The Impact of Inquiry Instruction and a Computer Simulation on Secon= dary Earth Science Pupils’ Understandings of Moon Phases

Using Technology to Pro= mote Conceptual Change in Secondary Earth Science Pupils’ Understandings of Moon Phases

Randy L. Bell= , University of Virginia

Ian C. Binns,= University of Virginia

Lara K. Smeta= na, University of Virginia

Abstract

This study assessed the u= se of a computer simulation and whole-class display system to promote scientific co= nceptions of shapes, sequence, and causes of moon phases for secondary earth science students. The instructional intervention was done by two preservice teachers and included 16 days of daily moon observations followed by in-class inquiry-based instruction on moon phases. Two treatments were compared: instruction where pupils collected moon data from natural observations (NO;= n=3D27) and instruction where pupils collected moon data from Starry Night observations (SN; n=3D22). Data resources used to characterize pupils’ conceptions of moon phases included forced-choice questionnaires, open-ended questionnaires, and interviews. The between-group comparisons of the forced-choice questionnaires showed no significant difference. The responses on the open-ended questionnaires and the intervie= ws were analyzed using the constant comparative method. Pre- to post-instruction ga= ins from the open-ended questionnaires and the interviews were substantial for both treatments and for all targeted moon concepts. The SN group demonstrated hi= gher gains for all targeted moon concepts. Results indicate that the use of a computer simulation can be effective in achieving conceptual change.

Introduc= tion

A growing = body of research suggests that educational technology can have a positive impact on science instruction and learning. For instance, studies report more student-centered and individualized teacher-student interactions during technology-enhanced instruction (Sandholtz, Ringstaff, & Dwyer, 1992; S= wan & Mitrani, 1993), higher levels of student engagement and time on task (MacArthur, Haynes, & Malouf, 1986; Worthen, Van Dusen, & Sailor, 1994), and increased student achievement (Bayraktar, 2002; Sivin & Bial= o, 1994). However, other research suggests that the technology does not offer = an advantage over traditional approaches. Findings suggest that educational technology may negatively impact student engagement and learning (Olson &am= p; Clough, 2001; Waight, 2004). However, a recent study by Marshall and Young (2006) reports that teachers do not take advantage of what the technology has to offer. For instance, computer simulations offer many features that, when used appropriately, have the potential to be effective instructional tools. Specifically, attention should be give to those programs designed to allow learners to compare their ideas to scientific conceptions.

Computer simulations present simplified models of real world systems and processes. = They present an alternative option for learning activities that may be omitted if deemed too time-intensive, costly or complicated. With simulations, teachers can potentially better focus students’ attention on learning objectiv= es, because with simulations the real-world environments are simplified, causal= ity of events are more clearly visualized, and unnecessary cognitive tasks are =

reduced (de Jong & Van Joolingen, 1998). By allowing students to observe, explore, recreate, and receive immediate feedback about real world components and phenomena, simulations have the potential to make learning abstract concepts more interactive, authentic, and meaningful (Cholmsky, 2003; Ramasundarm, 2005).

For over t= wo decades, researchers have explored the effectiveness of computer simulations for supporting science teaching and learning. This body of research indicat= es that simulations can be effective in developing content knowledge and proce= ss skills. For example, a recent meta-analysis of 42 studies found computer si= mulations to be the most effective type of computer-assisted instruction, reporting a mean effective size of 0.391 SD for simulations as compared to traditional instruction (Bayraktar, 2002). Geban, Askar, and Ozkan (1992) found greater student achievement in chemistry content knowledge and process skills with simulated discovery labs than with hands-on, teacher-prescribed labs. Akpan= and Andre (2000) investigated the effectiveness of computer simulations versus = and in conjunction with hands-on frog dissection. They found that students in t= he simulation only and simulation-before-dissection groups performed best on an anatomy test, having constructed an experiential base on which to build fur= ther knowledge.

