Using drawing as a tool for investigating undergraduate conceptions of Earth scientists

Abstract The purpose of this study was to investigate undergraduate students’ conceptions of Earth scientists, using drawing as a tool, during introductory Earth science courses. We explored two research questions: 1) What student conceptions are evident in undergraduate students’ drawings of Earth scientists? and 2) How do undergraduate students’ conceptions of Earth scientists—as evidenced in their drawings—change as a result of completing an introductory Earth science course? We collected pre- and post-drawings of Earth scientists from 94 students in six introductory Earth science courses at two universities and coded the 188 drawings across 39 indicators. We used Chi Square Goodness of Fit Tests to identify significant shifts between pre- and post-drawings and effect sizes with Cohen’s W. Twelve indicators demonstrated higher frequencies and eleven indicators showed significant change (with small, medium, and large effects) between the pre- to post-drawings collected at the beginning and end of each course. The results suggest that unsophisticated conceptions exist, especially at the beginning of an introductory Earth science course. Yet overall, findings indicate that in the courses from which we drew our data, most students had an informed view of Earth science as a field and the work of Earth scientists. Additionally, student ideas about Earth scientists primarily investigating rocks, minerals, and soil, expanded to include other areas of investigation such as the atmosphere and bodies of water. We propose that drawings can serve as an additional valuable tool in an instructor’s toolbox for understanding students’ conceptions of Earth scientists, which has important implications for instruction and curriculum design.


Introduction
Undergraduate students bring a set of beliefs, originating from prior experiences, including a variety of stereotypes about scientists and scientific work (Christidou et al., 2016).Research suggests that undergraduate student conceptions about scientists may affect their attitudes toward science and future career choices (Christidou et al., 2016;DeWitt et al., 2013;Farland-Smith et al., 2014, Finson, 2003).Especially, negative conceptions of scientists or images that do not fit students' beliefs about themselves may interfere with student interest in learning (Christidou et al., 2016;Farland-Smith, 2012;van der Hoeven Kraft, 2017).For example, in a study conducted by Hong and Lin-Siegler (2012), students' interest in solving complex physics problems increased after reading about the struggles of scientists and how they achieved their goals.This provides an illustration of how student conceptions can change to make science and scientists more relatable using personal experience.
We can better understand undergraduate student conceptions about scientists by studying them, then take the next step to design learning experiences that challenge students' preexisting beliefs and help them develop realistic ideas and positive attitudes toward science and scientists (Avraamidou, 2013).Prior work tells us that K-12 students hold naïve conceptions of several geoscience phenomena such as volcanoes (Anderson & Libarkin, 2016), plate tectonics (Hoyer & Hastie, 2023), rivers (Sexton, 2012), and groundwater (Arthurs, 2019).In an undergraduate geoscience education research, Manzanares et al. (2023) investigated alternative conceptions that students bring to a mineralogy course and found that experiences in prior science courses and outdoor activities impacted these conceptions.Understanding the origin and nature of alternative conceptions provides deeper insight that allows us to better support undergraduate students when learning about geoscience topics (Manzanares et al., 2023).
One popular practice for understanding students' conceptions of scientists is a drawing tool, the "Draw a Scientist Test" (DAST) (Chambers, 1983), which asks students to draw a picture of a scientist at work.More recently, researchers have developed drawing tools that target student conceptions of discipline-specific scientists.For example, recent studies used a Draw an Engineer test (DAET) and found that student conceptions of engineering work mostly depicted physical labor, such as building things (Hammack et al., 2020;Kuvac & Koc, 2022;Thomas et al., 2020).Additional work using the Draw an Environmental Scientist Test (DET) revealed images of environmental scientists as generally male and committed to environmental protection (Dikmenli et al., 2010;Joo et al., 2008;Yang et al., 2021).
Many college students (STEM and non-STEM) pursue Earth science courses as a requirement for their major, including education majors, which suggests the importance of addressing student conceptions of Earth Science and Earth scientists.However, to date, there are no drawing tools to reveal undergraduate students' conceptions of Earth scientists.Past research suggests that K-12 teachers who avoid or lack enthusiasm for teaching science usually have negative beliefs about science (Lotter et al., 2007;McNeill et al., 2013, Menon, 2020).Teachers' beliefs about science, science teaching, and science learning are critical to future science instruction; therefore, it is important to shape instruction that supports the development of positive attitudes toward science (Knight & Cunningham, 2004) within undergraduate science coursework.Further, understanding the conceptions undergraduate students hold generally as they enter Earth science courses is important for designing interventions to help develop a realistic understanding of the nature of the work of Earth scientists.However, this area remains unexplored through drawing in Earth science.The purpose of this study is to address this gap by investigating changes in undergraduate students' drawings of Earth scientists after their participation in introductory Earth science courses.

