Examining undergraduate and graduate student reasoning when interpreting infrared spectra

Abstract

As students progress through the chemistry curriculum, their interaction with and understanding of instrumentation increases. Integral to this educational journey is the acquisition of skills in interpreting data generated by a wide variety of instruments. Recent efforts have aimed at delineating student assumptions and cognitive constraints in the interpretation of spectral data across various educational levels, notably focusing within organic chemistry settings. However, there is currently limited work focusing on how upper-level undergraduate chemistry students engage with spectral data, particularly pertaining to infrared (IR) spectra. In this qualitative study, we investigate the strategies employed as upper-level undergraduate analytical chemistry students and graduate chemistry students interpret IR spectroscopic data, directly engaging in the scientific practice of analyzing and interpreting data. Sixteen semi-structured interviews were conducted using one task from a larger mixed-methods eye tracking study. Guided by data-frame theory, the findings of this research underscore the critical role of instructor modeling in facilitating the integration of data and frame to derive meaningful conclusions when interpreting IR spectra. This study contributes to a deeper understanding of the interpretation of spectral data, thereby informing pedagogical practices in chemistry education.

---

Introduction

With the publication of the National Research Council's Framework for K-12 Science Education and its practices, crosscutting concepts, and core ideas, a call was issued for the alignment of postsecondary curricula with three-dimensional learning (National Academies Press, 2012; Cooper et al., 2015). This has prompted the chemistry education community to conduct extensive research, concentrating on students' involvement in these scientific practices, crosscutting concepts, and core ideas (Stowe and Cooper, 2019; Xue and Stains, 2020; Erduran, 2022; Lieber and Graulich, 2022; Berg and Moon, 2023; Urbanek et al., 2023; Zhou and Moon, 2023). This previous work has significantly advanced our understanding of how students construct arguments from evidence and engage in data analysis and interpretation, as well as underscores the imperative for instructors to explicitly model and support argumentation practices. While the majority of these investigations have occurred within the context of undergraduate general and organic chemistry classrooms, limited attention has been directed towards upper-level undergraduate chemistry courses (Crandell et al., 2020; Dood et al., 2020; Deng and Flynn, 2021; Watts et al., 2021; Erduran, 2022). Notably, many of the studies within the context of organic chemistry employ spectral data as the primary data source, with a prevalent focus on nuclear magnetic resonance (NMR) spectra with occasional supplementation with infrared (IR) spectra and other data (Cartrette and Bodner, 2010; Topczewski et al., 2017; Connor et al., 2019; Stowe and Cooper, 2019; Connor et al., 2021). IR spectroscopy is an important technique employed widely across various areas of chemical industry (Hamilton et al., 2024). Recognizing its significance, the American Chemical Society Guidelines explicitly highlight its importance, emphasizing its need for use in the United States undergraduate chemistry curriculum (Committee on Professional Training, 2023). A recent survey of undergraduate chemistry laboratories highlighted the prevalence of IR spectrometers as one of the more frequently used instruments across different disciplinary laboratories (Connor and Raker, 2022). Given the ubiquitous nature of spectral data across a wide variety of chemical domains, it becomes imperative to expand research in this area, contributing to a more comprehensive understanding of how students engage in data analysis and interpretation with IR spectra beyond introductory coursework.

Previous research has shown that the decision and/or ability to access or not access particular prior knowledge and experience can impact how one analyzes data (Heisterkamp and Talanquer, 2015; Connor et al., 2019). A previous study on graduate student and faculty interpretation of IR and NMR spectra found that experience in solving problems containing spectra was an important factor for success (Cartrette and Bodner, 2010). Additionally, an eye tracking study on student reasoning when relating molecular structures to IR spectra found that students at different educational levels (undergraduate and graduate) rely on different assumptions about structure–property relationships and exhibit different gaze patterns when viewing spectral data (Cullipher and Sevian, 2015). Another study furthered this area of investigation by using student assumptions to identify types of mental models held, with implications to incorporate multiple variables when teaching spectral analysis, including the comparison of spectra and structure (Wright and Oliver-Hoyo, 2020). Finally, one study on organic chemistry students’ interpretation of IR and NMR spectra found that invalid chemical assumptions and heuristic reasoning strategies constrained student reasoning (Connor et al., 2019). In connection with this work, a subsequent report from Connor and colleagues focused on information processing during the interpretation of NMR spectra and complementary IR spectra and how students develop understanding of relevant concepts and principles when interpreting spectra (Connor et al., 2021). This work additionally revealed that undergraduate students often focused their attention on the fingerprint region within IR spectra, and that this was significantly different in comparison to the visual behavior of graduate student participants (Connor et al., 2021).

Much of the previous research focusing on student interpretation of IR spectra has predominantly centered on categorizing student assumptions and identifying constraints in student reasoning (Cullipher and Sevian, 2015; Connor et al., 2019; Wright and Oliver-Hoyo, 2020). While these previous studies have furthered our understanding of the assumptions and conceptual ideas held by students, they do not directly investigate the role of the data itself in the development of these assumptions and ideas. For example, student assumptions related to specific bonds present within a molecular structure may impact how they interpret a corresponding IR spectrum. Likewise, students’ interpretation of IR spectra may influence their assumptions about the structure of a given molecule. As a novel contribution, this study uses data-frame theory to explore how student knowledge is operationalized via frames as they interpret IR spectra.

Theoretical framing

Klein and colleagues posited data-frame theory as the act of sensemaking that involves a symbiosis of data and frame (Klein et al., 2006), with data being any observation or information available to a learner and a frame being an explanatory structure that accounts for data and guides the search for any additional data (Klein et al., 2007). A frame can be a plan for how to approach a task, a story that explains events, or the reasoning one uses to make decisions (Klein et al., 2006, 2007). The data elicits and helps to construct the frame, while the frame connects and filters the data (Klein et al., 2007). This cyclic process of sensemaking ultimately led Klein and colleagues to define data-frame theory as a symbiotic process – neither the data nor the frame occur first, and both the data and the frame inform each other (Klein et al., 2007).

When interacting with new data, a learner will use specific pieces of data as anchor points, which leads to the establishment of a frame; anchor points are specific observations or portions of the data that are used to establish the initial frame (Klein et al., 2006). For sensemaking to occur, data and frame must align – when data does not fit the frame, the sense maker may begin to question their frame and attempt to determine how the data does not fit the frame (Klein et al., 2006). Once the sense maker questions the frame, Klein and colleagues posit that they may engage in one of two pathways, shown in Fig. 1: either maintaining the frame or reframing. Maintaining the frame involves keeping the same anchor points by either elaborating on the frame or preserving the frame (Klein et al., 2006). To elaborate on the frame, the sense maker will seek to incorporate new data into their existing frame through linkage to existing prior knowledge and/or by expanding their frame to include new data; in contrast, to preserve the frame, the sense maker will explain away or disregard data (Klein et al., 2006). The reframing pathway involves the selection of new anchor points which leads to the establishment of a new frame. This allows the sense maker to gain a new lens through which to view the data (Klein et al., 2006). Both pathways are cyclic, and the process will repeat until the sense maker feels that the data and frame align (Klein et al., 2007).