Positive r= esults have also been reported for computer simulations used to promote more complicated goals such as inquiry and conceptual change. For example, Mintz (1993) reported positive finding for students using a computer simulation to design, implement and analyze the results of increasingly complex experimen= ts. Huppert, Lomask, and Lazarowitz (2002) investigated the impact of computer simulations on the development of higher-level inquiry skills. They found t= hat high school students using a simulated yeast cell lab outperformed those completing a hands-on lab. Zietsman and Hewson (1986) reported success in achieving conceptual change in physics using simulations that involved stud= ents experiencing a series of discrepant events that attacked the plausibility of commonly held alternative conceptions. Gorsky and Finegold (1992) also found computer simulations to be an effective mean= s of promoting cognitive dissonance as part of the conceptual change model. Other studies verified the challenge of achieving stable conceptual change. Tao a= nd Gunstone (1999) explored the effectiveness of using simulations to promote lasting conceptual change about force and motion topics. They determined th= at conceptual change is highly context-dependant and simulations may be effect= ive when students are able to reflect on and reconstruct their conceptio= ns. Overall, the literature suggests that computer simulations show promise for= learning about complex scientific phenomena and that future research should continue= to explore the most effective instructional uses. This study seeks to do so by investigating the effectiveness of a planetarium computer simulation in developing student understanding of moon phases.

Recent ref= orm documents include standards that explicitly address earth-moon-sun relationships. The National Science Education Standards state that middle grade earth science classes should build upon a conceptual understanding constructed in the elementary grades = for understanding the relationships among the earth, moon, sun, and solar syste= m; pupils should be able to use this model to explain moon phases (National Research Council, 1996). The = Benchmarks for Science Literacy acknowledges the difficulty that students have with these abstract ideas (American Association for the Advancement of Science, 1993). Indeed, the literature shows that despite the emphasis across grade levels, students struggle in developing scientific conceptions of moon phas= es. Consistent with conceptual change theory, research has shown that students’ alternative conceptions about lunar phenomena are tenacious= (Schoon, 1992; Stahly, 1999; Treagust, 1988).

Researcher= s have found traditional teacher-centered and textbook-based instruction to be ine= ffective in addressing these prevalent alternative conceptions, citing ambiguous textbook images and terminology as impediments to developing scientific conceptions of moon phases (Dai, 1991; Dove, 2002). Instead, the reform documents and recent empirical studies encourage constructivist methods, pupil-centered activities and experiential opportunities designed to help learners visualize the earth-moon-sun relationships (Jones, Lynch, & Reesink, 1987; Trundle, Atwood, & Christopher, 2002, 2004, 2006). Specifically, a literature review = by Kavanagh, Agan and Sneider (2005= ) recommends that instruction about moon phases involve making observations of the sky a= nd modeling the phases of the moon and the Earth-Moon-Sun system. However, the= se activities are time-intensive and can be frustrating for pupils since data collection is dependent on factors that lie outside of pupils’ contro= l, such as weather, availability to make observations at particular times and locations, etc.

Computer s= imulations such as the planetarium program Sta= rry Night (used in this study) have the potential to alleviate these complications. Using a computer simulation like Starry Night provides advantages such as the ability to collect complete and more detailed data sets in a shorter period of time, the ease = of making precise observations from any location on earth and from any vantage point, and the elimination of challenges posed by weather conditions, night= time viewings, and landscape obstructions (Trundle & Bell, 2003). A series of empirical studies have demonstrated that Starry Night can provide an effective context for preservice teachers to devel= op scientific conceptions of moon phases (Bell & Trundle, 2006; Trundle &a= mp; Bell, 2005). However, it is not known how preservice teachers who have deve= loped scientific conceptions of moon phases through the use of computer simulatio= ns use such simulations in their own instruction, or what their pupils learn f= rom this instruction.

