Introduction

There is a significant agreement among science faculty about the positive impact that the implementation of evidence-based pedagogical approaches and supportive learning environments have on student learning. Many science faculty have already adopted different forms of active learning in the class in order to increase student success, engagement, and retention. On the other hand, the content for science courses have not yet changed at the same rate. In other words, we changed “how to teach”, but we have not significantly changed “what to teach”. In order to revisit and improve what science students should learn, the President's Council of Advisors on Science and Technology (PCAST)¹ and the National Research Council² called instructors during the first two years of college to focus on deeper and transferable knowledge.

As a response to this call to action, the Next Generation for Science Standards (NGSS) has provided a very detailed framework and assessment tools for such implementation in K-12, which they can be easily applied to introductory science courses in college³.  Following the standards provided by the NGSS, instructors should focus on core ideas, crosscutting concepts and scientific practices. Activities that include these three aspects aim at what it is has been called the three-dimensional learning⁴.

The aim of this paper is to shed some light on how instructors can successfully address several scientific practices during the first semester of a General Chemistry course in one of the most prevalent Chemistry’s core concepts, the “Atomic/Molecular Structure and Properties” or what is commonly known as structure-property relationships; in particular, how can ChemEd X Data website best help in some of the most open-ended or data-driven types of questions. The several student activities presented and analyzed here pertain to different topics and they are spread throughout a typical first semester of General Chemistry and should not require any major curriculum redesign.

 

The tool: ChemEd X Data website

 

ChemEd X Data^5,6 (http://chemdata.r.umn.edu/chemedXdata/) was created to help students practice activities that address scientific practices, during introductory courses of chemistry. More specifically, ChemEd X Data was designed for chemistry students to navigate, filter and graphically represent chemical and physical data. The kind of activities that ChemEd X Data offers to students must be placed within the data-first or data-driven learning approach^7,8. Even though data-first activities, data-driven activities and problem-based learning can sometimes be used interchangeably, there is some variability in terms of how data are made available to students. On one end, the data may be a table of already pre-selected values. In many of these cases the activity will be more focused on interpreting and manipulating the given data rather than navigating, plotting and choosing them. Therefore, it may exclude the possibility for open-endedness. On the other end of the spectrum, the student could be offered an unstructured spreadsheet of data or access to a database or a web portal such as Pubchem or Wikipedia^9,10. In the latter case, the “open-endedness” may overwhelm students, or their performance may rely too much on their information literacy skills. The ChemEd X Data website tries to find a balance between the two extremes, where a large quantity of unstructured data is available while still trying to lower the technical barrier by making it easy to navigate, sort, rank and instantly plot different types of physical and chemical data. The three figures below show snapshots of ChemEd X Data most basic capabilities.

Figure 1 Flexible selection of compounds and their properties to discover trends in structure-property relationships: boiling and melting points, heat capacities, enthalpies of combustion, as a function of mass, shape, functional group…

 

Figure 2 Upon selecting a set of compounds, the user may click on the “View 3D!” button to visually explain the observed physical properties with molecular properties such as symmetry, charges and dipoles.

 

Figure 3 Navigate the periodic table to identify periodic trends or trends among inorganic compounds across a period or a group.

Implementation

 

The results presented in this paper were collected during the fall semester of a “General Chemistry 1” course at the University of Minnesota Rochester (UMR). The course was structured as a “flipped classroom”¹¹. In this type of delivery, students were assigned to watch three or four videos (with a length of approximately 5 to 10 min each) and turn in the answers to a series of preclass questions. However, the days when students were asked questions related to creation of knowledge no videos nor any pre-class questions were assigned in regards to the knowledge they were going to create. The majority of students were in their sophomore year. UMR only offers two Bachelor of Sciences, both related to Health Sciences, and students must take at least one semester of introduction to Organic structure during their first year before enrolling in General Chemistry. This curriculum approach is called “organic first”¹² and it is intended, in part, to improve their quantitative skills during their first year of college before enrolling in General Chemistry¹³. This approach also means that students are already very familiar with organic functional groups and organic molecular structures, which allowed the design of activities involving a wider range of organic compounds. Instructors at other institutions without an organic-first approach, should reformulate the activities that match students’ previous knowledge of molecular structure.