Fig. 1 Visual representation of data-frame theory, modified from Klein and colleagues (Klein et al., 2006) and Berg and Moon (Berg and Moon, 2023).

Klein and colleagues also estimate that each sense maker can use up to three frames simultaneously, and that a person may compare more than one frame to help make sense of the data (Klein et al., 2006). Berg and colleagues posited that to do this, the sense maker must engage in decentering from their central perspective – this process involves recognizing alternative perceptions of the same problem, which has been shown to support more productive argumentation and support student–teacher interaction (Berg and Moon, 2023). Previous research has also shown that experts and novices engage in data analysis and interpretation in similar ways (Berg and Moon, 2023; Zhou and Moon, 2023). What differentiates experts and novices is the prior knowledge and experience to which they have access – an expert will possess a deeper or more connected collection of knowledge and experiences which can inform the frame they establish during the sensemaking process (Berg and Moon, 2023; Zhou and Moon, 2023). The sensemaking process contains many skills that students must develop and is an important component in learning how to interpret data. Previous studies have shown that data analysis and interpretation is a dynamic process, where data-frame theory can be leveraged as a lens to learn how an individual's frame impacts how they view and interpret data (Klein et al., 2006; Slominski et al., 2020; Berg and Moon, 2023; Zhou and Moon, 2023).

Research question

This research study uses data-frame theory as theoretical framing to characterize students’ sensemaking processes and examine interactions between frames and data in the context of problems containing IR spectra, guided by the following research question: How do the frames established by upper-level chemistry students in instrumentation-focused laboratories and chemistry graduate students impact how they interpret IR spectral data?

Methods

A qualitative approach was employed to analyze retrospective think-aloud responses from students during a stimulated recall interview conducted after the completion of an eye tracking task involving a problem containing an IR spectrum (Lyle, 2003; Hyrskykari et al., 2008; Holmqvist et al., 2011). These methodological decisions were reached based on the positionality of the members of the research team, as well as the context in which this study was conducted (Secules et al., 2021). Specifically, the decision to use data-frame theory was informed by the research expertise of authors MBA and DTH related to the method of eye tracking – data-frame theory was selected due to its potential to be a useful framework for analyzing both the visual quantitative gaze data and corresponding qualitative interview data for tasks containing spectral information. While this was part of a much larger mixed-methods eye tracking study containing five tasks, this manuscript will focus solely on the qualitative data from one task. Eye tracking data will not be included in the present manuscript.

Ethical considerations

This study is one component of a larger investigation centered on the interpretation of IR spectra, with data collection occurring during Fall 2022, Spring 2023, and Summer 2023 at a large research institution in the southern region of the United States. Institutional Review Board (IRB) approval was obtained prior to participant recruitment and data collection.

Participants

A total of 16 participants were recruited to the study, including four undergraduate students enrolled in a Quantitative Analysis laboratory course, six undergraduate students enrolled in an Instrumental Analysis laboratory course, and six graduate students of various levels and divisions from the Department of Chemistry at the institution of study. All participants have been assigned pseudonyms to protect their identity, chosen by the research team to not specifically reflect their demographic information. One undergraduate participant enrolled in the Instrumental Analysis course indicated that they guessed on their answer to the task; thus, their response and subsequent data analysis was not featured within the current manuscript, which includes a total of 15 participants. The recruitment and purposeful sampling of participants spanned a diverse range to increase the variation in student responses and facilitate the comparison of eye tracking data across distinct levels of experience within the larger investigation (Palinkas et al., 2015). Demographic data from all 15 participants in this study has been included in Appendix 1.

Task design and development

All tasks were designed to mirror problems or challenges that students may encounter in the Analytical and Instrumental Analysis classroom and/or laboratory. Expert validation through review of the tasks occurred via consultation with instructors of the respective classroom and laboratory courses at the institution of study (Cook and Hatala, 2016). While the molecules included within the tasks for this study were all organic molecules, expert reviewers confirmed that the tasks were all appropriate within the context of the study, as they were similar to problems used in these courses for instruction on spectral interpretation. All tasks were designed via the conceptual framework of representational competence, where each task featured selected skills, characterized by variations in the quantity and types of skills per task; these skills included ability to interpret representations, translate between different representations, use representations via application to solve problems, and generate representations (Kozma and Russell, 2005).

This manuscript will focus on qualitative findings for student responses to the fourth task (shown in Fig. 2) of the five total tasks. After an initial pilot study, this task underwent modification based on findings that indicated that students often formulated assumptions about the provided molecular formula and subsequently overlooked spectral peaks that did not align with these assumptions. The initial pilot task has been shown in Appendix 2. In response to this insight, the task (shown in Fig. 2) was restructured to incorporate a molecular formula that could be two distinct molecules, and data-frame theory was adopted as an analytic framework for the qualitative interview data. This task asks participants to choose the molecular formula that best corresponds to the provided IR spectrum. There are two options given: (A), propane C₃H₈ and (B) propyne C₃H₄. Based solely on molecular formula, C₃H₄ could also potentially contain two double bonds and alternatively be considered, with the molecular formula also corresponding to propadiene. In this task, the participants must not only choose between option A and option B, but also identify the structure of option B using the spectrum provided. The spectrum provided has a weak peak at 2250 cm⁻¹ which indicates that there is a carbon–carbon triple bond. This peak, along with the lack of a strong carbon–carbon double bond peak at 1650 cm⁻¹, indicates that the appropriate answer is option B, and that the molecule is propyne.

Fig. 2 Task provided to students on a computer screen, asking them to circle the molecule (A) or (B) corresponding to the provided IR spectrum. Spectrum sourced from NIST Chemistry (Wallace, 2022).

Data collection

The task involved in this investigation constituted the fourth among five tasks assigned to participants, completed in sequence on a computer screen equipped with an attached eye tracker. Following each task, participants participated in a retrospective think-aloud, stimulated recall interview, during which they were presented with a recorded video of their eye movements (stimulated recall) and asked to describe their cognitive processes while solving the task (retrospective think-aloud) (Lyle, 2003; Hyrskykari et al., 2008; Holmqvist et al., 2011). The interview sessions varied in duration, ranging from 20 to 60 minutes. Both audio and video data were recorded during the interviews, with the position of the video camera allowing for the identification of student gestures and pointing actions toward the screen. Audio recordings were transcribed using Otter.ai (Otter.ai, 2024), and all transcripts were subsequently deidentified.

