Scaffolding and Explicit Error Feedback

Scaffolding prevents learners from feeling overwhelmed when confronting complex systems and procedures [21, 25]. Given that chemical resonance is often when organic chemistry students start putting all of their foundational knowledge together to solve problems, using digital media to scaffold the experience seemed like it may benefit some learners with less firm foundational knowledge.

The scaffolds that can be toggled are:

1. Always show the implicit hydrogens in the bond-line drawing 2. Always show the lone pairs in the bond-line drawing

3. Show implicit hydrogens when an error occurs on an atom 4. Show lone pairs when an error occurs on an atom

5. Show error type (e.g. too many electrons) and specify location of the error (e.g. highlighting the atom)

All but the last of these scaffolds specifically help learners who struggle with reading bond-line diagrams fluently, but have learned the basics of how to read them. The Software has four tiers of decreasing scaffolding, each of which are announced by tutorial dialogs when the user moves to a more advanced tier (see Figures 3.13 and 3.14 for examples). The Software promotes users to a more challenging tier after the user solves five problems in a row from the current tier on the first try, not including

Figure 3.13: The tutorial dialog at the start of the first scaffolding tier.

Figure 3.15: Example of a problem at the first scaffolding tier. Implicit hydrogens are shown, and if there were any lone pairs, they would also be shown.

the initial beginner problems. The designation of five-in-a-row in one try as signifying mastery is a somewhat arbitrary example of how this promotion criteria might work; with further user tests it may be that a better criteria accounts for other statistics or varies the criteria for promotion depending on the scaffolding tier.

The four scaffolding tiers in the Software, from most scaffolding to least, are: 1. All scaffolds toggled on — user sees implicit hydrogens and lone pairs always,

and receives complete information about an error

2. Like previous tier except implicit hydrogens only appear for errors — user has to recognize where implicit hydrogens are hiding when drawing arrows

3. Like previous tier except lone pairs also only appear for errors — the non-error state of the drawing looks like a normal bond-line drawing

4. Like previous tier except implicit hydrogens and lone pairs are never shown — the user always works with normal bond-line drawings

5. Like previous tier except errors are generic — if user’s solution is not a valid resonance structure, user must find the error themselves

The last tier represents a more exam-like experience in that the learner must diagnose their own error. Though reduced feedback may impede initial understand- ing, learners eventually need the opportunity to build their own schemata [15]. In general, less feedback forces the learner to think more deliberately about their own performance and knowledge [15].

Figure 3.16: Example of a problem in a scaffolding tier where the implicit hydrogens are only shown when the Software identifies an error; other carbon atoms in this molecule do not have their implicit hydrogens visible.

However, the Software still provides feedback about whether or not a submission is significant even at the most difficult tier, to prevent learners from confusing error- containing invalid solutions with insignificant but valid solutions. Eventually, another practice tool for ranking the relative significance of resonance structures may be beneficial for more expert learners.

When an error message appears in the lower left corner, it gives the user a cue of brief text as to what the error is and, unless the user is in the most difficult scaffolding tier, the atom in question turns red (see Figure 3.7). For a user who recognizes what the error is, the brief text should be enough for them to figure out how to correct their solution; for a user who does not see how the error applies, tapping the message brings up an error dialog with more specific information about what the error means (see Figure 3.17 for an example of the dialog).

A more detailed error message requires a tap, rather than appearing right away, for two primary reasons.

First, longer textual explanations of the error require more space; some users will not read longer explanations carefully, unless they deliberately seek them out [55]. More text taking up more space may require putting the error message in a more prominent place, which could distract the user from looking at the molecule as the primary way to investigate an error in the solution. If the error text demands user attention, users who might recognize the error once they see it in the resulting structure might be denied the opportunity to see the error for themselves before reading the text.

Figure 3.17: The detailed error dialog for an atom with too many electrons.

Second, the tiny amount of extra friction nudges users to process the brief cue about the error and look at the molecule first before deciding that they need more information. Access to a more detailed error message is not restricted, but it is not the default, which encourages the user to more actively investigate the error [58].