All the activities were delivered using the Moodle Learning Management System and students completed them individually either during class time or as homework outside of class. Some of the exercises required the use of ChemEd X Data but not all. Thanks to the laptop program, all students had a laptop during class time.

 

A common sequence of questions for several topics

Students completed a set of activities that pertain to three different topics with structure-property relationships, namely, periodic table trends, effect of size and charge on ionic bond, and factors affecting the boiling point of molecular solids. All these topics however, allowed us to build a common sequence of questions or progressions that students answered in those three different topics. The five-question sequence or progression can be simplified as:

1.     Identify molecular factors

2.     Connect structure and property by investigating data

3.     Predict and apply relationship

4.     Explain observations

5.     Conflicting factors and identify limit of predictability
 

Students in the past had special difficulty in questions two and five. Number two is necessarily open ended and hard to implement in a regular class, and number five is somewhat counterintuitive to students, therefore they may deserve a deeper justification.

 

Investigating data: Control of Variable Strategy

This activity consists of being able to select experimental data to show evidence of a specific trend or law, or what is commonly known as designing a controlled experiment or control of variable strategies (CVS)¹⁴.  See the work by Schwichow et al¹⁴ for an excellent overview on the pedagogical background in control of variable strategies. This activity represents one step further from what is commonly known as data-first problems⁸. In data-driven activities students are typically shown a specific arrangement of data. In the CVS case, however, students select the substances they want to represent, they must build a controlled experiment by keeping the rest of variables constant and varying the only one being tested. Therefore, there is not a unique valid answer and the problem becomes open-ended. As it has been previously described⁵, ChemEd X Data was specifically designed to practice this last kind of activity. Table 2 shows an example from students of a correct and incorrect selection of molecules to show evidence of the effect of linear/branching on the boiling point of organic molecules.

 

Table 1 An example of a correct (first row) and an incorrect (second row) selection could be when students are presented with the question “Select 4 compounds to show evidence of how the linear/branched chain may affect boiling points”.

                                                                                                                   

 

Conflicting factors and limit of predictability

In General Chemistry, there are several curricular examples where there is typically more than one factor deciding an outcome. In introductory levels, the instructor may decide to either avoid examples with conflicting factors or have students learn to recognize the limit of predictability at a given level of theory. For example, students may learn that hydrogen bonds are stronger than London forces, but that is only true if other factors such as mass and shape are kept constant. Several works have been published on how to better understand and teach this topic^15–18. If students are asked to predict whether methanol or hexane have the highest boiling point, they should first recognize that mass and intermolecular forces (IMF) are in conflict and that, at this level of theory, they do not have enough tools to predict an outcome. Alternatively, one can also reverse the question; if students are told that hexane has a higher boiling point than methanol they should be able to recognize that the mass is playing a larger role than the strength of intermolecular forces. This pedagogical route may be the recommended one. It is a good opportunity to address the scientific practices “Analyzing and interpreting data” as well as “Constructing Explanations and Engaging in Argument from Evidence”³ in introductory courses and it potentially eliminates any misconceptions or frustration that some students may feel when they realize that they are not being told “the total truth” or that chemistry has too many exceptions^19,20. Table 1 lists the different cases in a typical “General Chemistry 1” semester where this conflict among factors is given.

 

Table 2 Topics typically present in a first semester of a college General Chemistry course where more than one structural or energetic factor may affect the property. The different factors may potentially be in conflict with each other.