Draw a scientist tests
Early developers of "draw a scientist" tests include Chambers (1983), Finson et al. (1995), and Thomas et al. (2001).Chambers' original work spanned eleven years and is well known as the Draw a Scientist Test (DAST).Using simple procedures that involved asking children to "draw a picture of a scientist, " Chambers collected drawings from 4,807 children in 186 classes from kindergarten to fifth grade.Chambers developed a list of seven indicators of standard images of scientists (including lab coats, eyeglasses, facial hair, symbols of research including scientific instruments and laboratory equipment, symbols of knowledge including books and filing cabinets, technology, and relevant captions such as formulae, taxonomic classification, and "Eureka!") and used these indicators to analyze the drawings.Chambers identified that: 1) stereotypical images of scientists appeared among grade school students, and 2) stereotypical images appeared with greater frequency as students advanced through the grades.These were especially significant findings for scientists and science educators at the time.Finson et al. (1995) advanced Chamber's work by expanding his original list of seven indicators to include eight "alternative images" that were either evident in earlier work (Mead & Metraux, 1957) or that, with time, had become aspects of increased interest (race and gender).The "Draw a Scientist Test-Checklist" (DAST-C) included explicit instructions for scoring and guidance for analysis.Finson et al. (1995) conducted a field test of the DAST-C using preand post-drawings from 47 eighth-grade students divided into control and treatment groups.The treatment group completed a career-oriented interdisciplinary program designed to purposely expose students to alternative images of scientists.The researchers concluded that the DAST-C was useful in looking at grouped and overall element scores.The DAST-C enhanced analysis of the drawings by making an item analysis possible and response patterns more easily observable.Analysis of data from eighth-grade students in the treatment group revealed a significant shift from stereotypical images of scientists to more realistic images of the variety of persons involved in science as their contact with scientists increased.Thomas et al. (2001) reported on a further modification of the DAST-C with their development of the Draw a Science Teacher Test-Checklist (DASTT-C), which they expected would illuminate the knowledge and beliefs of preservice elementary teachers enrolled in science teaching methods courses.The checklist developed by Thomas et al. (2001) included attributes reflective of teacher-centered classrooms (teacher at the center of instruction and learning) and student-centered classrooms (students at the center of learning and teacher facilitating activities)." Thomas et al. (2001) tested and adjusted the DASTT-C across multiple iterations that reviewed the illustrations of 850 preservice teachers.This resulted in the addition of a descriptive narrative to confirm the evaluator's understanding of images in the drawings and assist with scoring.The narrative consisted of two questions: 1) What is the teacher doing?and 2) What are the students doing?Through validation of the DASTT-C, they found that for preservice teachers, the process of reflective drawing supported long-term professional contemplation about teaching elementary science.
Test development and expansion have continued with multiple disciplines adding to the range of test types.Rule et al. (2007) investigated preservice teachers' conceptions of clay scientists, Joo et al. (2008) used a Draw an Environmental Scientist Test to examine the conceptions of environmental scientists among high school students, and most recently, Kuvac and Koc (2022) developed a Draw an Engineer Test.However, to our knowledge, a drawing test for specifically eliciting student conceptions of Earth scientists does not exist.

Earth science conceptions and drawing
Although we are unable to find earlier studies that investigate student conceptions specifically about Earth scientists, much prior work has focused on student conceptions (and alternative conceptions) about Earth science.From a review of the literature, Dove (1998) documented the most common student (K-16+) alternative conceptions about rocks, earthquakes, volcanoes, Earth's structure, weathering and erosion, and soil.He concluded that some teaching practices, such as the use of everyday language, oversimplification, and textbook stereotyping, may be contributing factors to forming alternative conceptions.Dove's (1998) work describes "alternative conceptions in Earth science" (p.183) but targets geologic conceptions without expanding to broader Earth topics, including weather, climate, and oceans.Francek (2013) built on Dove's work with a compilation and review of over 500 "geoscience misconceptions" (p.31).Organized into eleven categories and broken down by age groups, he found that plate tectonics and erosion/weathering had the most misconceptions, especially at high-school and middle-school levels.The misconceptions were numerous and directly related to geoscience content rather than what Earth scientists do (e.g., only continents move, not oceans) while targeting specific geologic concepts.Henriques wrote in 2002 that "It is interesting to note that there were no articles in the Journal of Geoscience Education addressing meteorology misconceptions" (p.203) but found from a review of the literature on children's ideas about weather that there were many, including those about phase changes of water and the ability of matter to be invisible, but still exist.Like Francek's (2013) identification of 500 geoscience misconceptions, Feller (2007) compiled 100 misconceptions about the ocean with data collected from undergraduate classes.Here we see a few misconceptions related to scientists, e.g., Jacques Cousteau saved the ocean, we regularly monitor all areas of the ocean, and we have the technology to dive to any ocean depth.
Beyond broadly identifying student conceptions in Earth science, several studies additionally address conceptions about specific Earth science phenomena, e.g., the role of rivers in canyon formation (Sexton, 2012), sequence stratigraphy (Herrera & Riggs, 2013), the formation of eskers and erratics (Arrhenius et al., 2021), tornado wind speed (Van Den Broeke & Arthurs, 2015), and groundwater (Arthurs, 2019).Broadly speaking, these studies find a multitude of alternative conceptions among students regarding varied Earth science processes.
There is also rich literature on the use of drawing/sketching geoscience concepts in undergraduate geoscience courses.Smith and Bermea (2012) used student sketches to uncover students' alternative conceptions about plate tectonics and Harris and Gold (2018) assessed students' representations of the greenhouse effect using student-generated concept sketches.Johnson and Reynolds (2005) advocate for both instructors and students producing concept sketching as a method for identifying student conceptions, as formative assessment to guide homework, and for assessing student understanding in exams.Building on this, Arthurs et al. (2020) found that incorporating delimited sketch activities supported student knowledge incorporation about groundwater and moved students toward more expert-like conceptions.Finally, Ormand et al. (2017) promote the use of predictive sketching in the Spatial Thinking Workbook as a way for students to iteratively build skills for visualizing cross-sections of geological block diagrams.

