We based our research design on the model of educational reconstruction (MER). The MER was initially developed as a model for instructional planning in school settings and for curriculum development (Kattmann et al., 1997). This model soon proved to be useful in a much wider scope of applications, and became an important framework for research and development in science education (Duit, 2007; Duit et al., 2012; Reinfried et al., 2009). The MER has since been adopted as a major theoretical perspective in science education research by various science education groups in Europe.

The MER is based on a constructivist epistemological approach. A balance between science-related and education-oriented issues is considered a necessity for effective teaching and learning. The primary focus of science-related teaching (e.g., in university lectures) tends to be on the scientific nature of a certain topic. Following scientific conventions and routines, generations of teachers used to present scientific content in a simplified (reduced) manner in science instruction, but the MER focusses on a quite different approach: the key message of this model for education-oriented teaching is that a new structure for science instruction has to be found in an iterative process between the analysis of the scientific content and learners' perspectives, preconceptions, and experiences.

The MER integrates three significant components of science education research: (1) the clarification and analysis of scientific content, (2) research on teaching and learning, with a particular emphasis on the role of students' pre-instructional conceptions in the learning process, and (3) the design and evaluation of teaching and learning environments (Duit, 2007; Duit et al., 2012). In our study, all three components were applied (see Fig. 1): we took into account the definitions pertaining to the topic of hydrogeology, and the interpretation of the research results regarding pupils' and students' conceptions of groundwater. Based on these, we devised the design of our multimedia learning program. The ascertainment of the effectiveness of our multimedia tool began with an examination of the groundwater concepts of our target groups in order to investigate the extent to which conceptual change and knowledge gain was possible by working through the learning program.

Numerous studies have shown that children come to class with a wide variety of preconceptions in relation to scientific concepts, many of which are inadequate (e.g., Vosniadou, 2013; Hammann and Asshoff, 2014; Kattmann, 2015). Everyday preconceptions are often resistant to change, especially if they appear to be intuitively correct. Because people are familiar with these preconceptions and they have become firmly established in everyday life, they are often considered to be adequate or at least not harmful. Preconceptions or “framework theories” (Vosniadou, 2014) are abstract, naive knowledge structures resulting in deep ontological commitments in terms of how we understand the world. They can impede knowledge restructuring and be resistant to change.

Conceptual change theory is widely accepted in science education, and numerous studies have led to remarkable insights into the thought patterns and conceptions of children and adolescents in various subfields of science. A number of studies show that new information is incorporated into existing ideas for as long as possible and thus retained, even if there are obvious contradictions. Researchers agree that it is one of the most important aims of science instruction to develop students' pre-instructional conceptions towards the intended scientific concepts. Vosniadou (2014) holds that these framework theories do not seem to disappear, but continue to exist and interfere with access to scientific concepts, even among skilled adults. Therefore, from a constructivist point of view, science learning cannot be understood as the replacement of an incorrect by a correct concept (Vosniadou, 2007). Referring to these complex learning processes, Duit and Treagust (2003) and Kattmann (2005) preferred to use the term “conceptual reconstruction” instead of “conceptual change”.

With respect to groundwater, research has shown that common conceptions of groundwater are seldom based on scientific findings and that there is a strong prevalence of incorrect hydrogeological concepts. The following represent dominant preconceptions (Dickerson and Dawkins, 2004; Dickerson et al., 2005, 2007; Ben-zvi-Assarf and Orion, 2005; Reinfried, 2005, 2006a, b; Schultz, 2006; Schwartz et al., 2011):

groundwater is stored in underground lakes;

In their study of 17- and 18-year-olds, Dickerson et al. (2005) asked for an indication of size in order to better classify the conceptions of these adolescents. Over 60 % of respondents imagined groundwater lakes and rivers to be similar to water bodies on the surface of the earth, and to be of considerable size (see also Cheek, 2010).

The idealized notion pertaining to the quality of groundwater is also worth mentioning. Reinfried (2006b) and Reinfried et al. (2012) reported from their research involving 13-year-olds, that many of the respondents generally believed that groundwater, and especially spring water, was clean and drinkable. According to Suter et al. (2007), this notion is also shared by adults. There appears to be a lack of awareness concerning threats to groundwater quality and its conservation.

The abovementioned misconceptions of groundwater as an underground lake, river, or accumulation of water in cavities are persistent and outlast academic tuition. Groundwater is an abstract phenomenon that is neither visible nor can it be experienced. It therefore tends to be explained by means of well-known structures and occurrences above the surface of the earth. Aside from this tendency to explain the world by means of analogies, we also often resort to metaphorical explanations. In keeping with the theory of experience-based understanding from Lakoff and Johnson (2003), for example, we frequently refer to water veins in the ground in analogy to the veins transporting blood through our body.