Property to be predicted	Several factors that may come into conflict	Examples where content in introductory courses does not predict an outcome
Melting point or heat of fusion of molecular solids	Intermolecular forces, molecular shape and mass	Highest boiling point? Methanol vs Hexane
Periodic table trends: atom size and ionization energy of elements.	Nuclear charge, electronic shielding and orbital size.	What element has a larger radius among diagonal elements? (Li is bigger than Mg but O is smaller than Cl)
Ionic lattice energy of a crystalline solid	Cation and anion size and their ionic charge	What compound has a higher ionic lattice energy, KF or NaCl?
Electronic configuration of elements (ground state)	Hund’s rule and Aufbau principle	For chromium and molybdenum, the ground state is “s1 d5”, but for tungsten is “s2 d4” and carbon is “s2 p2”
VSEPR theory: molecular angle	Lone pairs and double bonds affecting the regular angle	Molecules where lone pairs and double bonds are present in the central atom: SO2, O3, NOCl
Heat capacity of molecules	Molecular mass, number of covalent bonds and intermolecular forces.	Highest heat capacity? propanol (more mass) vs butane (more bonds)
Miscibility of two liquid substances	Intermolecular forces and molecular shape	What is more soluble in water, butanol or dichloromethane?

 

 

Results and Discussion

 

The analysis of student performance in the “structure-property relationship” activities may be enriched if it is compared with students’ performance at the end of the course as well as in previous college courses.

As the figure below shows, three very distinct groups of students can be identified when representing “final course grade” vs “College GPA held at the beginning of the semester”. These three groups labeled as low performers (LP, black squares), intermediate performers (IP, green circles), and high performers (HP, red triangles) are the three clusters obtained using the k-means clustering analysis as it is implemented in the R package.  All the points in the graph correlate fairly well to a linear behavior, with a Pearson’s product-moment correlation coefficient of 0.807. As it has been noted elsewhere²¹, albeit not being very informative in regards to students profile, the current students’ GPA is a good predictor for future course performances. These three groups of students, LP, IP and HP, will be used throughout the analysis of activities presented below.

Figure 4 The clustering of “students’ GPA at the beginning of the course” vs “course final grade” at the end of the course gives three groups of students labeled as Low Performers (LP, black squares), Intermediate Performers (IP, green circles), and High Performers (HP, red triangles).

The next three subsections present the graphical analysis of students answers to the sequence of the five questions in three General Chemistry topics: Periodic Table Trends, Ionic Lattice Energy of Crystalline Solids, and Boiling Point of Molecular Liquids. As it is indicated in the example questions, students had to use ChemEd X Data in some parts of the activity, but not all.

Periodic Table Trends

 

In this first activity, students investigate the periodic table trends, and answered a set of questions that follow the common progression indicated above. For space reasons the entire questions are not shown verbatim.

 

1)    Identify Atomic factors: Identify Z, shielding and orbital size.
Example: The charge of oxygen nucleus is ___, and the charge of carbon nucleus is___, the number of electrons of neutral oxygen is ____ …., oxygen valence electrons are in n=____...

2)    Connect Structure and Property.
Example: Use ChemEd X Data and choose a series of elements that show how the element radius is affected by the nuclear charge.

3)    Predict and Apply Relationship: Explain factors for regular PT trend
Example: Explain what factors make oxygen smaller than carbon.

4)    Explain Observations: Explain factors for diagonal elements
Example: Use ChemEd X Data to explain that Oxygen is smaller/bigger than Chlorine because Oxygen has
higher Z/lower shielding/smaller valence orbitals (choose the correct answer).

5)    Identify limit of predictability
Example: Which pair of elements present a conflict of factors that does not allow you to predict what element is bigger: Choose one or more. “Li vs O”, “O vs Mg”, “Li vs Mg”, “O vs Cl”

 

Figure 5 Sankey diagram showing the flow of students’ answers to a series of questions related to periodic table trends. The last column labeled “College performance” shows the fraction of students belonging to High Performers (HP), Intermediate Performers (IP), and Low Performers (LP) that answered correctly or incorrectly the last question on “identifying the limit of predictability”.

In the Sankey diagram of the figure above, each column represents a question in the activity. The horizontal stripes connecting the columns indicate the fraction of students who answered that specific question correctly (green) or incorrectly (red). For example, in the first question labeled as “Identify molecular factors”, the vertical line tells us that about 70% of students answered it correctly. Moving to the next question, labeled as “Connect structure and property”, that initial 70% of correct answers increases to 85%. One of the main advantages of this diagram is that it shows the flow of students’ answers among adjacent questions.