Theoretical perspective
In using student drawings to evaluate student conceptions about Earth scientists, we contend that the drawings accurately and authentically represent these conceptions.We use a theoretical construct of contextualization (Halldén et al., 2008), rooted in a constructivist tradition to underpin these assumptions.A constructivist theory of learning maintains that learners construct knowledge rather than just passively taking in information (Bada & Olusegun, 2015) and that these knowledge constructions assemble as networks of multiple knowledge elements (Hiebert & Carpenter, 1992).By layering contextualization on top of this, we additionally consider the contexts in which the students are operating and of which conceptions are the product (Nilsson & Ryve, 2010).Nilsson and Ryve (2010) argue that when using a constructivist perspective, context refers to the cognitive context shaped by the learner's personal interpretations.They identify three cognitive contexts.
First, the conceptual context considers the personal constructions of concepts embedded in a subject area.Undergraduate students arrive in science courses with different educational backgrounds (Kokkelenberg & Sinha, 2010).Many have not been in an Earth science course since middle school and may rely on what they remember from that experience to contextualize their drawing.Students may have completed a variety of other science courses, such as biology and environmental science, or have other geoscience experiences, such as using a well or collecting rocks and fossils.Some students may have active interests in environmentalism or read news about climate change.Others may simply conjure up ideas about what they think represents a scientist or "Earth." These experiences, or lack thereof, influence student understanding of what Earth scientists do and who they are, which may be evident in student drawings.
The second context is situational and refers to interpretations made in the interaction between the learner and the immediate surroundings (Nilsson & Ryve, 2010).Instructors typically fill Earth science classrooms with representations of the discipline.If there are posters of the water cycle, displays of fossils, or drawers full of rocks, these certainly contextualize students' conceptions.Additionally, a student who lacks ideas about Earth scientists may assume that the instructor is an Earth scientist; why not just draw them?We contend that timing matters as well, for example, if students draw an Earth scientist on the first and last day of class this situation also highly contextualizes the drawings, and we must consider this context along with any conclusions made.
The third context we consider is the cultural context, which refers to the construction of discursive rules, conventions, and behavior patterns (Nilsson & Ryve, 2010).Prior work (Chambers, 1983;Finson, 2002;McCann & Marek, 2016) has powerfully demonstrated that a persistent stereotype of scientists exists as predominantly white and male (Finson, 2003;Monhardt, 2003), although Miller et al. (2018) found changes in K-12 students' stereotypes as more female scientists were depicted in younger grade-level students drawings.The geosciences widely recognize this stereotype and actively challenge it, especially through efforts like The Bearded Lady Project: Challenging the Face of Science (Currano et al., 2023).Although work to dispel stereotypes is great, it takes time to reach broad populations and become normative; thus, it is important to recognize that when asked to draw a scientist, a cultural context may construct a prescriptive scenario, especially for students who operate largely outside the domain of science.
We cannot ask questions about student conceptions outside of context; therefore, using a theoretical construct of contextualization explicitly recognizes that multiple contexts inform students' conceptions and how they manifest in student drawings.Explicitly recognizing this theoretical perspective, we ask the following research questions: 1. What student conceptions are evident in undergraduate students' drawings of Earth scientists? 2. How do undergraduate students' conceptions of Earth scientists-as evidenced in their drawings-change as a result of completing an introductory Earth science course?

Research design
We used an exploratory sequential mixed methods design (Creswell et al., 2011) in which we collected data in two phases.We analyzed the first data set using qualitative methods (coding) to determine what student conceptions were evident in the data.We used these results to develop a scoring checklist that we applied to the entire data set after a subsequent phase of data collection.After scoring, we conducted quantitative analyses with the intent to explain relationships found in the qualitative data.We used Chi-Square Goodness of Fit Tests to test for significant differences between pre-and post-drawings for each indicator in the checklist and calculated the effect size (Cohen's W) for all variables with a significant difference.

Study setting and participants
This study evolved from a larger study that sought to investigate preservice teachers' self-efficacy for teaching science courses after their experience in an Earth science course involving a research-based experience to develop skills for scientific inquiry.As part of this study, we collected drawings of Earth scientists from the students (preservice teachers) at the beginning and end of the semester.Because we found these data to be compelling, with a greater potential for investigating how student conceptions of Earth scientists change, we continued to pursue this additional line of research.To do so, we collected data at two large public universities in seven introductory Earth science courses.Both universities are mid-sized with Earth science (geoscience) and education programs.We collected data from three types of Earth science courses and two kinds of geology courses.  1 provides the distribution of courses, instructors, and students in these courses.We include course descriptions in the supplementary online materials.
We collected data from 94 undergraduate students enrolled in these courses.Table 2 presents representative participant demographic information and an average of prior science courses (high school and college) taken prior to participation in the study (not including the course in which they were currently enrolled).As described in the following section, this information is from a subsample (n = 67) of our participants collected during Phase 2 of the study.We collected data during the Spring and Fall semesters of 2021, during the COVID-19 pandemic.Thus, our data collection during the Spring of 2021 occurred within courses taught using a virtual format; however, by the Fall of 2021, all classes were back in-person, and we collected data accordingly.At both universities, the course instructor read the recruitment script, and participants who volunteered to participate signed electronic consent forms, accessed either with a link or with phones via a projected QR code.The Institutional Review Board at University 1 approved the study and all procedures.

Data collection and development of the Draw an Earth Scientist Test and Scoring Checklist
We collected data at the beginning and end of each semester (as close to the first and last day of class as possible yet left to the preferences of individual instructors).As we collected data, we concurrently improved our process for data collection and developed the scoring checklist.Please refer to Figure 1.