The purpos= e of this investigation is two-fold—first, to characterize secondary earth science pupils’ conceptions of moon shapes, changes in the moon’= ;s shape, and the cause of moon phases before and after instruction; secondly,= to assess the effectiveness of using a computer simulation versus natural observations in regard to pupil learning. The following research questions guided this investigation:

What are secondary earth science pupils’ conceptions of moon phases before and after completing an inquiry-based unit on lunar concepts?
What is the impact on secondary earth science pupils’ conceptions of moon phases using the Starry Night simulation versus natural observations of the moon?

Methods<= o:p>

Participants and Context

The two te= acher participants in the study were preservice secondary earth science students enrolled in a teacher education program at a major mid-Atlantic university. Dean was completing the requirements for his Bachelors of Science degree in Environmental Science and Masters of Teaching degree concurrently. Rob had = previously earned a Bachelors of Arts degree in Earth Science at a different universit= y, and was enrolled in the post-graduate Masters of Teaching program. Both participants were ultimately seeking licensure to teach Earth Science. Thei= r science backgrounds included courses in environmental science, geology, meteorology, and oceanography. Both had completed one undergraduate level astronomy cour= se. The study took place in the fall semester of their fifth year of the progra= m during their normal student teaching experience.

For his student teaching experience, Dean taught five ninth-grade earth science classes at a rural high school. The school served approximately 800 students with 13% minority students and 18% free and reduced lunch students. Rob tau= ght five ninth-grade earth science classes at a suburban high school. The school served approximately 1000 students with 7% minority students and 8% free and reduced lunch students.

For each s= tudent teacher, two of the highest academic level classes were selected to partici= pate in the study. Given the two participants’ teaching schedules, this was the only choice possible to ensure that the resulting four classes were comparable in regard to overall academic achievement. The pupil participants included 49 volunteers from the four secondary earth science classes (out o= f a total of 87 pupils, for a participation rate of 56%). Each of the four clas= ses was randomly assigned to one of two treatments: pupils in the natural observation (NO) treatment (n =3D 27) collected moon data from natural observations and pupils in the Star= ry Night (SN) treatment (n =3D 22) collected moon data from the Starry Night computer simulation. = Otherwise, the moon phase instruction for the two groups was equivalent.

Both instr= uctional treatments for the Moon Watch Project began with a 16-day observational per= iod. All pupils were required to use an observation log (Figure 1) to record dra= wings of the shape of the moon, the date and time of the observation, and the direction they looked to see the moon. Pupils in the natural observation gr= oup were expected to make daily observations outside of class and then share th= eir observations in brief discussions at the beginning of each class session. T= hese discussions averaged 15 minutes and addressed the required aspects of each observation, patterns pupils had noticed, and predictions for their future observations. The teachers made a point to only facilitate the discussions, saving explicit instruction about moon phases for after the observation per= iod.

Figure 1. Representative moon log for natural observation group.

Pupils in = the Starry Night group used the same observation log and recorded the same information as the pupils in the natu= ral observation group, but collected this information from the computer simulat= ion. Each class began with a 15-minute discussion during which they made observa= tions as a group using Starry Night, = which the teacher projected from the computer onto a screen for the class to see.= The teachers led the class in finding the moon and then asked pupils to draw its shape and record the date, time, and viewing direction. As with the natural observations treatment, pupils were encouraged to develop and share patterns and to make predictions for future observations. Again, the teachers served only as facilitators during the 16 days of data collection.

At the con= clusion of the 16-day observational period, pupils in both groups participated in t= wo lessons designed to summarize and make meaning of the data they had collected. Pupi= ls began the first class discussing shapes, sequences, and causes of moon phases, in= the context of their observation logs. For both treatments, pupils volunteered = to draw their phases and sequence on the board. The class reviewed these drawi= ngs and corrected each other. Pupils also shared their ideas about the cause of moon phases. For this lesson, the teacher facilitated the discussion, but d= id not correct pupils’ responses, although the pupils did correct each other.