The last column in the figure above is not a question on this activity, it is a first attempt at understanding how performance in higher-level activities may relate to course performance and college GPA. In this case, because of the limitations of the Sankey diagram, we can only see what portion of students who obtained correct answers in the last question “Identify limit of predictability” belonged to each of the three college performance levels, LP, IP, and HP. The first thing that the results tell us is that students do not need to be High Performers (HP) to perform well in these specific higher-skill activities, but they need to be at least Intermediate Performers (IP).

 

Ionic Lattice Energy of Crystalline Solids

 

In this second activity related to how the charge and size of ions affect the energy of the ionic lattice, students answered a set of questions that follow the common progression indicated above. Again, for space reasons the entire questions are not shown verbatim.

1)    Identify molecular factors: ion charge and trends in size
Example: The size of Na is smaller/larger than the size of Cs. The charge of ion Na is smaller/larger/equal than ion Cs… (circle the correct option)

2)    Connect structure and property: identify the role of size and charge in ionic lattice energy.
Example: Based on your observations in ChemEd X Data, the smaller the ion the stronger/weaker the ionic lattice is (circle the correct option)

3)    Predict and apply.
Example: When comparing NaF and CsI, what compound will have a stronger ionic lattice?

4)    Explain observations.
Example: The energy of the ionic lattice for KF is larger than for LiF. Can this observation be explained by this theory?

5)    Limit of predictability.
Example: In what cases the theory cannot clearly predict an outcome?
1) Comparing LiF and NaCl 2) Comparing KCl NaBr 3) Comparing KF and LiF

Figure 6 Sankey diagram showing the flow of students’ answers to a series of questions related to the ionic bond. The last column labeled “College performance” shows the fraction of students belonging to High Performers (HP), Intermediate Performers (IP), and Low Performers (LP) that answered correctly or incorrectly the last question on “identifying the limit of predictability”.

 

Boiling Point of Molecular Liquids

The last of the activities, the boiling point of molecular liquids, is probably the topic where ChemEd X Data can be more useful. In the two cases above, periodic table trends and energy of the ionic lattice, there is a more limited dataset that students can explore, just few elements and few periods of columns. Therefore, it is not until this present topic when some of the exercises are truly open-ended and each student uses a completely different dataset to justify a trend or an exception. A sample of the sequence of questions is the following

 

1)    Identify molecular factors: Connect functional groups and IMF
Example: Identify each of the following functional groups with the intermolecular forces they present (H-bond, dipole-dipole or London forces)

2)    Connect structure and property. Create relationship: building CVS
Example: Use ChemEd X Data to select a set of compounds that show evidence of how mass affects the boiling point of molecular liquids.
(see next section for a more detailed description of these activities)

3)    Predict and apply relationship. Rank functional groups by boiling point.
Example: Based on your investigations, rank the strength of intermolecular forces as they affect the boiling point of their compounds: H-bonds, London forces, dipole-dipole interactions.

4)    Explain observations: mass or dipole: CH_xCl_y vs CH_xF_y
Example: The link below plots the boiling point of CH₃Cl, CH₂Cl₂, CHCl₃, CCl₄.
http://chemdata.r.umn.edu/chemedXdata/#stamp=1414679798226
In this series, what is the property that controls over the others.  Is it the dipole? That is, the most polar has the highest boiling point? Is it the mass? That is, the most massive has the highest boiling point?

5)    Limit of predictability: Alkanes vs alcohols
Example: Alkanes only present London/dispersion forces, which are much weaker than hydrogen bonds. Does it mean that an alcohol because it presents hydrogen bond will always have a higher boiling point than an alkane? Investigate if a heavy alkane has a lower boiling point than a light alcohol. Explain and submit the URL with the selection of molecules.

 

Figure 7 Sankey diagram showing the flow of students’ answers to a series of questions related to the boiling point of molecular liquids. The last column labeled “College performance” shows the fraction of students belonging to High Performers (HP), Intermediate Performers (IP), and Low Performers (LP) that answered correctly or incorrectly the last question on “identifying the limit of predictability”.