Phase 1
We collected data for Phase 1 from 27 participants (a total of 54 drawings) in a virtual format due to the COVID-19 pandemic.As part of a week's assignment, we asked participants to simply "draw a picture of an Earth scientist." Participants used materials they had on hand; some used colored pencils or markers, while others provided pen or pencil drawings.
In the literature on drawing/sketching in the geosciences coding the drawings is a common method used for identifying frequencies of salient aspects (Arthurs et al., 2020;Harris & Gold, 2018;Kirby et al., 2022;Smith & Bermea, 2012;Soltis et al., 2021).Taking this into account and concurrent with data collection in Phase 1, we surveyed the broad "Draw a Scientist" and conceptual change literature to explore how studies framed their scoring methods.We considered four different frameworks.The most common method for scoring drawings uses a dichotomous checklist in which scorers mark salient items as either present or absent.This is the original method used with Chambers' DAST (1983), scored across seven literature-informed indicators representative of standard images of scientists.Finson et al. (1995)  A second method for scoring drawings uses broad constructs to frame dichotomously scored indicators.We see this type of scoring with the DASTT-C (Thomas et al., 2001), which divides scoring into three sections: teachers, students, and environment, with individual indicators related to activity and position.Although this produces a total score through tallying all scores, researchers can consider each broad construct independently.Minogue (2010) successfully used the DASTT-C with preservice elementary education students and noted that "as with most assessment rubrics, there was some ambiguity and thus unintended room for subjectivity" (Minogue, 2010, p. 776).
Farland-Smith (2012) developed a third method for a modified Draw a Scientist Test with more extensive directions to purposefully elicit representations of the scientist's location, appearance, and activity.She scored the three constructs (location, appearance, and activity) on a continuum from 0 to 2, labeled "sensationalized, " "traditional, " and "broader than traditional, " respectively.Farland- Smith (2012) claimed that this method of scoring separation allowed for "perceptions to be viewed in three different lenses, creating a kaleidoscope of…perceptions versus a checklist" (Farland- Smith, 2012, p. 115).Similarly, Thomas et al. ( 2020) used a scoring method with a modified Draw an Engineer Test (mDAET) with four domains, including an engineer's use of mathematics, use of science, gender stereotype, and work.She scored these on a continuum from 0 to to solicit gender data, we provided participants with a blank to fill in ("Gender_________"). these data represent participant responses.2, with 0 representing vague, superficial understanding and 2 as detailed, explicit understanding.Descriptions provided in the rubric served to manage objectivity and discourage interpretation or assumed meanings.
We additionally reviewed the Student Perceptions about Earth Sciences Survey (SPESS) (Jolley et al., 2012).Although not a drawing test, this tool for measuring shifts in students' perceptions of Earth and ocean sciences is geoscience specific and thus lends insight into the norms defined by the community.From the SPESS, we gleaned a list of ten potential items for inclusion in a checklist (e.g., nature, human impact, investigation, natural processes, etc.) Once data collection was complete, and guided by this insight, we conducted open coding of the drawings to analyze the content and create codes (Miles & Huberman, 1994) for use in a checklist for scoring the drawings.
In our first scoring attempt, we created checklist #1.We used a modified version of Farland-Smith ( 2012) with three constructs-location, appearance, and activity-scored on a modified continuum from 1 to 3 that we labeled: 1 -cannot be categorized; 2 -naive or traditional; and 3 -broader than traditional.The first and second authors individually scored all 54 drawings and compared scores.Notes kept at this stage of the study read, "I'm not sure that the scoring sys-tem…is effective at getting to the heart of what we are trying to find out…I saw so many nuances in the drawings that were difficult to categorize and likely represent interesting conceptions (that we are missing)." Thus, we abandoned this as a method for scoring.Subsequently, we created checklist #2 using the Finson et al. (1995) model.We organized the codes into five broad categories: physical appearance, location, activity/position, instruments/tools, and objects for study that incorporated codes from our content analysis.This early version of our final checklist used dichotomous scoring (present/absent) across 32 indicators.The first and second authors used checklist #2 to individually rescore the 54 drawings, compare scores, and resolve differences until we reached 100% agreement.

Phase 2
The first author presented preliminary results in an oral presentation at the Earth Educators Rendezvous held in July 2021.Although our purpose was to share particularly interesting findings with the community, we were especially interested in garnering feedback on our methods and generating interest in collaborations that would increase our data collection at additional universities.There was broad interest in the study, with four individuals volunteering to collect data, although not all these materialized.The most fervent feedback regarded interest in eliciting and analyzing representations of race and ethnicity.This is challenging when analyzing stick figure drawings, but we received a suggestion to supply participants with skin-tone colored pencils in hopes of spontaneously producing these representations.
Based on these experiences and in preparation for Fall 2021 in-person data collection, we modified some processes and made minor revisions to the checklist.First, we created a standardized template to use in lieu of a blank sheet of paper (see supplemental online materials).We added a section for participants to provide brief demographic data and indicate previously completed Earth science courses.We outlined a central area (big box) for the drawing and added two questions to the bottom: Where is the Earth scientist?and What is the Earth scientist doing?These two questions aligned with the methods developed by Thomas et al. (2001) and attempted to resolve some ambiguity we experienced during the first round of coding with checklist #2.Based on feedback received at the Earth Educators Rendezvous, we added a sixth category to checklist #2 ("Demographics of the Earth scientist"), made minor revisions to existing indicators (e.g., added "unclear" to the location category and collapsed some categories), and purchased twelve sets of skin-tone colored pencils for use at University 1.