A teacher-= led demonstration followed the discussion. A light source set up on one side of= the room represented the sun, the pupils sat in the middle of the room to repre= sent the earth, and the teacher walked around the pupils holding a lunar globe to show the phases. Pupils drew the shapes they observed and noted the sequenc= e of the moon phases. After the demonstration, the class discussed the drawings = and sequences, comparing the shapes drawn during the demonstration to those dra= wn on the board at the start of the period. This comparison was repeated for t= he sequencing. The teacher resisted providing the correct shapes and sequence; instead, the class arrived at a general consensus.

The follow= ing class period began with a review of the previous lesson. The teacher quickl= y recapped the conclusions the pupils reached in regard to the shapes and sequence. Ne= xt, the teacher listed on the board the pupils’ ideas about the cause of = the moon phases from the previous lesson. The teacher challenged pupils to figu= re out which, if any, of these ideas fit their data and the model of the sun-earth-moon system. Pupils then participated in a hands-on modeling acti= vity using an exposed light bulb to represent the sun, a Styrofoam ball held at arm’s length to represent the moon, and their own heads to represent = the earth. Throughout the activity, the teacher and the pupils discussed the relationship between their observations and this hands-on model with the go= al of further delineating the shapes, sequence, and cause of moon phases. Fina= lly, the teacher led a discussion in which pupils selected the best explanation = for the cause of moon phases from the list of ideas on the board. In all four classes, the pupils reached consensus on scientific conceptions of moon pha= ses, including shapes, sequences, and cause.

The resear= chers collected all lesson plans and one of the researchers observed and recorded detailed notes for all lessons during the instructional treatment period. T= hese data were compared for each treatment to ensure that both teachers incorpor= ated similar instructional strategies and devoted equal time to each of the two instructional approaches.

Data Collection

The resear= chers used two approaches to characterize pupils’ conceptions of moon phase= s. First, participants completed pre- and post-instruction administrations of a modif= ied version of the Lunar Phases Concept Inventory (LPCI). The LPCI is a validat= ed 21-item instrument designed to assess conceptual understandings of lunar phases (Lindell & Olson, 2002). The modified LPCI used in this study was the result of a content analysis of the original LPCI conducted by the particip= ating student teachers, their mentor teachers, and an expert on moon phases. Their analysis indicated that five items addressed content that the student teach= ers did not plan to cover during the Moon Watch Project. The researchers removed these questions resulting in a total of 16 forced-choice items (Appendix A)= .

In additio= n to the LPCI, a subset of pupils (18 from the natural observations group and 14 from the Starry Night group) agreed = to complete a modified version of the protocol developed by Trundle et al. (2002, 2004, 2006) to assess moon phase conceptions through written responses to open-en= ded questions, a card sort, and responses to task-based interviews. Typically, = the protocol involves both pre- and post-assessments. Due to time constraints, = the mentor teachers did not allow a pre-instruction administration of the full = protocol. Instead, the researchers developed a questionnaire that consisted of the open-ended items from the Trundle et al. (2002, 2004, 2006) protocol, plus = an additional item asking pupils to explain the cause of moon phases in their = own words (Appendix B). Pupils completed this questionnaire immediately prior to the Moon Watch Project.

At the con= clusion of the Moon Watch Project, the subset of 32 pupils completed the post-instruction open-ended questionnaire, a sequencing card sort activity,= and interview (Appendix C). During the interviews, pupils were challenged to use three-dimensional models of the sun, moon, and earth to demonstrate and exp= lain their conceptions of the causes of moon phases. Additionally, pupils were a= sked to describe if and how their understanding of moon phases had changed over = the course of the Moon Watch Project, as well as what caused these changes.