 

Student performance in Controlled Variable Strategies (CVS)

 

Question number two in any of the sequences of questions above is the ideal opportunity when students can build knowledge by investigating trends and creating controlled experiments in truly open-ended data-driven exercises. This is particularly true when investigating the different factors affecting the boiling point of molecular liquids and it may justify a more detailed analysis.

ChemEd X Data offers a wide set of compounds and many ways to sort them. In this specific case, students were asked three different questions:
 

1.     Select a set of molecules on ChemEd X Data in order to run a controlled experiment to investigate the effect of shape (linear or branched) on boiling points. Explain what you found out.

2.     Select a set of molecules on ChemEd X Data in order to run a controlled experiment to investigate the effect of mass on boiling points. Explain what you found out.

3.     Select a set of molecules on ChemEd X Data in order to run a controlled experiment to investigate what intermolecular forces are stronger (H-bonds, dipole-dipole or London forces) and therefore have a higher boiling point. Remember that in a controlled experiment the different molecules must keep shape and mass as constant as possible so that you are only assessing the effect of the intermolecular force.

 

As the table below shows, students perform very differently in CVS questions depending on what molecular property they are trying to investigate. When assessing the branch effect (first row), the majority of students are able to select a correct set of molecules in which the mass and IMF are kept constant while the level of branching of the alkyl chain is different and therefore being tested. While twice as many high performing students can answer that question correctly than low performing ones (81.3% vs 46.2%) there is no significant difference in the final course grade between students who can answer this question successfully and those who cannot (p = 0.12). On the other hand, when trying to build a controlled experiment to test the effect of intermolecular forces (last row), a much lower fraction of students in all three categories can correctly answer this question. The fraction of high performers who can do it well more than triples the fraction of low performers who can (31.3% vs 7.7%). And the final grade between successful students and unsuccessful ones is statistically significant (p = 0.016).

The results in this table seem to indicate that students are not necessarily challenged by the open-endedness of the question, nor navigating the website, but, rather, by the fact that some molecular and atomic factors are more difficult to identify, select and keep constant in a controlled experiment than others. In this case, when testing IMF, it is difficult to keep the mass approximately constant while changing the functional group.

 

Table 3 Results showing how students perform in three different exercises requiring Controlled Variable Strategies to examine the effect of branch/linear chains, mass and intermolecular forces (IMF) on the boiling point of molecular liquids. Students are divided in three groups as defined above HP: high performing; IP: Intermediate performing, and LP: low performing.

	% of all students who answered correctly	% of HP students who answered correctly	% of IP students who answered correctly	% of LP students who answered correctly	Final % course grade average for students answering Correctly/Incorrectly; p-value
Select molecules to assess branch effect	72.4%	81.3%	74.2%	46.2%	83.1/79.2; p = 0.12
Select molecules to assess mass effect	46.6%	50.0%	48.4%	30.8%	83.1/81.1; p = 0.37
Select molecules to assess IMF effect	27.6%	31.3%	32.3%	7.7%	85.8/80.6; p = 0.016

 

 

Conclusions

The conclusions can be highlighted in five main points.

1)    A unified approach to structure-property relationships allows instructors to aim at specific skills of different cognitive levels throughout a General Chemistry semester.

2)    The ChemEd X Data website facilitates the practice of some of these skills using data-first activities.

3)    Sankey diagrams show us how students correct and incorrect answers flow through a sequence of questions.

4)    The type of questions that students have the most trouble with are: “Control of Variable Strategy” (step 2) and “Identifying the limit of predictability” (step 5)

5)    Students can better identify the limit of predictability when using ChemEd X Data.

6)    CVS exercises are only possible with data-first activities. Student performance will depend on the type of molecular factor being tested.

 

Bibliography

 

(1)       Executive Office of the President Council of Economic Advisers. Preparing the Workers of Today for the Jobs of Tomorrow http://www.whitehouse.gov/administration/eop/cea/Jobs-of-the-Future (accessed Jun 21, 2013).