Phase 3
During Phase 3, we collected an additional 134 drawings during Fall 2021 from participants in courses previously described.Due to the amount of data we collected and our desire to increase the validity and reliability of our findings, we invited two student researchers to join the study (the third and fourth authors).We adapted the training process highlighted in Thomas et al. (2020), which included 1) providing a brief introduction of the project and checklist development process, 2) explaining the checklist, including categories, and 3) providing and explaining two previously scored example drawings.We randomly chose two previously scored drawings and asked the new researchers to use checklist #2 to score them.We discussed discrepancies between the two new scores and revisited each category and subcategory to answer all questions.After discussion, we shared the original scores and again discussed discrepancies until we reached an understanding.We next selected two additional, unscored drawings and individually scored them, then compared and discussed discrepancies.We repeated this process until we consistently scored the drawings with a high degree of consensus.
After completing the initial training, we continued by scoring a set of 30 drawings in two pairs, consisting of one new researcher and one original researcher.Each pair met to resolve differences in coding, and then we all met to reach a consensus between the two pairs.We discussed multiple issues that surfaced.For example, a recurring discrepancy between researchers was the interpretation of drawings showcasing celestial bodies.One of the drawings had a picture of the sun.Two researchers coded the presence of a celestial body, which contrasted with the other two researchers, who indicated no celestial body.The researchers discussed whether the presence of the sun was merely decorative or if the participant meant to convey that it was an object of study.After discussion, we decided that even though the student included a celestial body in the drawing, there was no clear evidence that the Earth scientist was studying it.We also used the accompanying narratives, when possible, to make determinations.For example, in another drawing, the participant drew the moon and stars and explained that "the Earth scientist is out at night observing the moon." We decided to code for the presence of a celestial body because the participant indicated that the moon was an object of study.We held similar discussions for each category regarding what counted as the presence or absence of an indicator as depicted in the drawings.
Throughout the process, we kept memos to document issues along with aspects of the checklist that were ambiguous or subject to interpretation.We used these memos to 1) finalize the checklist, and 2) create a detailed table of code descriptions (see supplemental online materials) to use as an arbitrator of coding decisions.We investigated inter-coder reliability on this set of 30 coded drawings by calculating the Cohen Kappa coefficient for each indicator in two ways: between the members of one pair (average across all indicators = 0.803) and between the two pairs (average across all indicators = 0.820).We include the Kappa coefficients for the final checklist in the supplementary online materials; these numbers show that the coders were almost entirely in agreement (Sim & Wright, 2005).With this accomplished, the third and fourth authors continued to independently score the remaining 134 drawings using the final checklist.Like procedures we established during training, the two researchers compared results and resolved differences until they reached a consensus.

Statistical analysis
We used Statistical Package for Social Sciences (SPSS) to conduct the statistical analysis.We ran descriptive statistics and used the Chi-Square Goodness of Fit Test to test for significant differences between the pre-and post-drawings for each indicator in the checklist.We determined the Chi-Square value for each indicator by using the observed frequency of each indicator in the post-drawings and the expected frequency (how often each indicator appeared) in the pre-drawings.Because one of the assumptions of the Chi Square Goodness of Fit Test is that the expected frequency should be at least five, we excluded any sub-categories with fewer than five occurrences.Twenty-eight of 39 indicators met the assumptions for analysis using the Chi Square Goodness of Fit Test.In addition, we calculated the effect size (Cohen's W) for all variables with a significant difference (p < 0.05) between pre-and post-drawings.We considered effect sizes of 0.10 as small, 0.3 as medium, and 0.5 as large (Cohen, 1992).

Results
We present the results from the following five major categories: 1) physical appearance, 2) location, 3) activity/position, 4) instrument/tools, and 5) objects for study.Descriptive statistics for the pre-and post-drawings as well as the results of Chi-Square Goodness of Fit and Cohen's W tests are in the supplemental online materials.We are not including the categories "unclear" or "other" here and note that some aspects of the drawings were double coded (e.g., a drawing that had the location coded as "laboratory/office" was also coded as "indoor").We are not reporting results on the demographics of the Earth scientists for reasons described in the next section.

Research question #1
We asked, "What student conceptions are evident in undergraduate students' drawings of Earth scientists?"Using descriptive statistics across both pre-and post-drawings, with an arbitrary cutoff of at least 30 occurrences, we found most evident the representations listed in Table 3.The most frequent indicators across drawings included Earth scientists observing/collecting/analyzing data (n = 135; activity/position) outdoors or at a field site (n = 100; location).Students most frequently depicted Earth scientists with rocks, minerals, or soil (n = 72; objects for study).

Research question #2
We asked, "How do undergraduate students' conceptions of Earth scientists-as evidenced in their drawings-change as a result of completing an introductory Earth science course?"We saw significant change across 11 of the included indicators that did not violate the assumption of the Chi Square test with fewer than five occurrences.Five of the significant changes were in the positive direction (more post-test representations) and six of the significant changes were in the negative direction (fewer post-test representations).The largest positive change between pre-and post-drawings occurred with illustrations of clouds, weather, and atmospheric phenomenon as an object for study (Cohen's W = 0.689, large effect).Drawings of celestial bodies, such as the Sun, Moon, and other planets as objects for study also increased (Cohen's W = 0.362; medium effect).There was a small positive increase in the number of students who drew Earth scientists studying mountains (or volcanoes) and bodies of water (Cohen's W = 0.221 and 0.209; both small effects).
In the activity/position category, there was a small increase in the number of drawings that showed Earth scientists observing/collecting/analyzing data, but this number was already high in the pre-drawings (Cohen's W = 0.204; small effect).In the negative direction, there were fewer lab coats and glasses/goggles included as part of the Earth scientists' physical appearance (Cohen's W = 0.304 and 0.284; medium and small effect).Additionally, the students did not depict as many Earth scientists sharing/communicating/ teaching/demonstrating as an activity/position (Cohen's