Data Analysis=

A split-pl= ot ANOVA was used to assess differences in pre- and post-instruction mean LPCI scores for each treatment as well as the differences between mean scores of the two instructional approaches (natural observations versus observations with Starry Night). Classroom observati= ons and lesson plan analysis indicated that the teachers implemented the two instructional treatments consistently. Furthermore, a t-test on the pre-instruction LPCI means showed no differences = for the two teachers’ pupils within each treatment (NO p =3D .604; SN p = =3D .618). Therefore, pupils in Dean and Rob’s natural observation classes were combined, as were pupils in Dean and Rob’s Starry Night classes for subsequent analyses (n =3D 27 for natural observation treatment, and n =3D 22 for Starry = Night treatment).

The resear= chers used the constant comparative method to analyze the data recorded on coding sheets from each pupil’s post-instruction open-ended questionnaire and interview (Strauss & Corbin, 1994). A coding sheet and coding system ba= sed on previous research (Trundle, 2003; Trundle et al., 2002, 2004, 2006) were used to analyze and summarize the pupils’ moon phase drawings and interview transcripts. Analysis involved comparing pupil views as captured = on the coding sheets to accepted scientific views of shapes, sequences, and ca= uses of moon phases. Finally, achievement gains for the Starry Night and natural observations groups were compared to assess the relative effectiveness of the two instructional approaches.

= ;

Findings

Modified LPCI=

Prior to instruction, the mean scores on the LPCI were 7.04 for the natural observations group and 7.68 for the Starry Night group. After instruction, the means scores on the LPCI were 10.70= for the natural observations group and 10.82 for the Starry Night group. A split-plot factorial ANOVA showed that th= ere was a significant difference between the pre-instruction means and post-ins= truction means for both the natural observations and Starry Night groups (F_LPCI = (1,47) =3D 57.003, p < .001). The analysis also showed that there was not a significant difference between the two treatments (F_treatment (1,47) =3D 0.314, p =3D .578).

Table 1

Analysis of Va= riance for Treatments

Source	df	F	p
Between subjects<= /p>
Treatment	1	.314	.578
Error	47	(5.567)	<= /p>
Within subjects
LPCI	1	57.003*	< .001
Treatment x LPCI	1	.346	.559
LPCI x Error	47	(4.921)	<= /p>

Note. Values in parentheses repres= ent mean square errors.

*p < .001

Open-ended Questionnaire and Interviews<= /u>

A subset o= f pupils (18 from the natural observation group and 14 from the Starry Night group) participated in post-instruction interviews= . In these interviews, the researchers first asked participants to explain their responses to the post-questionnaire. Pupils described their responses to the pre- and post-instruction open-ended questionnaires, which included their d= rawings of the moon phases and sequences and written descriptions of the cause of m= oon phases. Next, pupils used three-dimensional models to explain what they tho= ught caused moon phases. Then, each participant completed a sequencing card sort activity. The interviews concluded with pupils explaining how their underst= anding of moon phases had changed over the course of the Moon Watch Project and wh= at aspect(s) of the instruction influenced these changes.

Shapes

Pre-Ins= truction

The first = question of the pre-instruction open-ended questionnaire asked pupils to predict the= shapes of the moon phases they would see during their moon observations. To be classified as scientific, all drawings had to reflect accurate scientific shapes, but did not have to include all of the observable phases. The major= ity of pupils in both instructional treatments included nonscientific shapes in their drawings (Table 2). Seventy-eight percent of the pupils in the natural observations group and 57% of the pupils in the Starry Night group included at least one non-scientific shape a= mong their pre-instruction drawings. All of these pupils included either non-scientific crescent moons or non-scientific gibbous moons. Non-scientif= ic crescent phases were drawn with either over- or under-articulated terminato= rs (line separating lit and dark parts of the moon). Non-scientific gibbous mo= ons were drawn as a partial lunar eclipse. Figure 2 shows a representative pre-instruction drawing of the moon phase shapes. Note that Tim includes an under-articulated terminator for two crescent moon drawings (#5 and #7) and= a partial eclipse representation for the gibbous moon (#2).