These metaphors and body-related constructions can also be traced throughout historical conceptions of groundwater: as early as 2500 years ago, Pythagoras described the earth as resembling the human body, and Leonardo da Vinci and Johannes Kepler compared the earth's water to the blood of an organism (cf. Reinfried, 2006a, p. 54; 2006b, 40–42). The idea of an underground water network existed up until the mid-19th century (subaerial river model), and it was not until the beginning of the 20th century that the present-day conception was established. In colloquial language, however, millennia-old metaphors persist regardless of modern geological knowledge.

These metaphors are reinforced by mainstream popular science television, literature, and textbooks. Without much reflection on the consequences, some authors display an aquifer in the geologic tradition as a homogenous blue area, which is then interpreted by laypeople in the sense of the abovementioned misconceptions (Schwartz et al., 2011). Inadequate or incorrect visual representations of groundwater in textbooks further impede the development of scientifically accurate concepts. Shepardson et al. (2009) criticized the prevailing misrepresentations of the water cycle in American textbooks, where water is displayed as a stylized landscape with mountains and coastlines. As many pupils are unable to relate these images to their actual surroundings, such representations are impractical for conveying a deeper understanding of the water cycle and the role of groundwater. Reinfried (2006a) also sees pictures in textbooks as a source of misunderstandings. Arrows depicting the groundwater movement from land to sea, for example, could be interpreted by pupils to represent rivers or water veins. Wampler (1998, 2000), Dickerson et al. (2007), and Duffy (2012) also identified illustrations, which are either too simplified or downright negligent. As our recent analysis of 23 textbooks confirms, all of these criticisms can also be applied to Austrian textbooks.

Teachers are not always capable of compensating for the shortcomings of textbooks as their own conception of groundwater is often similar to the preconceptions of their pupils (Dickerson and Dawkin, 2004; Duffy, 2012). In their study conducted as part of the Arizona Water Festival within the scope of a school program in 2009, Schwartz et al. (2011) discovered that pupils performed better when their teachers had taken part in a training workshop on the subject.

The multimedia learning software deals with various questions concerning groundwater in unconsolidated rocks where it occurs in the pores between the mineral grains. In order to develop an adequate model of groundwater (cf. Hölting and Coldewey, 2013; Davis and de Wiest, 1966; Hilberg, 2015), adolescents need to understand the following:

Rainwater seeps into the ground through cavities between mineral grains, and accumulates in permeable and porous sediments above an impermeable layer. The characteristics of the pore space, and therefore its suitability as a groundwater aquifer, depend on the grain size. Larger grains constitute larger pore spaces while smaller grains are surrounded by smaller pore spaces. It generally applies that the more pore space available, the more groundwater can be transported and stored therein. Very small grain sizes (silt and clay) constitute pore spaces that are too small to allow water to percolate and hence form an aquiclude.

Theories of multimedia learning (Mayer, 2009; Moreno, 2006) constitute an important basis for designing such a learning program. One of their key messages is that meaningful learning can be promoted by taking into account the architecture of human information processing and the characteristics of the working memory. Mayer (2005, 2009) and Mayer and Moreno (2003) recommended several principles of multimedia learning, which we adopted when designing our multimedia learning program: we implemented a good balance between auditory and visual presentations of information. The texts are kept short (no scrolling) and the criteria for comprehensibility according to Langer et al. (2011) were taken into consideration in the text presentation. With regard to motivation, a geologist guides the user through the program in the role of a “pedagogical agent” (Mayer, 2005). She offers explanations, asks questions, and gives instructions for the interactive tasks as well as feedback on the test questions.

Experiences and results of studies with other multimedia learning programs on biological topics were also taken into consideration (Unterbruner and Unterbruner, 2002, 2005; Unterbruner et al., 2008). The learning program is characterized by a clear structure and a row of information units followed by test questions. Three test questions conclude each thematic sub-unit and are designed to give users feedback on how well they have grasped the learning contents, and to fuel their motivation. Working through a chapter takes between 15 and 20 min.

The program is interactive, cognitively activating, and devised to be worked through independently. Cognitive activation is to be achieved by means of a problem-oriented approach on the one hand (e.g., Unterbruner and Pfligersdorffer, 2007; Zumbach et al., 2014), and through interactive elements on the other. Various interactive elements require the user's active participation, for example by using a magnifying glass to enlarge smaller details.

In accordance with the iterative approach of MER, we based our theoretical considerations regarding the design of the multimedia learning program on the investigations about students' pre-conceptions by Dickerson and Dawkins (2004), Dickerson et al. (2005, 2007), Ben-zvi-Assarf and Orion (2005), Reinfried (2005, 2006a, b), Shepardson et al. (2009), Schultz (2006), and Schwartz et al. (2011) on the one hand, and on basic hydrogeological concepts on the other. In the latter case, we focussed on the most relevant scientific aspects to our target group of young people, who may never have dealt with the topic of groundwater before. Accordingly, results from conceptual change research by Strike and Posner (1992), Vosniadou (2007), and Sinatra (2005) broadly influenced our storyboard design. The dramaturgy of the multimedia program/storyboard will be described in detail below (see Table 1).