Discussion
This study aims to investigate undergraduate students' conceptions of Earth scientists and how they change after completion of an introductory Earth science course using drawings.While there are other studies that have used drawings as a tool to investigate students' conceptions of scientists from different disciplines (e.g., Dikmenli et al., 2010;Hammack et al., 2020;Joo et al., 2008;Kuvac & Koc, 2022;Nairn, 1996;Thomas et al., 2020;Yang et al., 2021), this study contributes uniquely by focusing on conceptions about Earth scientists.We return to our theoretical perspective as a lens for viewing the results, specifically within three cognitive contexts (Nilsson & Ryve, 2010).These are: 1) the conceptual context that considers the personal constructions of concepts embedded in a subject area; 2) the situational context that refers to interpretations made in the interaction between the learner and the immediate surroundings; and 3) the cultural context that refers to the construction of discursive rules, conventions, and patterns of behavior.

Observing, collecting, and analyzing data
More than any other indicator, students included observing, collecting, and analyzing data as an activity central to the work of Earth scientists (e.g., Figure 2).By including "observing, " "collecting, " and "analyzing" in a single category, this category was broad; however, it captured students' conceptions of Earth scientists working in a field that is primarily observational and data-driven rather than experimental.From a conceptual context perspective, this suggests that Earth science instructors-both at the K-12 and undergraduate levels-are helping students build concepts about the centrality of data to understanding Earth systems.This not only characterizes the discipline but distinguishes it from science disciplines that focus more on experimentation (although no science discipline is exclusively one or the other).The Next Generation Science Standards (NGSS) reflect this distinction with Science and Engineering Practice (SEP) #4: Analyze and Interpret Data.The application and distribution of SEP #4 across different science disciplines is uneven; it is associated with 16% of the Earth & Space Science performance expectations compared to 13% in Life Science and only 7% in Physical Science (Kastens, 2015).Earth science research is observational, with data collection playing a key role.Experimentation occurs; however, it contrasts with higher levels of experimentation in chemistry and physics.Evidence of this distinct concept in our data is encouraging.It suggests that students build accurate conceptions about how Earth scientists work, build theories, and base claims.
The small positive increase in the number of students who drew Earth scientists observing, collecting, or analyzing data as an activity in the post-drawings (Cohen's W = 0.204; small effect) suggests that throughout the introductory Earth science courses in this study, data were an important component.Whether by involving students in data collection, or providing examples of Earth science investigations that generate data, the conceptual contexts that instructors are building continue to strengthen student ideas of Earth science as a discipline that collects considerable amounts of data.In the object of study category (versus the activity/position category) depictions of data decreased, however.This difference resulted from how we coded the categories; as an object of study, we coded for data if an Earth scientist was studying data (e.g., data displays, graphs, spreadsheets), whereas we included data generation (for example in the field or in a lab) when we coded for observing, collecting, and analyzing data.

Outdoor and field sites
Coinciding with observing, collecting, and analyzing data is the finding that many drawings (n = 100) included Earth scientists outdoors or at a field site (Figure 3).Students may be aware that Earth scientists conduct fieldwork, and this could contribute to a cultural context promoting ideas of Earth scientists working outdoors.Certainly, while accurate, it is also true that many Earth scientists work indoors.About half as many drawings included Earth scientists working indoors in a laboratory or office.We noticed that in some pre-drawings, students who presumably were unsure about how to depict an Earth scientist defaulted to drawing a stereotypical scientist, e.g., male in a lab coat, with goggles, glassware, or crazy hair-a description used by Farland-Smith (2012) doubtlessly to mean a disheveled, Einstein-like coiffure.In some cases, students added a globe to make the scientist an "Earth scientist" (e.g., Figure 4).The statistically significant changes in the negative direction with three indicators "indoors, " "lab coat, " and "glasses/goggles" reflect a move away from these stereotypical conceptions, which is favorable.However, we note that it is important for Earth science instructors to expose students to a broad spectrum of interesting Earth science topics.Not all students enjoy being outdoors and some enjoy laboratory work.Accurately representing the range of topics and work undertaken by diverse Earth scientists ensures that we provide an accessible and equitable curriculum that targets the interests of as many students as possible while realistically portraying the field.

Objects of study
As an object of study, many drawings depicted Earth scientists investigating rocks, minerals, or soil (e.g., Figure 3).This could result from a conceptual context, with students having ideas of Earth scientists studying "earthy things," or the context could be situational, especially if an instructor displays rocks and minerals around the classroom.Although there was not a significant shift in either direction between pre-and post-tests with the rocks/minerals/soil indicator, significant positive shifts in depictions of other Earth features were evident.Depictions of bodies of water along with mountains and volcanoes increased; with these, the data suggests that the course of instruction had a small effect.Drawings of celestial bodies, such as the Sun, Moon, and other planets, also increased (medium effect).We propose that these increases resulted from the situational context, specifically in the Earth Space Science course that includes lessons on Sun and Moon patterns.The most considerable increase, with a large effect, was illustrations of clouds, weather, and atmospheric phenomenon (Figure 5).Specific courses likely fostered this through a situational context because the course taught by Instructor 3 explicitly focuses on the atmosphere and Instructor 1 includes a weather unit in her courses.Across all these increases, however, our data suggest that introductory Earth science courses can expand student conceptions about what constitutes Earth systems and phenomena.