Table 2

Interviewed Pupils’ Responses Coded as Scientific

Targeted moon phase conceptions	Pretest	Posttest<= /span>	Gain (Posttest-Pretest)
Scientific moon phase drawings
Natural Observations (n =3D 18)	22%=	72%=	50%=
Starry Night (n =3D 14)	43%=	100= %	57%=
Scientific waning and waxing sequence drawings
Natural Observations (n =3D 18)	11%=	50%=	39%=
Starry Night (n =3D 14)	36%=	86%=	50%=
Both scientific phases and sequence drawings
Natural Observations (n =3D 18)	0%<= o:p>	33%=	33%=
Starry Night (n =3D 14)	21%=	86%=	65%=
Scientific cause of Moon Phases
Natural Observations (n =3D 18)	0%<= o:p>	44%=	44%=
Starry Night (n =3D 14)	0%<= o:p>	71%=	71%=

Figure 2. = Tim’s pre-instruction drawings of moon phases showing alternative shapes for the crescent and gibbous phases.

Post-Instruction

The first = question of the post-instruction open-ended questionnaire asked participants to draw= the phases that they observed and then add the phases they predict they would h= ave seen had they made observations for a full cycle.

The post-instruction results reflected substantial improvements in both groups’ abilities to draw scientific moon shapes. Specifically, 72% of the pupils in the natural observations group and 100% of the pupils in the = Starry Night group included scient= ific shapes. Thus, the gains in ability to draw scientific shapes favored the Starry Night group over the natural observations group (57% vs. 50%). Figure 3 shows Tim’s post-instructi= on drawings of the moon, which included all scientific shapes.

Figure 3. = Tim’s post-instruction drawings of moon phases showing a dramatic improvement over the pre-instruction drawings.

Sequence

Pre-Ins= truction

The second= and third questions from the pre-instruction open-ended questionnaire addressed= the sequence of moon phases. The second question asked pupils whether they thou= ght the moon phases would appear in a predictable pattern or sequence. The third question asked pupils who indicated that the moon phases would appear in a = predictable pattern or sequence to draw the moon phases in the sequence they expected to observe. To be classified as scientific, pupils had to agree that the moon phases would appear in a predictable pattern or sequence and had to include both scientific waxing and waning sequences in their drawings. Pre-instruct= ion results from the second question revealed that all of the pupils in the nat= ural observations group and all but one pupil in the Starry Night group indicated that they expected the moon phases= to appear in a predictable pattern. In response to the third question, only 11% of the pupils in the natural observations group and 36% of the pupils in the Starry Night group drew scientific sequences (Table 2). Figure 4 shows a representative pre-instruction drawin= g of the moon phase sequence. Note that Zach incorrectly drew waning moon shapes= in a waxing sequence and waxing moon shapes in a waning sequence. This was typ= ical of the majority of participants who drew non-scientific sequences.

= ;

Figure 4. = Zach’s pre-instruction drawings of moon sequence showing alternative shapes and sequences.

Post-Instruction

The post-instruction results reflect improvements in both groups’ abiliti= es to draw scientific sequences. Consistent with the pre-test results, all par= ticipants in the natural observation group expected the moon phases to appear in a predictable pattern. Additionally, all participants in the Starry Night group also expected the moon phases to appear in a predictable pattern. Moreover, 50% of the pupils in the natural observations group and 86% of the pupils in the = Starry Night group drew both scientific waxing and waning sequences. As with t= he shapes, the gains for sequence favored the Starry Night group over the natural observations group (50% vs. 39%). Figure 5 shows Zach’s post-instruction drawing of the moon sequence. Not only = did Zach provide the correct sequence, he also drew all scientific shapes and correctly named each phase.

Figure 5. Zach’s post-instruction drawings of moon sequence.

Summary of Shape and Seq= uence Results

Prior to instruction, none of the pupils in the natural observation group and 21%= of the pupils in the Starry Night = group were able to draw both scientific phases and sequences prior to instruction= . After instruction, 33% of the pupils in the natural observation group and 86% of = the pupils in the Starry Night grou= p could draw both scientific phases and sequences (Table 2). When both shape and sequence are considered, the gains heavily favored the Starry Night group over the natural observations group (65% vs. 39%).