First of all, we decided not to start our program by activating preconceptions and previous knowledge in order to avoid reinforcing existing misconceptions (cf. Sinatra, 2005). As a primary problem, we identified that most people have no concrete notion or, at best, a very vague idea of the structure and composition of the ground (i.e., weak coherence; cf. Sinatra, 2005). Groundwater may be an abstract phenomenon, but contrary to the issue of climate change, it can sometimes become quite tangible (e.g., in building trenches). However, we assume that most people do not make the connection between the observation of these phenomena and groundwater. Our primary aim, i.e., to convey an accurate understanding of groundwater, thus requires the best possible visualization of the composition of the “ground beneath our feet”. Therefore, the program begins with conveying said knowledge, but without making any direct reference to groundwater at first. The subject of groundwater is subsequently developed based on that knowledge. We accordingly developed a dramaturgy for the storyboard based on the following central questions (cf. Hilberg, 2015):

What makes up the ground beneath our feet?

In line with the international studies described above, the results of the drawing exercises from the pre-test showed that the dominating preconceptions of students and pupils were the academically incorrect conceptions of an underground river (students: 30 %, pupils: 47 %) and an underground lake (students: 31 %; pupils: 15 %). Other concepts were rarely mentioned. The scientifically accurate conception of water within porous and permeable rocks was drawn by 11 % of students, and only 3 % of pupils (see Table 3).

In their verbal descriptions, 60 % of pupils vs. 89 % of students described the concept correctly. This discrepancy can be attributed to the fact that most of the verbal descriptions of groundwater provided were very short and generic (e.g., “Water in the ground”), and did not express nor allow conclusions as to the underlying conceptions.

The scientifically adequate concept of groundwater was significantly more prevalent in the post-test. The percentage of correct and partially correct drawings rose from 9 to 42 % for pupils and from 20 to 49 % for the students. The evaluation of the graphical representations produced by the participants showed a statistically highly significant shift from inadequate preconceptions to the correct conception. An evaluation of the verbal descriptions of groundwater yielded similar results, although from a much higher baseline (Fig. 7, see also Table 3).

When examining the preconceptions of the underground river and lake in detail, the Wilcoxon test showed that these perceptions were significantly reduced in the post-tests for both pupils and students (Fig. 8).

The degree to which this effect can be attributed to the effectiveness of the multimedia learning program becomes evident in a comparison of the experimental and control groups. The concept scores (i.e., sum of points achieved in the concept tasks, max. 4) of the pre- and post-tests of both groups were calculated and analyzed. This showed a significant improvement in the scores of the participants from the experimental group while the scores of the control group saw a slight decrease (pupils – experimental group: +1.20 points, control group: -0.03 points; students – experimental group: +1.27 points, control group: -0.07 points) (Fig. 9).

In order to verify the overall increase in knowledge, all items testing knowledge were combined to a total knowledge score. Every correct answer was worth 2 points, resulting in a maximum total knowledge score of 24 points in both the pre- and the post-test. The overall increase in knowledge (or decrease, as the case may be) was determined by the difference between the total knowledge scores from the pre- and post-test.

A comparison with the control group was once again used to show that the increase in knowledge was, in fact, attributable to the use of the multimedia learning program. On average, the scores of students from the experimental group increased by 3.29 points while those students from the control group only achieved an increase by 0.89 points. The ANOVA revealed a highly significant difference between the two groups (F(1, 86) = 12.35; p < 0.01; η2 = 0.13). In the case of the pupils, the experimental group achieved an increase by 5.31 points compared to 3.82 points (F(1, 120) = 5.88; p < 0.05; η2 = 0.05) in the control group.

Regarding the fundamental geological concepts of porosity and sediments, the increase in knowledge was shown to be particularly high in both experimental groups. Pupils and students performed best in regards to sediments, flow rates in gravel, sand, and clay and in depicting the groundwater surface.

The ANOVA also showed that the participants in the experimental group were significantly more confident in their answers in the post-test compared to the participants of the control group.

We also examined whether the increase in knowledge varied between participants with a higher level of prior knowledge compared to those with little or no prior knowledge. In the experimental groups of both pupils and students, we observed that participants with little prior knowledge achieved an increase in their knowledge scores in a significantly greater number of instances than those who possessed a higher level of prior knowledge to begin with.

The multimedia learning program was evaluated very positively. From a maximum of 60 possible points (4 points per item) in the evaluation questionnaire, the average score given by pupils was 51.4 points (s = 5.47) while students gave an average of 55 points (s = 3.87).

The results of the individual scales related to interest, comprehensibility, enjoyment, subjective achievement, and usability are summarized in Table 4.