “grid” icon below the axis title line, at the beginning of that line. This will take you to a spreadsheet page that displays data values for all countries (rows) and years (columns) for that particular variable. At the top of the spreadsheet grid are several hyperlinks. The About link provides definitions, sources and other meta-data. In the online version, a list of all Gapminder variables can be accessed via the DATA tab at the top of the Gapminder Home screen. Each variable name in the DATA table is a link to associated information. Exporting Trendalyzer graphs: Trendalyzer does not have an export function that allows copying the graphics into documents or presentations. However, graphics can be copied and pasted using screen capture features of Windows 7 and Apple operating systems. The graphics included in this paper were captured using the Windows 7 Snipping Tool. Apple computers have a similar screen capture utility, but in my experience the resolution of the graphics captured using the Windows tool was superior. Alternatively, you can click on the Share graph button available in the online version of Trendalyzer. This will save your graph to an online storage area and provide you with a unique alphanumeric string that is a URL to that graph exactly as is existed at that point in time. You can send this hyperlink in an email or embed it in a document or PowerPoint presentation. Both of these procedures are modeled in the figure captions of this paper. Exporting Gapminder data: While it is possible to view all of the raw data values in the Gapminder database, it was not possible to download these data for analysis by other software when this paper was being written. However, the database managers indicate that a download capability is under development. Access to this download feature will be through the “Grid” data icon below each axis title in the Trendalyzer graph. Above the spreadsheet of data values is a Download link that will provide options for downloading data in various formats, including Excel, CSV (comma delimited text) or PDF. Given the large number of variables in the Gapminder database, there are many associations that can be explored. In this section I describe how this resource can help students learn about factors driving human population growth. The context of these exercises would be appropriate for an introductory course in ecology or environmental science. These exercises allow students to explore the ecological concepts of population dynamics (survivorship, fecundity, net reproductive rate, generation time, population growth rate). These exercises would typically be assigned to students after they have been introduced to these concepts and the associated calculations for these variables. It is important that students understand these ecological variables well because the human demographers who created the databases used in Gapminder use different variables that are conceptually related, but not identical. For example, the Gapminder variable Children per woman (total fertility) is similar to Net reproductive rate, but not the same. Likewise, the Gapminder variable Age at first marriage is related to Generation time (at least in traditional cultures), but is clearly not the same. Hence, students must make the connections between the variables they have learned in the population ecology unit and related Gapminder variables. Detailed documentation for Gapminder variables can be accessed as described above and students can compare and contrast the Gapminder variables and the standard population ecology variables. One operational definition of “understanding” is that students can think and act flexibly with what they know, applying knowledge in contexts different from that in which learning originally occurred (Harvard Graduate School of Education, Active Learning Practices for Schools). Students can demonstrate their understanding of core population ecology concepts by connecting them to similar variables in Gapminder. Because the temporal and spatial patterns of association are highly variable among the many countries in the database, students in a class can do these exercises independently (each analyzing data for a different country). I have students work in small groups that are assigned a specific region of contiguous countries that have similar climate, natural vegetation (biome), religion, etc. Each group pools results from individual country analyses to identify similarities and idiosyncratic variation and come to some conclusion or policy recommendation, which they must then defend based on their data analyses. They are also asked to make specific connections between the results of their analyses and information they have found from various sources (print, web-based) that are relevant to their question and their specific geographic region. Question 1: Population growth rate is the result of the difference between birth rates and death rates and between immigration and emigration rates (r = b – d + i − e). Gapminder provides data for birth rate and death rate, but not for immigration or emigration. However, the influence of migration can be inferred from changes in population growth that are not explained by changes in either birth or death rates (by process of elimination). For specific assigned countries, which variable is most strongly correlated with population growth rate (i.e., has the strongest potential influence on population growth), birth rate, death rate or are these two variable equally correlated with population growth? Use plots of each of these three variables over Time and scatterplots of Birth Rate (X) vs. Population Growth Rate (Y) and Death Rate (X) vs. Population Growth Rate (Y) to answer this question. Note: For any question that attempts to use correlation analysis to identify “influence,” a discussion of the difference between correlation and causation should be required. Question 2: The Demographic Transition Model describes a temporal change in birth rates, death rates, and population growth rate and size between conditions of high birth and death rates (with stable population size) and low birth and death rates (with stable population size). During the transition, death rates drop first followed by a drop in birth rates. Because death rates are