Causes

Pre-Ins= truction

The fourth question from the pre-instruction open-ended questionnaire addressed the ca= uses of moon phases. Pupils were asked to draw a diagram and describe in their o= wn words what they thought causes moon phases. Pupils provided a variety of responses to this question (Table 3). None of the pupils in either group met the four criteria required for a complete scientific understanding of the c= ause of moon phases. Two pupils in both groups met the criteria for scientific fragments. One pupil in the Starry = Night group met the criteria for scientific fragments with alternative whereas no= ne of the pupils in the natural observations group met these criteria. Overall= , half of the pupils in the natural observations group and 58% of the pupils in th= e Starry Night group responded with = alternative conceptions or alternative conception fragments about the cause of moon pha= ses.

Table 3 <= /o:p>

Frequencies of= Pupil Pre-instruction Conceptions of the Causes of Moon Phases

Conceptual Understandings	Criteria<= /p>	Natural Observations Group (n =3D 18= )	Starry Night Group (n =3D 14= )
Scientific	Included each of the four conceptions: · Half of the moon is illuminated by the sun= [SciHaf] · The portion of the illuminated half seen f= rom earth varies over time [SciSee] · The relative positions of the earth, sun, = and moon determine the portion of the lighted half seen from earth [SciEMS] · The moon orbits earth [SciOrb]	0%	0%
Scientific Fragments	Included a subset but not all four of the scientif= ic criteria	11%	14%
Scientific fragments with Alternative	Included a subset of the scientific criteria and o= ne of the alternative fragments listed below	0%	7%
Alternative Eclipse	The earth’s shadow causes the moon phases	11%	30%
Alternative Earth’s Rotation	The earth’s rotation on its axis causes the = moon phases	16%	7%
Alternative Clouds	Cloud cover causes moon phases	0%	7%
Alternative Heliocentric	Position of the moon around the sun.	0%	7%
Alternative Sun	Sun’s shadow on moon causes moon phases	6%	0%
Alternative Tides	Earth’s tides cause moon phases	6%	0%
Alternative Fragments	Included a subset or subsets of alternative concep= tual understandings	11%	7%
Incomplete Response	Does not give a complete response or gives no response.	39%	21%

Prior to instruction, there were pupils from both groups who did not provide enou= gh information to classify their responses as scientific or alternative. Thirt= y-nine percent of the pupils in the natural observations group and 21% of the pupi= ls in the Starry Night group eithe= r gave an incomplete response or no response at all. When asked during the post-instruction interview why they did not respond to this question, the majority indicated that they did not know how to begin to explain the cause= of moon phases.

Overall, t= he most common alternative conception prior to instruction was the alternative ecli= pse (11% natural observation, 30% Starry Night). Figure 6 shows Jennifer’s pre-instruction response to the cause of moon phases. She correctly states that the moon reflects the sun’s light, but then goes on to explain that the phases are caused by the earth blocking the light from the sun.

Figure 6. Jennifer’s pre-instruction explanation and drawing for the cau= se of moon phase.

Post-in= struction

The post-instruction results reflect improvements in both groups’ abiliti= es to explain the cause of moon phases. Specifically, 44% of the pupils in the natural observations group and 71% of the pupils in the Starry Night group described all four components of a complete scientific conception of the cause of moon phases (Table 4). The remaining pupils for both groups (56% for the natural observations group and 29% for = the Starry Night group) expressed a pa= rtial scientific conception, but held on to at least one alternative fragment. No pupils in either group expressed entirely alternative conceptions of the ca= use of moon phases or gave incomplete responses in the post-instruction assessm= ent.