Data analysis

The first author (DTH) conducted and transcribed all interviews via Otter.ai and manually cleaned the resulting transcriptions (Otter.ai, 2024). Regarding researcher positionality (Secules et al., 2021), author KH was likely considered to have insider perspectives as a part of the community within which the research was conducted. However, she did not conduct interviews or transcribe those interviews in this study, as these steps occurred prior to her joining the research team. Data familiarization (Braun and Clarke, 2021, 2023) was carried out by both the first and second authors (DTH, KH) by reading through each transcription to gain a better understanding of the content of the interviews.

Initially, a pilot study was conducted by the authors using tasks that contained IR spectra, similar to tasks in the present study. While not presented in the manuscript, findings from this initial pilot study indicated that participants analyzed data through a specific lens or perspective, and this led some participants to disregard data that did not fit their perspective. These initial findings led the authors to use data-frame theory to develop five tasks, including the task in the present manuscript (Fig. 2). Data-frame theory was selected for task development and analytic framing due to its distinction between frame and data, as well as the ability to observe how different frames impact data analysis. For data collected in the present study, open inductive coding was first used to initially characterize knowledge and/or strategies that the participants used to help them solve the task (Braun and Clarke, 2021, 2023). Next, deductive coding was carried out using an a priori codebook informed by data-frame theory to sort the initial inductive codes into three categories: data, frame, and conclusion (Klein et al., 2006; Braun and Clarke, 2023; Zhou and Moon, 2023). This codebook is shown in Appendix 3, with selected quotes to represent each category. Codes signifying direct observations regarding the spectrum and molecular formulae were grouped within the data category. Codes signifying students’ utilization of prior knowledge or experiences to interpret observations were grouped within the frame category. Finally, codes indicative of students formulating conclusions based on the data and their frame were grouped within the conclusion category.

This research was intentionally designed to follow an asset-based approach, meaning that the researchers avoided judgements of participant competence and instead focused on the frames participants established and how this impacted their data analysis (Bain et al., 2019; Tashiro and Talanquer, 2021; Crandell and Pazicni, 2023). To categorize participant responses following this approach, a model from a previous study was leveraged (Crandell and Pazicni, 2023) and adapted to include the productivity of the frame and the accuracy of the data interpretation, which occur on separate axes (Fig. 3). This adapted model was discussed among all members of the research team until consensus was reached. To use this model, all responses categorized as data or frame (Appendix 3) were disaggregated. The frame category was disaggregated into productive and less productive, while the data category was disaggregated into high accuracy and developing accuracy regarding data interpretation. Regarding productivity of the frame, frames that focused on the presence of a triple bond were regarded as productive by the research team. Frames that focused on the presence of double or single bonds were regarded as less productive, with the latter being regarded as more productive within the “less productive” categorization. The first and second authors (DTH, KH) independently sorted each student response and placed it into the model shown in Fig. 3 before coming to complete consensus. Throughout analysis, the research team did not assume that the level of accuracy of interpretation was indicative of the level of productivity of the established frame, or vice versa. If more than one frame was used during analysis, the frame that the participant used to develop their conclusion was the frame used to place the participants’ response on the model. If more than one point of data was used, which was the case for most participants, the level of accuracy of interpretation of each portion of data was aggregated with the others before being placed on the model.

Fig. 3 Operationalized model of the interaction between data-frame theory to include the productivity of the frame and the accuracy of the data interpretation.

In following an asset-based approach, it is important to acknowledge that the quadrants represented on the model in Fig. 3 are not viewed as better or worse, but only as a method to differentiate ways in which participants solved the task provided. The categories themselves are not inherent to the participants, but rather constructed by the researchers through our perceived understanding of the student responses using our own knowledge and experiences (Mauthner and Doucet, 2003; Bott, 2010; Jamieson et al., 2023). These categories do not define the participants’ knowledge nor the participants themselves. Finally, we are only able to code knowledge that was explicitly stated during the interview. Thus, data analysis and conclusions drawn reflect the spoken knowledge of the participants.

Regarding trustworthiness and credibility (Lincoln and Guba, 1986), initial individual open coding was conducted by the first and second authors (DTH, KH) before then being compared and discussed between the two coders until complete consensus was reached (Reynolds et al., 2011). The first and second authors (DTH, KH) then individually coded the responses using the codebook developed via data-frame theory and sorted each participant into a respective quadrant in the model shown in Fig. 3. Complete agreement between all members of the research team was required for each quadrant placement before establishing final categories for task responses.

Positionality and reflexivity

Before the discussion of results, the authors of this manuscript call attention to the fact that all research holds embedded measures of subjectivity based on individual biases and reflects the context in which it was designed and conducted (Foote and Bartell, 2011; Gillborn et al., 2018; Bayeck, 2022; Rodriguez and Navarro-Camacho, 2023). The perspectives of the participant-facing researcher (DTH) and data-facing researchers (KH, MBA) shape all components of this work. The perspective of author DTH is informed from his experience as a current graduate student, who previously worked as an analytical chemist in industry, while KH is informed by her experiences as a senior undergraduate student at the time of data analysis and interpretation of results. In consideration of the researcher-as-instrument dimension of positionality in this qualitative work (Secules et al., 2021), DTH's interests are shaped by and come from experience in industry, with an interest in investigating data interpretation stemming directly from feeling underprepared following his undergraduate degree in chemistry; KH's interests and values are shaped by a commitment to enhancing the college experience for herself and fellow students.

Undergraduate participant recruitment was conducted by DTH and MBA, which impacted decisions related to participation in the study. Graduate student recruitment was conducted by DTH, and as he is a current graduate student, participation in the study was influenced by his relationship with others within the department. Additionally, participation in the study was impacted by the identity of MBA as an assistant professor within the department and a chemistry education researcher mentoring both undergraduate and graduate students at the institution of study.

We have intentionally included this positionality prior to the discussion of results to provide readers with the opportunity to critically evaluate who the research is done by and for, acknowledging that our identities have impacted all aspects of this research including but not limited to, reviewed literature, study design, data collection, data analysis, and conclusions drawn (Secules et al., 2021).

Results

Participant responses were grouped into five unique themes based on (1) the selected answer to the task, (2) the resulting quadrant placement shown in Fig. 3, and (3) if they engaged in a reframing process. These categories are shown in Table 1, with their corresponding descriptions. It is worthwhile to note that the quadrant number should not be interpreted as a reflection of hierarchy – it is not the authors’ intent to use the numbering system as a ranking of a student's overall knowledge and/or problem-solving ability. This quadrant model was used to operationalize the theoretical framing to investigate patterns within the interview data on how students solved a specific task involving an IR spectrum. To best tell the story of this data, details for each theme will be presented in the order in which they are listed in Table 1, rather than by numerical ordering.

Table 1 Themes of task responses

Theme	Description
Quadrant 2	Less productive frame, highly accurate data analysis, selected option A.
Quadrant 1	Productive frame, highly accurate data analysis, selected option B.
Reframing	Reframed from less productive frame to more productive frame, highly accurate data analysis, selected option B.
Quadrant 3-A	Less productive frame, developing accuracy for data analysis, selected option A.
Quadrant 3-B	Less productive frame, developing accuracy for data analysis, selected option B.