(2)       A Framework for K-12 Science Education: Practices, Crosscutting Concepts, and Core Ideas; National Academies Press: Washington, D.C., 2012.

(3)       National Research Council. Next Generation Science Standards: For States, By States; 2013.

(4)       Laverty, J. T.; Underwood, S. M.; Matz, R. L.; Posey, L. A.; Carmel, J. H.; Caballero, M. D.; Fata-Hartley, C. L.; Ebert-May, D.; Jardeleza, S. E.; Cooper, M. M. Characterizing College Science Assessments: The Three-Dimensional Learning Assessment Protocol. PLOS ONE 2016, 11 (9), e0162333.

(5)       Eklund, B.; Prat-Resina, X. ChemEd X Data: Exposing Students to Open Scientific Data for Higher-Order Thinking and Self-Regulated Learning. J. Chem. Educ. 2014, 91 (9), 1501–1504.

(6)       Prat-Resina, X. Models360 and ChemEd X Data: web platforms to navigate, represent and interpret chemical information. Chem. Educ. EduQ 2016, No. 22, 22–30.

(7)       Grubbs, W. T. Data-Driven Exercises for Chemistry: A New Digital Collection. J. Chem. Educ. 2007, 84 (8), 1391.

(8)       Nichol, C. A.; Szymczyk, A. J.; Hutchinson, J. S. Data First: Building Scientific Reasoning in AP Chemistry via the Concept Development Study Approach. J. Chem. Educ. 2014, 91 (9), 1318–1325.

(9)       Walker, M. A.; Li, Y. Improving Information Literacy Skills through Learning To Use and Edit Wikipedia: A Chemistry Perspective. J. Chem. Educ. 2016, 93 (3), 509–515.

(10)     Pence, H. E.; Williams, A. J. Big Data and Chemical Education. J. Chem. Educ. 2016, 93 (3), 504–508.

(11)     Pienta, N. J. A “Flipped Classroom” Reality Check. J. Chem. Educ. 2016, 93 (1), 1–2.

(12)     Muthyala, R. S.; Wei, W. Does Space Matter? Impact of Classroom Space on Student Learning in an Organic-First Curriculum. J. Chem. Educ. 2013, 90 (1), 45–50.

(13)     Dame, L.; Aryal, B.; Huq, A.; Prat-Resina, X. Designing and Delivering an Interdisciplinary Quantitative Curriculum for a Health Sciences Program. Submitted 2016.

(14)     Schwichow, M.; Zimmerman, C.; Croker, S.; Härtig, H. What Students Learn from Hands-on Activities. J. Res. Sci. Teach. 2016, 53 (7), 980–1002.

(15)     Beauchamp, G. Further Analysis of Boiling Points of Small Molecules, CHwFxClyBrz. J. Chem. Educ. 2005, 82 (12), 1842.

(16)     Beauchamp, G. Boiling Points of Halogenated Ethanes: An Explanatory Model Implicating Weak Intermolecular Hydrogen−Halogen Bonding. J. Phys. Chem. A 2008, 112 (42), 10674–10680.

(17)     Glazier, S.; Marano, N.; Eisen, L. A Closer Look at Trends in Boiling Points of Hydrides: Using an Inquiry-Based Approach To Teach Intermolecular Forces of Attraction. J. Chem. Educ. 2010, 87 (12), 1336–1341.

(18)     Peckham, G. D.; McNaught, I. J. Teaching Intermolecular Forces to First-Year Undergraduate Students. J. Chem. Educ. 2012, 89 (7), 955–957.

(19)     Carter, C. S.; Brickhouse, N. W. What Makes Chemistry Difficult? Alternate Perceptions. J. Chem. Educ. 1989, 66 (3), 223.

(20)     Scerri, E. R. The Trouble with the Aufbau Principle. Educ. Chem. 2013, 50 (6), 24–26.

(21)     Noble, J.; Sawyer, R. Predicting Different Levels of Academic Success in College Using High School GPA and ACT Composite Score. ACT Research Report Series. ACT Res. Rep. Ser. 2002.