Table 4 <= /o:p>

Frequencies of= Types of Conceptual Understanding for Interviewed Pupils

Type of Conceptual Understanding	Natural O= bs. group (n =3D 18= )	Starry Night group (n =3D 14= )
Before In= struction	After Ins= truction	Before In= struction	After Ins= truction
Scientific	0%	44%	0%	71%
Scientific fragments	11%	0%	14%	0%
Scientific fragments with alternative	0%	56%	7%	29%
Alternative	50%	0%	58%	0%
Incomplete Response	39%	0%	21%	0%

Jennifer’s conception of the cause of moon phases improved dramatically as indicated b= y her post-instruction responses to the open-ended questionnaire and interview. H= er pre-instruction explanation of the cause of moon phases indicated that she believed the Earth’s shadow causes moon phases. Her post-instruction explanation on the open-ended questionnaire included two of the four criter= ia for a scientific conception of the cause of moon phases (SciOrb and SciEMS). The following quote is from Jennifer’s response on the post-instruction open-ended questionnaire, which was accompanied by the drawing in Figure 7.=

“The moon revolves around= the earth [SciOrb]. When it’s between the earth and sun, we can’t see the light, so it’s a new moon, but as the moon revolves back to the opposite side, we can see more a= nd more of it until it’s a full moon again [SciEMS]. Then we see less and less of it until we cannot see any (the new moon) and the cycle continues.”

&nbs= p; &= nbsp; &nbs= p; Jenni= fer, Post-Instruction Open-Ended Questionnaire

Figure 7. Jennifer’s post-instruction drawing indicating her understanding of SciOrb and SciEMS.

While Jennifer expressed on= ly two of the criteria for a scientific understanding in her response to the open-ended questionnaire, she clearly expressed all four criteria during th= e subsequent interview. A portion of the transcript from her interview is provided below= .

Researcher: I have models of all the things you were talking about [sun, earth, and moon]. Please use these models to show me and explain to me what causes moon phases.

Jennifer: [Arranges the model components so that the sun and earth are in a straight line, and holds the moon component in her right hand in between the earth and the sun. The models are arranged at approximately 0 degrees at a = new moon position.] If the sun, earth, and moon were arranged like this, we would have a new moon because the sun is lighting up the half of the moon [= SciHaf] that is facing it [Points at the half of the moon facing = the sun]. We can only see the half of the moon that is not facing the sun, which is not lit up by the sun [Sc= iSee], so that is why we have a new moon. As the moon orbits the earth [SciOrb], we see more of the lit s= ide as the moon goes from new to full moon and we see less of the lit side as the = moon returns to a new moon.

Researcher: I have some cards to show you. Anything that is colored orange is what we see. This is a full mo= on because the entire circle is colored orange. I would like for you to arrange the model so that we would have a full moon.

Jennifer: [Moves the model so that the earth component is between the moon and= sun in a straight line and holds the moon slightly above the plane created by t= he sun and earth to show that the earth is not blocking the light.] This is the full moon because the earth is between the sun and moon [SciEMS]. The sun lights up the ha= lf of the moon that is facing the earth. So half of the moon is still lit [SciHaf] and we see that complete = half [SciSee].

Jennifer, Post-Instruction Int= erview

Conclusi= ons and Discussion

Prior to instruction, a majority of the pupils in both the natural observations and = Starry Night group could not corre= ctly draw the shapes and sequences of the moon phases. Their conceptions of observable moon shapes were either incomplete, included nonscientific shapes such as partial lunar eclipses, or both. All pupils in the natural observat= ions group and 93% of the pupils in the = Starry Night group understood that moon phases appear in a predictable pattern; however, the majority could not draw sequences consistent with scientific conceptions. Finally, none of the pupils in either group met the four crite= ria required for a complete scientific understanding of the cause of moon phase= s.

After instruction, both groups’ conceptions= improved for all targeted moon phase concepts. The gains in pupils’ abilities = to draw scientific moon shapes were similar for both groups, but slightly favo= red the Starry Night group, in whic= h