---

Quadrant 2

All participants within the quadrant 2 theme selected option A as their answer to the task. These participants established a frame where they assumed option B contained double bonds based on the molecular formula. These participants then interpreted the spectrum to not contain a carbon–carbon double bond peak, leading them to select option A. When asked to describe any prior knowledge and why they selected option A, Skye stated,

“So prior knowledge like the [previous task], knowing how to draw the structures from [options] A and B. And so being able to draw those out, and then I identified the functional groups, and then after identifying the functional groups, I could look at the table and use the table to say, okay, I should see a peak here. Do I see it? … I didn't see any big stretch around 1600, which was [to] indicate double bonds. And so since…I didn't see a really big peak. So I thought, oh, there's probably no double bonds.”

Skye used their knowledge of functional groups to establish a frame that option B had double bonds and option A did not. They identified that there were no double bond peaks present in the spectrum, and they did not expand upon this explanation or mention peaks in relation to the molecular formula for option A. Rather, Skye chose their answer based on the lack of evidence for support of double bonds in option B. In comparison, Danni described their selection of option A as,

“I was looking at the formula and then I was like drawing it in my head, and then also using the spectrum given and comparing it to where I think that those peaks should have been, if it was one compared to the other… So I started with B since it had the least amount of hydrogen. So I knew…the four bonds that carbon has to have, I knew that just based off of the way that the structure was, in the way I had to do it in my head that there [weren’t] enough hydrogens for…each C to have…to make like the total four bonds. So I put two and two together to know that there had to be at least one C that was double bonded. And then they each have…their respective H's. So I knew that, I don't know if this is making sense, though, I knew that there had to be a peak around 1650 based on what the graph told me, and there wasn't one. So then I went and looked at what…I drew out formula A in my head, and then was looking for where a C–H peak should be. And I found that…on the graphs, I was like, it has to be [option A]…because the other one is lacking the peaks that should be corresponded with that formula.”

Danni used their prior knowledge of carbon requiring a full octet to establish their frame that option B has double bonds, since there were not enough hydrogens to complete the octet. When interpreting the spectrum to support their frame, however, Danni determined a lack of a peak at 1650 cm⁻¹ to support option B as the answer, ultimately deciding that there was evidence for C–H single bonds and indicated option A as the answer. When the data prompted Danni to question their frame, they chose to preserve the frame that C₃H₄ has double bonds and elaborate their frame to include C₃H₈ having C–H bonds. Like Skye, Danni relies on frames that are established based on anchor points that come from the molecular formula, using the spectrum to verify their frame.

Participants in quadrant 2 use productive resources (e.g., the octet rule and hydrogen deficiency) and interpret that the spectrum does not contain a peak at 1650 cm⁻¹, which indicates carbon–carbon double bonds. Coding within this quadrant should not be considered binary, as being wholly productive or unproductive, and establishing that C₃H₄ has double bonds was not coded as unproductive. In the case of quadrant 2 responses, this was coded as less productive for this specific task – while it led students to accurately interpret the lack of the corresponding peak indicating double bonds on the spectrum, it also led them to not consider looking for a triple bond peak at 2250 cm⁻¹. These participants instead chose to preserve or elaborate their frame, which led them to select option A without reframing.

Quadrant 1

All participants within the quadrant 1 theme selected option B as their answer to the task. Unlike the other quadrants, these participants established a triple bond frame from the beginning and continued with this frame throughout the entire task. When asked to describe why option B was selected, Jay said,

“There is [an] alkyne peak around 3300 and then in 2200, carbon-triple bond-carbon. So, I just went with that. And then I saw…the first one is not unsaturated. And the second one is unsaturated…I saw [the] first peak and then second saw [the] alkyne peak, so the first molecule was not alkyne, so the only option was the second molecule.”

Jay first mentions the specific peaks in the data with which they established their frame. They note the carbon–carbon triple bond peak, as well as the C–H alkyne peak. After analyzing the spectrum, they state that they looked at the molecular formulae and determined that A was saturated and B was unsaturated, indicating a selection of the final answer as option B because option A was not an alkyne. In comparison, George described their selection of option B:

“So, I looked at each of the compounds, and the first compound I'm expecting is with three carbon C–H stretches, which I did not see or seem to see. It wasn't as prominent as it would have been if it was a saturated compound. Also, from prior knowledge, I know that…alkene or alkyne will have a band above 3000, that's going to be obvious. Also with this table, I was able to see that it's possible we have a C triple bound C stretch at around 2250. Both prior knowledge and this [table].”

George used different prior knowledge than Jay to establish their frame and determined that the size of the C–H peak was not as strong as it should have been for a saturated compound. They then used the spectrum to determine that the peak above 3000 cm⁻¹ indicated that the molecule would be an alkene or alkyne. Finally, they reference that they used the chart to determine that it was a triple bond due to the peak at 2250 cm⁻¹.

All participants in quadrant 1 explicitly referenced the carbon–carbon triple bond peak at 2250 cm⁻¹ when describing the establishment of their frame. This peak was not mentioned in responses within quadrants 2, 3-A, or 3-B. Most students within quadrant 1 stated that they recognized this critical peak through first looking at the IR spectrum. Additionally, one student looked for evidence of this peak due to first establishing that C₃H₄ should indicate a triple bond peak based on the molecular formula. Prior knowledge of peak location in relation to functional group was a resource used often for those in this quadrant.

Reframing

All participants in the reframing theme selected option B as their final answer to the task. These participants were initially sorted into quadrant 1, as their final established frame was that C₃H₄ had a triple bond, referencing the peak at 2250 cm⁻¹ when interpreting the spectrum. However, they were separated into a unique theme because, unlike the previously mentioned participant responses within quadrant 1, all participants in this reframing theme initially assumed that C₃H₄ had double bonds based on the molecular formula (like those in quadrant 2). They then subsequently transitioned from quadrant 2 to quadrant 1 by establishing a new frame through the activation of different prior knowledge and interpretation of the spectrum. Vance described this process of reframing:

“I knew [option B was] similar to A… So I was like, oh, maybe it's a double bond instead of just all single bonded. Then when I was counting my hydrogens, I was like, Wait… This isn't adding up, what am I missing? And I was like, oh, it's a triple bond. And so that's why I needed to erase it and write it that way. Because triple bonds are straight generally, when you draw them… So once I knew…what bands I was looking for, for each molecule, I immediately went to look for what ones [were] given to me in the…table… I knew I had a C–H given to me and the C triple bond. So I knew that A would only have a peak for C–H groups. And then B will have two peaks that should have the triple bond peak and C–H peak… So I immediately saw that there was a C–H peak, and then once I found that…I needed to find the triple bond peak, which, as it states in the table is very weak. And it's around the 20 to 50 range. So I was like, okay, that this tiny little peak right in the middle is, is probably it, and I just was checking to make sure I wasn't missing anything.”