lower than birth rates for most of the transition period, population size increases. For assigned countries, evaluate temporal variation in birth rates, death rates and population growth rates and determine to which stage of the Demographic Transition Model each country can be assigned. Combine results for multiple countries in the same geographic region (selected to have similar environment, culture, and religion) to characterize that region. Population growth rate is reduced if the number of children per woman is reduced and(or) if the age of women when they are reproducing is increased. Discussions about reducing human population growth often focus on how to reduce the number of children that women produce over their lifetime (net reproductive rate, total fertility rate, “family size”). Many factors influence women's decisions about how many children they desire and how many they actually produce, including child mortality rate, female education, female job opportunity, women's age at 1st reproduction (marriage), and availability/use of contraception. For assigned countries in a geographic region, evaluate the temporal correlations between Children per woman (total fertility) and Gapminder variables related to those listed above. Describe similarities and differences (if any) among the countries within your region. Identify the variables most strongly correlated with historical changes in total fertility. Compare the current values for these variables in your assigned countries to the values typical of developed countries that have already attained stable population size to identify “mission accomplished” vs. “needs improvement” situations. Based on your data analysis recommend policies for manipulating the “needs improvement” variables to reduce total fertility rate and population growth rate. These policy recommendations must be non-coercive (do not require individuals to change their behavior, but may reward desired behaviors) and socially-acceptable (policies that you would be willing to accept for yourself, while recognizing potential for cultural differences). Note: Assigned readings given to students provide different perspectives on the relative efficacy of different policies and on how variation in the status of women among different countries influences fertility rates. For any of these different exercises, student groups could present results of their analyses to the entire class and explain how they support their conclusions and decisions. This would give the students the opportunity to explain how they analyzed data and identified relevant information from published sources to develop knowledge of their assigned geographic region and how they used this knowledge to develop their judgment regarding appropriate policy recommendations. The presentations from various groups give the class a sense of the range of variation among countries and regions that differ with regard to history, culture, religion, economic development etc. The modality of this sharing could be oral presentations to the class, PowerPoint presentations from student groups made available on a course web site or a class poster session. Rules could be established that require students to be selective in what is included in their presentations, forcing them to distinguish between information that is critical to justify their conclusions and information that might be interesting but not essential. The Gapminder exercises described above can provide an engaging context for developing student critical thinking skills. One definition of critical thinking is, “An intellectually disciplined process of actively and skillfully conceptualizing, applying, analyzing, synthesizing, and/or evaluating information gathered from, or generated by, observation, experience, reflection, reasoning, or communication, as a guide to belief and action. In its exemplary form, it is based on universal intellectual values that transcend subject matter divisions: clarity, accuracy, precision, consistency, relevance, sound evidence, good reasons, depth, breadth, and fairness” (Scriven and Paul 1987). The exercises described above require that students analyze data to draw well justified conclusions, relate their results to findings of others from online and published sources and, in some cases, formulate proposals to address real world problems that can be defended based on their analyses. These exercises touch on experiences that are virtually universal across all human populations, including life, death, childhood, parenthood, aging, sex, reproduction, marriage, education, and work / employment. Our students bring prior experiences, opinions, and assumptions about these universal human experiences that influence their world view and their inferences about how and why people in other countries and cultures live the way they do. Many of these prior assumptions and inferences are based on ignorance and incorrect. The combination of data analyses and readings that offer accurate information and alternate world views will challenge students to re-evaluate their naïve assumptions and inferences about people in other countries and cultures and to formulate new ideas about how the world works. This aspect of the Gapminder exercises goes well beyond the usual scope of an ecology or environmental science course (and well into the disciplines of anthropology and sociology, among others). However, purposefully including discussion of these social dimensions of human population growth adds a rich context for developing critical thinking skills to address a compelling, real-world problem. Students will need to formulate clear questions, identify and apply relevant information to develop answers that are supported by available facts, consider how alternate world views might lead others to different answers based on the same facts, evaluate the implications and practical consequences of these different world views, and communicate these differences in a manner that is logical, clear and respectful. That is, the Gapminder exercises described above give students an opportunity to practice most of the skills that define critical thinking (The Critical Thinking Community 2011) and to apply the ecological knowledge they have gained to addressing a major real-world problem.