Hand lenses
A result that we found interesting was the representation of hand lenses (or magnifying glasses; Figures 3 and 6).
Students represented hand lenses in enough numbers (n = 34 overall; no significant change between pre-and post-drawings) that it caught our attention, especially with how they depicted Earth scientists using hand lenses.
Several drawings included an Earth scientist using a hand  lens to examine large-scale objects (like a landscape or the entire Earth) or an Earth scientist just holding a hand lens without using it.Our interpretation is that students associate Earth scientists with hand lenses due to a cultural context but without a complete understanding of what an Earth scientist might do with a hand lens.We note that hand lenses (or magnifying glasses) have been used in the kind of rock and mineral identification labs commonly practiced in K-12 classrooms and Earth science courses for preservice teachers (Dove, 1996;Ebert & Elliott, 2002;Westerback & Azer, 1991) This NGSS performance expectation is associated with the Scientific and Engineering Practice Developing and Using Models and with the Crosscutting Concept Stability and Change.We strive to develop both aspects of science with undergraduate students to move beyond rote learning and facilitate deeper thinking.Yet our data suggests that in some cases, the conceptions that students develop through these experiences may remain rather superficial (e.g., Earth scientists use hand lenses).

Mythic and other elements
The literature notes a few other confusing elements that appear in students' drawings of scientists with regularity.
One is what Finson (2002) calls a "mythic" element (p.341), like Frankenstein or a mad scientist, and we noted a few examples of this (Figure 3).Finson (2002) also supposes that sometimes students just draw what they think is silly and having found numerous lighthearted drawings, we certainly agree.Like the hand lens example, there is also confusion over exactly what different kinds of scientists do.With a draw-an-engineer test, Kuvac and Koc (2022) found that some students linked engineers with physical labor and operating machines.Yang et al. (2021) found that a draw-an-environmental scientist test elicited ideas about environmental protection and Rule et al. (2007) found that a draw-a-clay scientist test (clay minerals) yielded images of pottery and potter's wheels.A curious and recurring aspect in the drawings we analyzed involved Earth scientists studying Earth from outer space, sometimes as an astronaut, but often just floating, occasionally with a telescope (Figure 6).

Gender and race
We also coded all the drawings for the gender of the Earth scientist.Following a suggestion made during the discussion period of our conference presentation of initial results, we coded the drawings for race as well.To elicit conceptions about race and ethnicity, we provided skin-tone colored pencils to students at University 1 during the Fall of 2021 (the cost and logistics of distributing skin-tone colored pencils limited our ability to do so outside of the home university).Each of these students completed pre-and post-drawings (52 drawings total).We found that the skin-tone colored pencils were largely unused, and only four drawings (produced by two students, e.g., Figure 1) exhibit obvious efforts to represent Earth scientists as nonwhite.
Other than difficulties encountered when trying to determine the gender and race of drawings that were not much more than stick figures, we also were cognizant that our own biases could greatly influence our coding of these attributes.We conscientiously worked against using norms to  characterize race and gender (for example, assuming a scientist with a ponytail is female).This reduced our ability to characterize these attributes to cases that were evident in the writing below the picture (where some students explicitly used "he, " "she" or "they" to represent Earth scientists).Because of these challenges, we do not believe our data on gender and race are valid, and we are not reporting on these data here.
Although it is important data to collect, our experience indicates that drawings are not an effective way to collect data on student conceptions of the race and gender of Earth scientists.Rule et al. (2007) claim success with eliciting greater representation of race, citing drawings of 23 nonwhite scientists from a sample of 87 preservice teachers; however, they do not explain their methods for doing so (e.g., skin-tone colored pencils) and report that a social justice course that the students took concurrently may have influenced the results.Finson (2002) notes that the "draw a scientist" literature demonstrates that most minority students draw images of Caucasian scientists while contending that students may hold perceptions of scientists different from what they draw, thus "one needs to view student drawings with the proverbial grain of salt" (p.341).When looking at longitudinal trends, however, there is evidence that female scientists are appearing with greater frequency.Miller et al. (2018) conducted a meta-analysis of drawing studies that spanned five decades and found that students depicted female scientists more often in later decades.If we make a subjective evaluation, we would say our results align with this finding, and many students in our study drew Earth scientists that looked like themselves.This is encouraging and follows recent progress in the overall geoscience cultural context.

Limitations
The scope of this study is a factor limiting the transferability of the results.Due to the sample size, the results may not be fully transferable to contexts outside of the universities and classes from which we drew our participants.In particular, a large portion of our sample consisted of education majors and was majority female, which could influence the results.By using a theoretical construct of contextualization, we explicitly recognize that different contexts can drive how students represent their conceptions of Earth scientists; therefore, within different contexts, the results will likely be different.As researchers, our own experiences also contextualized the analysis.We attempted to reduce individual biases and describe our methods as transparently as possible to recognize this; nevertheless, different researchers with different biases and in different contexts may view these drawings and see something different.