Vance used hydrogen deficiency as their reasoning to reframe from a double bond to a triple bond in the molecular formula for C₃H₄. Like previous participants, they did use the molecular formula to guide how they viewed the data; however, because their frame assumed a triple bond, they looked for the peaks which led them to find evidence as verification for the triple bond. This participant reframed from a less productive to more productive frame, and this reframing was due to prior knowledge activated (hydrogen deficiency) based on the molecular formulas provided. Caden provided another example of reframing for this task:

“The C double bond C…wasn't where I thought it should be. So then I was like, well, maybe it can be a triple bond. And I was like, oh, yeah, that's a thing [points to peak on the spectrum at 2250 cm ⁻¹ ].”

Caden initially established a frame that C₃H₄ contained a double bond, drawing out the structure to help them with the task. Upon analyzing the spectrum, they noticed that there was not a carbon–carbon double bond peak indicated. This discrepancy prompted them to question their initial frame and led to them activating prior knowledge of triple bonds, which prompted them to reframe to C₃H₄ containing a triple bond and subsequently noticing the triple bond peak at 2250 cm⁻¹.

Participants in the reframing theme initially established a less productive frame, that there were double bonds in the structural formula of C₃H₄. However, when the data did not agree with their initially established frame, they reframed. This led them to establishing the frame that there was a triple bond in the structural formula, leading them to select option B.

Quadrant 3-A

All participants in the quadrant 3-A theme selected option A, C₃H₈, as their answer to the task, establishing their frame based on the molecular formula first before interpreting the spectrum for verification. All within this quadrant recognized that option A contained more hydrogens than option B, which was used as an anchor point for the frame they established. When asked to discuss this selection, Jake described,

“That's the thing like I…forgot. How many double ones will be there? How is it like…triple bond or double bond? That is the first thing I thought about… Which one [option] is the double bonded? … And then I thought, it's C ₃ H ₈ , should have some double bonds. And C ₃ H ₄ will have triple bonds. So then I will look for the double bonds in the spectra and the wave numbers for it. I didn't see any triple bond wave number in the spectra. And then I saw the double bonded wave number, as well as the C–H in the 2850 and 3000.”

Jake established a frame based on differences in the number of hydrogens between the two options and concluded that this difference was due to the presence of a double bond in option A and a triple bond in option B. They selected option A based on their interpretation that the spectrum did not indicate a triple bond peak but did indicate a double bond peak. While Jake does include in the frame that C₃H₄ has a triple bond, the overall frame was still categorized as less productive due to the frame including that there were multiple triple bonds, instead of one, and that C₃H₈ had multiple double bonds. Birch also established a frame based on differences in the number of hydrogens; however, they concluded that this was due to a difference in the number of double bonds present in the molecules:

“Yeah, so A has more hydrogens, which it has way less double bonds from what I can tell. So it's not going to have nearly as strong of a response when it comes to the double bond peaks considering it like, pretty confidently there was definitely like a pretty strong C–H, [option A] just seemed like the right one given the fact that there was so little response for the double bond.”

Birch's frame that there were fewer double bonds in the structural formula of option A was elaborated upon by the assumption that this would cause option A to have a stronger C–H peak than a double bond peak, when compared to option B. They selected option A based on their interpretation that the double bond peak within the spectrum was weak, which supported their frame.

All participants in the quadrant 3-A theme selected option A because they established less productive initial frames based on the molecular formulas as anchor points. They did not mention any consideration of alternative frames because they interpreted the spectrum to support their established frame.

Quadrant 3-B

All participants in the quadrant 3-B theme selected option B as their answer to the task, by establishing a frame that option B contained double bonds. They were separated from quadrant 3-A, in part, due to the anchor points from which they established the frame as well as how the frame guided their data analysis. The participants in this quadrant established that option B, C₃H₄, contained double bonds based on spectral interpretation of the presence of a carbon–carbon double bond peak. Angel reported choosing option B because,

“Initially, I didn't know because I was like, well how am I supposed to know how many hydrogens are, then I was like…double bonds and triple bonds of course. And I looked for that. And I think it had a double bond, so I picked option B. I didn't think it was A… Because the only thing that's changing between A and B is number of hydrogens. So I mean, if there is a double bond, there'd be a peak there. So I looked for that. And if there was one, then it has to B… It's at 1650.”

Angel selected option B because they established a frame that there was a double bond, based on the difference in the number of hydrogens in the two molecular formulae. When asked why they established this frame, they referenced that there was a double bond peak on the IR spectrum at 1650 cm⁻¹. While they mention the potential for triple bonds, they established the frame of double bonds using the carbon–carbon double bond peak in the IR spectra as their reasoning. Similarly, Blair describes their choice of option B:

“Because I thought the double bond C–C is like, there's like a little spike. It's not strong, but I don't know what strong or medium is. This one [points at the area of the spectrum at 1500 cm ⁻¹ ].”

Blair also determined that there was a double bond between the carbons based on peaks in the IR spectrum, pointing to the peak at 1500 cm⁻¹ as being evidence for the double bond. They also mention that they were unsure of what strong or medium meant from the chart, which was noted by several participants throughout the interviews.

All participants in the quadrant 3-B theme established a frame which led them to interpret the spectrum as indicating a carbon–carbon double bond. Angel used the difference in the number of hydrogens within the molecular formulae to assume double or triple bonds as their anchor point, while Blair used the peak at 1500 cm⁻¹ as their anchor point. Since their frame was supported by their interpretation of the spectrum, they did not consider reframing or discarding the data. They also chose option B, which is the correct answer; however, their reasoning was that this was because the molecule contained double bonds.

Discussion and conclusions

The findings from this study contribute in several ways to our understanding of how students interpret spectroscopic data, particularly IR spectra. Specifically, this work provides evidence that there are many factors that impact how students analyze and interpret IR spectra, and that assessing student understanding via only binary accuracy (i.e., correct/incorrect) may not adequately measure student understanding. Rather, characterizing student understanding based on productivity of the frame has been proposed to provide more insight into conceptual understanding and reasoning via scientific practices (Heisterkamp and Talanquer, 2015; Eckhard et al., 2022; Noyes et al., 2022). Several participants in this study selected the “correct” answer, while still establishing that this molecule contained double bonds. Additionally, previous studies have found that students are adept at making accurate observations of data (Berg and Moon, 2023; Zhou and Moon, 2023), and that students are able to observe and interpret information specifically contained within IR spectra (Stowe and Cooper, 2019). In contrast, and in alignment with work by Connor and colleagues (Connor et al., 2019; 2021), our work found that students may have difficulty in differentiating between peaks on IR spectra that are close in wavenumber despite providing tables of information in connection with the spectra. This is also connected to work by Wright and Oliver-Hoyo, who propose the benefit of having students reason about IR spectra conceptually alongside wavenumber data (Wright and Oliver-Hoyo, 2020). In our work, only those who had established a frame that C₃H₄ had a triple bond referenced the weak peak at 2250 cm⁻¹, and the participants who reframed to assume a triple bond peak referenced hydrogen deficiency and the lack of a double bond peak as the data that caused them to question their frame. These students did not explicitly mention the 2250 cm⁻¹ peak until after reframing. This has a potential implication that the weak triple bond peak was ignored to preserve the frame, or, given its size, simply ignored as inconsequential, which would align with previous work (Meister and Upmeier zu Belzen, 2021). Future research should consider further investigating this phenomenon.

The results also align with literature to support the evidence that experts and novices perform similar sensemaking processes (Klein et al., 2007; Berg and Moon, 2023; Zhou and Moon, 2023). In our work, both undergraduate and graduate students used data to establish a frame which they used to guide further analysis. There was also not a clear delineation between the types of prior knowledge used and accuracy of data interpretation between the groups of students. There were undergraduate and graduate participants in quadrants 1 and 3, and participants from both undergraduate laboratory courses in quadrant 2 (Fig. 3). Previous studies on IR spectral interpretation have primarily focused on organic chemistry students as novices and graduate students as experts (Cullipher and Sevian, 2015; Connor et al., 2019; 2021). It is possible that, as the upper-level undergraduate students in our study had already taken organic chemistry, they had enough experience to no longer be considered novice. These findings are not meant to question other studies and prior work that has previously provided evidence of differences between course levels, but rather to highlight the level of knowledge that upper-level undergraduate students hold. Additionally, our work provides evidence that upper-level undergraduate students may need to be considered as a different population when analyzing data related to interpreting spectral information, in comparison to students from an introductory-level undergraduate course.

Implications for practice

For instructors, it is important to model data analysis and interpretation, emphasizing that all data within an IR spectrum must be examined, particularly above the fingerprint region. Modeling how to move between data and frame, how to establish anchor points, and how to reframe is encouraged. Berg and Moon showed how simulated peer review could be used to allow students to see differences in their frames and discuss these differences (Berg and Moon, 2022, 2023). There is potential that peer review could be used to model reframing in classroom and laboratory settings when interpreting IR spectra. In connection with our findings, instructors could potentially leverage several modalities of assessment of student understanding of IR spectral interpretation. For example, open response items could be used to elicit student ideas related to explanations of why they selected a specific response when choosing a molecule in connection with a provided spectrum (such as in the task presented in Fig. 2). Instructors could also use open response through creative exercises (CEs), providing students with a prompt containing an IR spectrum and/or structures and asking them to provide as many clear, distinct, accurate, and relevant statements to the prompt as possible (Ye and Lewis, 2014; Wang and Lewis, 2020). Additionally, instructors could require students to first draw each structure as an initial component prior to choosing an answer. If limited to multiple-choice assessment formats, instructors could leverage the use of a reasoning tier within a two-tiered assessment (Treagust, 1986), with an example shown in Appendix 4. Specific responses to this reasoning tier would then allow the instructor to provide feedback and directly model how to move between data and frame through each of the tiered response options.

Limitations

As this study was part of a larger eye tracking study, students answered interview questions while watching a retrospective video recording of their eye movements when initially viewing the task. While this has the potential to enhance aspects of discussion surrounding their thought processes, it may have also limited potential discussion of prior knowledge, as students focused on what they were looking at on the screen when describing their thought processes. In addition, while the undergraduate participants were recruited from the laboratory course, the eye tracking study occurred in a research space distinct from the laboratory environment. The tasks in this study also featured only organic compounds, and as IR spectroscopy is used in many disciplines beyond organic chemistry, there is limited knowledge on how this work transfers beyond the context of the spectral analysis of organic compounds, particularly those that rely heavily on wavenumbers below 1500 cm⁻¹. The task itself only provided two molecular formulae, and while it was designed to provide multiple potential molecules for option B, this still may have narrowed the potential activated resources and potential frames established. The students were directed to engage in data analysis involving only these stimuli, which may have inhibited their potential resource activation and framing. Finally, the use of one task when analyzing the data limits the generalizability of the study, and the findings may be specific to this isolated task.

Author contributions

In an effort to normalize the practice of transparency in the preparation of this work, the specific contributions of all authors are described as follows: study design – DH, MBA; data collection – DH, MBA; development of data analysis plan – DH, MBA; data analysis – DH, KH; interpretation of results – DH, KH, MBA; writing – DH, KH, MBA; editing – DH, KH, MBA; funding and resources – MBA.

Data availability

Qualitative data within the current manuscript have been collected from human participants. Thus, under IRB regulations at the institution of study for IRB-21-87, data are not available due to confidentiality reasons in order to maintain anonymity.

Conflicts of interest

There are no conflicts of interest to declare.

Appendices

Appendix 1

Demographic information of all research study participants (N = 15) is indicated in the following table.

Gender	n	Race/ethnicity	n	Year of study	n	Major of study	n
Man	3	African American/black	3	Sophomore	1	Chemistry	1
Non-binary	2	Asian/Asian American	4	Junior	4	Biology	5
Woman	9	Hispanic/Latine/Chicane	3	Senior	4	Biochemistry	3
Chose not to specify	1	White/Caucasian	6	Graduate student	6	Graduate student	6
		Chose not to specify	1

Appendix 2

The original task prior to post-pilot task edits is shown below. The number of answer options was reduced from three (shown below) to two options in the final version of the task. The spectrum remained the same, with the appropriate answer of propyne. Option A was changed to propane.

Appendix 3

Codebook used for deductive coding, containing categories of data, frame, and conclusion (Klein et al., 2006; Zhou and Moon, 2023). Codes signifying direct observations regarding the spectrum and molecular formulae were grouped within the data category. Codes signifying students’ utilization of prior knowledge or experiences to interpret observations were grouped within the frame category. Finally, codes indicative of students formulating conclusions based on the data and their frame were grouped within the conclusion category.

Category	Definition	Representative quote
Data	Direct observations regarding the spectrum and molecular formulae (Klein et al., 2006)	Jay: “There is [an] alkyne peak around 3300 and then in 2200, carbon-triple bond-carbon.”
Frame	Utilization of prior knowledge or experiences to interpret observations (Klein et al., 2006)	Vance: “I knew [option B was] similar to A… So I was like, oh, maybe it's a double bond instead of just all single bonded.”
Conclusion	Formulation of conclusion based on data and frame (Zhou and Moon, 2023)	Danni: “And I found that…on the graphs, I was like, it has to be [option A]…because the other one is lacking the peaks that should be corresponded with that formula.”

Appendix 4

An example of a tiered assessment, containing both the multiple-choice item and reasoning tier, providing context for why students selected a specific response to the initial item.

Acknowledgements

The authors would like to thank the students who took time to participate in this research study and the instructors who allowed us to recruit participants. The authors would also like to thank the Department of Chemistry at the University of North Texas for support.

References

1. Bain K., Rodriguez J.-M. G. and Towns M. H., (2019), Chemistry and Mathematics: Research and Frameworks To Explore Student Reasoning, J. Chem. Educ., 96(10), 2086–2096 DOI:10.1021/acs.jchemed.9b00523.
2. Bayeck R. Y., (2022), Positionality: The Interplay of Space, Context and Identity, Int. J. Qual. Methods, 21, 16094069221114745 DOI:10.1177/16094069221114745.
3. Berg S. A. and Moon A., (2022), Prompting hypothetical social comparisons to support chemistry students’ data analysis and interpretations, Chem. Educ. Res. Pract., 23(1), 124–136 10.1039/D1RP00213A.
4. Berg S. A. and Moon A., (2023), A characterization of chemistry learners’ engagement in data analysis and interpretation, Chem. Educ. Res. Pract., 24(1), 36–49 10.1039/D2RP00154C.
5. Bott E., (2010), Favourites and others: reflexivity and the shaping of subjectivities and data in qualitative research, Qual. Res., 10(2), 159–173 DOI:10.1177/1468794109356736.
6. Braun V. and Clarke V., (2021), One size fits all? What counts as quality practice in (reflexive) thematic analysis? Qual. Res. Psychol., 18(3), 328–352 DOI:10.1080/14780887.2020.1769238.
7. Braun V. and Clarke V., (2023), Is thematic analysis used well in health psychology? A critical review of published research, with recommendations for quality practice and reporting, Health Psychol. Rev., 17(4), 695–718 DOI:10.1080/17437199.2022.2161594.
8. Cartrette D. P. and Bodner G. M., (2010), Non-mathematical problem solving in organic chemistry, J. Res. Sci. Teach., 47(6), 643–660 DOI:10.1002/tea.20306.
9. Committee on Professional Training, (2023), ACS Guidelines for Bachelor's Degree Programs, https://www.acs.org/education/policies/acs-approval-program/guidelines.html (accessed 2025-02-21).
10. Connor M. C., Finkenstaedt-Quinn S. A. and Shultz G. V., (2019), Constraints on organic chemistry students’ reasoning during IR and ¹H NMR spectral interpretation, Chem. Educ. Res. Pract., 20(3), 522–541 10.1039/C9RP00033J.
11. Connor M. C., Glass B. H., Finkenstaedt-Quinn S. A. and Shultz G. V., (2021), Developing Expertise in 1H NMR Spectral Interpretation, J. Org. Chem., 86(2), 1385–1395 DOI:10.1021/acs.joc.0c01398.
12. Connor M. C. and Raker J. R., (2022), Instrumentation Use in Postsecondary Instructional Chemistry Laboratory Courses: Results from a National Survey, J. Chem. Educ., 99(9), 3143–3154 DOI:10.1021/acs.jchemed.2c00415.
13. Cook D. A. and Hatala R., (2016), Validation of educational assessments: a primer for simulation and beyond, Adv. Simul., 1(1), 31 DOI:10.1186/s41077-016-0033-y.
14. Cooper M. M., Caballero M. D., Ebert-May D., Fata-Hartley C. L., Jardeleza S. E. and Krajcik J. S. et al., (2015), Challenge faculty to transform STEM learning, Science, 350(6258), 281–282 DOI:10.1126/science.aab0933.
15. Crandell O. M., Lockhart M. A. and Cooper M. M., (2020), Arrows on the Page Are Not a Good Gauge: Evidence for the Importance of Causal Mechanistic Explanations about Nucleophilic Substitution in Organic Chemistry, J. Chem. Educ., 97(2), 313–327 DOI:10.1021/acs.jchemed.9b00815.
16. Crandell O. M. and Pazicni S., (2023), Leveraging cognitive resources to investigate the impact of molecular orientation on students’ activation of symmetry resources, Chem. Educ. Res. Pract., 24(1), 353–368 10.1039/D2RP00164K.
17. Cullipher S. and Sevian H., (2015), Atoms versus Bonds: How Students Look at Spectra, J. Chem. Educ., 92(12), 1996–2005 DOI:10.1021/acs.jchemed.5b00529.
18. Deng J. M. and Flynn A. B., (2021), Reasoning, granularity, and comparisons in students’ arguments on two organic chemistry items, Chem. Educ. Res. Pract., 22(3), 749–771 10.1039/D0RP00320D.
19. Dood A. J., Dood J. C., de Arellano D. C.-R., Fields K. B. and Raker J. R., (2020), Analyzing explanations of substitution reactions using lexical analysis and logistic regression techniques, Chem. Educ. Res. Pract., 21(1), 267–286 10.1039/C9RP00148D.
20. Eckhard J., Rodemer M., Langner A., Bernholt S. and Graulich N., (2022), Let's frame it differently – analysis of instructors’ mechanistic explanations, Chem. Educ. Res. Pract., 23(1), 78–99 10.1039/D1RP00064K.
21. Erduran S., (2022), Argumentation in Chemistry Education, The Royal Society of Chemistry.
22. Foote M. Q. and Bartell T. G., (2011), Pathways to equity in mathematics education: how life experiences impact researcher positionality, Educ. Stud. Math., 78(1), 45–68.
23. Gillborn D., Warmington P. and Demack S., (2018), QuantCrit: education, policy, ‘Big Data’ and principles for a critical race theory of statistics, Race Ethn. Educ., 21(2), 158–179 DOI:10.1080/13613324.2017.1377417.
24. Hamilton D., Castillo A. and Atkinson M. B., (2024), Survey of Instrumentation Use in Industry: What Does Industry Want New Chemists to Know? J. Chem. Educ., 101(5), 1883–1890 DOI:10.1021/acs.jchemed.3c00990.
25. Heisterkamp K. and Talanquer V., (2015), Interpreting Data: The Hybrid Mind, J. Chem. Educ., 92(12), 1988–1995 DOI:10.1021/acs.jchemed.5b00589.
26. Holmqvist K., Nyström M., Andersson R., Dewhurst R., Jarodzka H. and van de Weijer J., (2011), Eye Tracking: A Comprehensive Guide To Methods And Measures, Oxford, UK: Oxford University Press.
27. Hyrskykari A., Ovaska S., Majaranta P., Räihä K.-J. and Lehtinen M., (2008), Gaze Path Stimulation in Retrospective Think-Aloud, J. Eye Mov. Res., 2, 1–18 DOI:10.16910/jemr.2.4.5.
28. Jamieson M. K., Govaart G. H. and Pownall M., (2023), Reflexivity in quantitative research: a rationale and beginner's guide, Soc. Personal. Psychol. Compass, 17(4), e12735 DOI:10.1111/spc3.12735.
29. Klein G., Moon B. and Hoffman R. R., (2006), Making Sense of Sensemaking 2: A Macrocognitive Model, IEEE Intell. Syst., 21(5), 88–92 DOI:10.1109/MIS.2006.100.
30. Klein G., Phillips J. K., Rall E. L. and Peluso D., (2007), A Data-Frame Theory of Sensemaking. Expertise Out of Context: Proceedings of the Sixth International Conference on Naturalistic Decision Making, Psychology Press.
31. Kozma R. and Russell J., (2005), Students Becoming Chemists: Developing Representationl Competence, pp. 121–145 DOI:10.1007/1-4020-3613-2_8.
32. Lieber L. and Graulich N., (2022), Investigating students’ argumentation when judging the plausibility of alternative reaction pathways in organic chemistry, Chem. Educ. Res. Pract., 23(1), 38–54 10.1039/D1RP00145K.
33. Lincoln Y. S. and Guba E. G., (1986), But is it rigorous? Trustworthiness and authenticity in naturalistic evaluation, New Dir. Program Eval., 1986(30), 73–84 DOI:10.1002/ev.1427.
34. Lyle J., (2003), Stimulated recall: a report on its use in naturalistic research, Br. Educ. Res. J., 29(6), 861–878 DOI:10.1080/0141192032000137349.
35. Mauthner N. S. and Doucet A., (2003), Reflexive Accounts and Accounts of Reflexivity in Qualitative Data Analysis, Sociology, 37(3), 413–431 DOI:10.1177/00380385030373002.
36. Meister S. and Upmeier zu Belzen A., (2021), Analysis of Data-Based Scientific Reasoning from a Product-Based and a Process-Based Perspective, Educ. Sci., 11(10), 639 DOI:10.3390/educsci11100639.
37. National Academies Press, (2012), A Framework for K-12 Science Education: Practices, Crosscutting Concepts, and Core Ideas, National Academies Press DOI:10.17226/13165.
38. Noyes K., Carlson C. G., Stoltzfus J. R., Schwarz C. V., Long T. M. and Cooper M. M., (2022), A Deep Look into Designing a Task and Coding Scheme through the Lens of Causal Mechanistic Reasoning, J. Chem. Educ., 99(2), 874–885 DOI:10.1021/acs.jchemed.1c00959.
39. Otter.ai, (2024), https://otter.ai/ (accessed 2024-09-12).
40. Palinkas L. A., Horwitz S. M., Green C. A., Wisdom J. P., Duan N. and Hoagwood K., (2015), Purposeful sampling for qualitative data collection and analysis in mixed method implementation research. Adm. Policy Ment. Health, 42(5), 533–544 DOI:10.1007/s10488-013-0528-y.
41. Reynolds J., Kizito J., Ezumah N., Mangesho P., Allen E. and Chandler C., (2011), Quality assurance of qualitative research: a review of the discourse, Health Res. Policy Syst., 9, 43 DOI:10.1186/1478-4505-9-43.
42. Rodriguez A. J. and Navarro-Camacho M., (2023), Claiming Your Own Identity and Positionality: The First Steps toward Establishing Equity and Social Justice in Science Education, Educ. Sci., 13(7), 652 DOI:10.3390/educsci13070652.
43. Secules S., McCall C., Mejia J. A., Beebe C., Masters A. S., L. Sánchez-Peña M. and Svyantek M., (2021), Positionality practices and dimensions of impact on equity research: a collaborative inquiry and call to the community, J. Eng. Educ., 110(1), 19–43 DOI:10.1002/jee.20377.
44. Slominski T., Fugleberg A., Christensen W. M., Buncher J. B. and Momsen J. L., (2020), Using Framing as a Lens to Understand Context Effects on Expert Reasoning, CBE—Life Sci. Educ., 19(3), ar48 DOI:10.1187/cbe.19-11-0230.
45. Stowe R. L. and Cooper M. M., (2019), Arguing from Spectroscopic Evidence, J. Chem. Educ., 96(10), 2072–2085 DOI:10.1021/acs.jchemed.9b00550.
46. Tashiro J. and Talanquer V., (2021), Exploring Inequities in a Traditional and a Reformed General Chemistry Course, J. Chem. Educ., 98(12), 3680–3692 DOI:10.1021/acs.jchemed.1c00821.
47. Topczewski J. J., Topczewski A. M., Tang H., Kendhammer L. K. and Pienta N. J., (2017), NMR Spectra through the Eyes of a Student: Eye Tracking Applied to NMR Items, J. Chem. Educ., 94(1), 29–37 DOI:10.1021/acs.jchemed.6b00528.
48. Treagust D. F., (1986), Evaluating students’ misconceptions by means of diagnostic multiple choice items, Res. Sci. Educ., 16(1), 199–207.
49. Urbanek M. T., Moritz B. and Moon A., (2023), Exploring students’ dominant approaches to handling epistemic uncertainty when engaging in argument from evidence, Chem. Educ. Res. Pract., 24(4), 1142–1152 10.1039/D3RP00035D.
50. Wallace W. E., (2022), Infrared Spectra, in Linstrom P. J. and Mallard W. G. (ed.), NIST Chemistry WebBook, NIST Standard Reference Database Number 69, National Institute of Standards and Technology, https://webbook.nist.gov/chemistry/ (accessed 2024-09-12).
51. Wang Y. and Lewis S. E., (2020), Analytical chemistry students’ explanatory statements in the context of their corresponding lecture, Chem. Educ. Res. Pract., 21(4), 1183–1198 10.1039/D0RP00063A.
52. Watts F. M., Zaimi I., Kranz D., Graulich N. and Shultz G. V., (2021), Investigating students’ reasoning over time for case comparisons of acyl transfer reaction mechanisms. Chem. Educ. Res. Pract., 22(2), 364–381 10.1039/D0RP00298D.
53. Wright L. C. and Oliver-Hoyo M. T., (2020), Student assumptions and mental models encountered in IR spectroscopy instruction, Chem. Educ. Res. Pract., 21(1), 426–437 10.1039/C9RP00113A.
54. Xue D. and Stains M., (2020), Exploring Students’ Understanding of Resonance and Its Relationship to Instruction. J. Chem. Educ., 97(4), 894–902 DOI:10.1021/acs.jchemed.0c00066.
55. Ye L. and Lewis S. E., (2014), Looking for links: examining student responses in creative exercises for evidence of linking chemistry concepts, Chem. Educ. Res. Pract., 15(4), 576–586 10.1039/C4RP00086B.
56. Zhou J. and Moon A., (2023), “To Be Honest, I Didn’t Even Use the Data”: Organic Chemistry Students’ Engagement in Data Analysis and Interpretation, J. Chem. Educ., 100(1), 80–90 DOI:10.1021/acs.jchemed.2c00840.

---