Conclusions and implications
In this study, we found that student drawings include Earth scientists making observations, collecting data, and analyzing data.As a representation of student conceptions, this suggests that students overall have realistic conceptions of Earth scientists, although notable exceptions exist.Some of these exceptions, such as drawings of stereotypical scientists holding a globe, indicate that some students arrive with little to no idea of what Earth science is when they enter our courses.This is certainly understandable for non-STEM majors who are fulfilling a requirement and haven't taken an Earth science course since middle school.However, for instructors of introductory Earth science courses, it also prompts us to consider where we start on the first day of class; activities designed to familiarize students with the subject could be beneficial, especially in making connections and increasing engagement.Involving students in data collection and analysis throughout an Earth science course is a way to create a situational context that recognizes the importance of data to the discipline.We believe that many instructors do this, and our study provides some evidence that this promotes accurate student conceptions of Earth scientists.We propose that this may also improve public advocacy for mitigating human-induced climate change as it supports an understanding of how scientists base claims on observations and evidence.
Exposing students to a variety of Earth scientists and showcasing diverse people working in a variety of settings can continue to change the cultural context.This study suggests that introductory Earth science courses can move student conceptions of Earth scientists away from stereotypes and expand their conceptions to include outdoor fieldwork and additional "geospheres" (atmosphere and hydrosphere).Research prior to ours (e.g., Hong & Lin-Siegler, 2012;Miller et al., 2018;Rule et al., 2007), clearly demonstrates that education experiences can positively impact the conceptions that students hold of scientists, whether the focus of educators' efforts is gender or racial equity or simply making scientists look more like "regular people" (Finson, 2002, p. 343).Helping students see people like themselves in a variety of Earth science scenarios (e.g., at weather stations, aboard ships, at museums, in classrooms, at drill sites, and in data labs) may ignite interest in different aspects of the discipline and may increase the workforce in geoscience-related careers.Moreover, demonstrating that the discipline is not necessarily "outdoors intense" and is accessible to people with disabilities promotes awareness of equity.The most successful strategies for doing so appear to include the use of role models, activities, and targeted career exploration (Finson, 2002;Sherman-Morris et al., 2019;Sheffield et al., 2021).
Through the Science and Engineering Practices, the NGSS provide opportunities for involving students in the practices of scientists.As Earth science educators build upon existing pedagogies to increase student engagement with data, promote systems thinking, and facilitate involvement in research, students will experience Earth science as Earth scientists themselves.This molding of the situational context in classrooms to include authentic experiences will help create accurate conceptions of Earth scientists, which is especially important for preservice teachers who may pass these conceptions on to future students.In this study, we did not collect data on student experiences in introductory Earth science courses, still, our findings suggest that through course completion, student conceptions evolved to include more realistic representations.
The geoscience education community is interested in ensuring that access to Earth science education is just and equitable and that our professional community is diverse and inclusive.Initiatives such as Unlearning Racism in the Geosciences (URGE), Geosciences Associated Societies Committed to Embracing and Normalizing Diversity Research Coordination Network (Geosciences ASCEND), and the National Association of Geoscience Teachers' (NAGT) Diversity, Equity, and Inclusion (DEI) Committee and Statement on Racial Injustice are testament to this collaborative endeavor.Our methods for eliciting student conceptions of race and gender were ineffective (we were hesitant to "plant the seed" about race and gender and hoped that these features would appear organically).We suggest that alternative techniques, such as student interviews, might be a better way to investigate these conceptions.Interviews alone would not yield the same unique data that drawings capture; perhaps conducting both is a way to prompt these ideas.Recent work that explicitly introduces diverse geoscientists in the classroom setting has had positive results; for example, Sheffield et al. (2021) used a classroom intervention called "Scientist of the Week, " which involved projecting a photo of a diverse geoscientist along with information about their field of study and contributions.We welcome feedback and encourage conversation on authentic methods for investigating student conceptions of race and gender in the Earth scientist community and beyond.
Overall, we found drawing to be an effective method for investigating student conceptions and how they change.The pre-drawings revealed multiple conceptions that students bring to our classes, several of them accurate, but many limited in scope, some stereotypical, a few mythical, and often unsophisticated.By asking students to repeat the process at the end of the semester, we successfully captured how these conceptions changed.We propose that using drawing to investigate student conceptions of Earth scientists is an additional valuable tool in an instructor's toolbox for formative and summative assessment that can provide important feedback for instruction and curriculum design.
expanded this to 15 indicators and an open-ended question and claimed this made identifying and recording indicators more efficient and readily quantifiable for data analysis.

Figure 1 .
Figure 1.Phases leading to the development of the draw an earth scientist test and scoring checklist.
, open to non-science majors at University 2, was Dynamic Earth.The fifth course was an introductory physical geology course at University 2 intended for geology majors, civil and construction engineering majors, and other interested science majors.Collectively, six different instructors taught these courses.Table The first course was an introductory Earth science course at University 1, offered to elementary education majors and titled Earth Space Science.The second and third courses were two courses at University 2 offered to elementary education majors, titled Earth Systems: Issues and Applications, and Exploring Earth Science: The Atmosphere.The title of the fourth course

Table 1 .
study setting: distribution of courses, instructors, and participants.

Table 3 .
indicators with over 30 occurrences in participant drawings (in order of frequency).

Table 4 .
direction of statistically significant change (and cohen's W).
Develop a model to describe the cycling of Earth's materials and the flow of energy that drives this process.[Clarification Statement: Emphasis is on the processes of melting, crystallization, weathering, deformation, and sedimentation, which act together to form minerals and rocks through the cycling of Earth's materials.][Assessment Boundary: Assessment does not include the identification and naming of minerals.] . However, the Next Generation Science Standards (NGSS) have explicitly left rock and mineral identification out of the Performance Expectations for Earth and Space Sciences, focusing instead on processes inherent in Earth systems (NGSS Lead States, 2013).A related Performance Expectation for middle school students (MS-ESS2-1) reads: