Design and Evaluation of a Braided Teaching Coursein Sixth Grade Computer Science Education

Arno PasternakFritz-Steinhoff-Gesamtschule Hagen and

Technische Universitat Dortmund44227 Dortmund, Germany

arno.pasternak@cs.tu-dortmund.de

Jan VahrenholdTechnische Universitat DortmundFaculty of Computer Science44227 Dortmund, Germany

jan.vahrenhold@cs.tu-dortmund.de

ABSTRACTWe report on the design and evaluation of the rst yearof a Computer Science course in lower secondary educationthat implements the concept of braided teaching [7]. Be-sides being a proof-of-concept, our study demonstrates thatstudents can indeed be taught Computer Science (as op-posed to Information and Communication Technology) asearly as in sixth grade while at the same time not falling be-hind with respect to Information Technology Literacy. Wepresent quantitative and qualitative results and argue thatComputer Science can be taught just like any other sciencesubject worth full curriculum credit.

Categories and Subject Descriptors: K.3.2 [Computersand Education]: Computer and Information Science Educa-tion

General Terms: Design, Experimentation.

Keywords: Lower Secondary Computer Science Educa-tion, Braided Teaching, Computer Science vs. ICT

1. INTRODUCTIONA recent study [10] conducted by the ACM and the CSTA

reveals that in contrast to the crucial role Computer Scienceand the technologies enabled by it plays in the 21st century(see also [11]), Computer Science as a subject of study stillplays a minor if non-existent role in K-12 education. Whilethe study presents comprehensive data only for the U.S.,reports from other countries (e.g. [1, 5, 12]) indicate thatthere are only few success stories in general.

One important conclusion that can be drawn from thisstudy is that policymakers are not technology-oblivious perse: 73% of the school districts in the U.S. have adopted acurriculum that aims at Information and CommunicationTechnology (ICT) skills [10, p. 44]. In contrast, only 37%of the districts have adopted a curriculum that goes beyondthis and aims at concepts in Computer Science [10, p. 37].

Permission to make digital or hard copies of all or part of this work forpersonal or classroom use is granted without fee provided that copies arenot made or distributed for profit or commercial advantage and that copiesbear this notice and the full citation on the first page. To copy otherwise, torepublish, to post on servers or to redistribute to lists, requires prior specificpermission and/or a fee.SIGCSE12, February 29March 3, 2012, Raleigh, North Carolina, USA.Copyright 2012 ACM 978-1-4503-1098-7/12/02 ...$10.00.

As part of their National Call to Action the authorsof the study request to develop courses to implement newcomputer science standards [10, p. 14] and present eortsfrom the states of Georgia and Texas. The authors empha-size that a rudimentary knowledge of computing is insuf-cient in the digital age; instead, they demand ComputerScience to be taught on a continuum from basic comput-ing concepts that can be attained at elementary and middleschool levels to deeper knowledge, skills, and practices moreappropriate for secondary school [10, p. 25].

An established way of teaching a continuum of conceptsis along what is known as Bruners spiral curriculum [2]. Inthe case of Computer Science in lower secondary education,we have proposed to follow this approach in what we termedbraided teaching [7]: a contextualized approach to coursedesign that organizes topics along strands which then aretaught in an interleaved way and along a spiral curriculum(see [7] and the references therein for a discussion of thisconcept and related work).

Definition 1. A strand is a sequence of items addressedin class that satisfies the following criteria: (1) The itemscan be assigned to a well-defined subject matter (by theirstructure or their content). (2) The subject matter is iden-tifiable and recognizable to the students throughout the se-quence. (3) The subject matter is being presented from morethan one point of view or embedded in more than one con-text. (4) The sequence of items is addressed in more thanone teaching unit.

Two questions left open are whether this concept can beinstantiated in practice and whether positive eects can beobserved. We report on the rst year of a prototypical im-plementation of the concept and, based upon qualitative andquantitive evidence, armatively answer both questions.

2. DESIGN AND IMPLEMENTATION OFA BRAIDED TEACHING COURSE

One of the major impediments in implementing and eval-uating Computer Science courses in lower secondary educa-tion is that in the authors state these courses are not worthfull curriculum credit and that thus the student populationin these courses is somewhat biased. Fortunately, the ad-ministration at the Fritz-Steinho-Gesamtschule allowed fora prototypical implementation of a Computer Science coursein sixth grade that is worth the same (i.e. full) curriculumcredit as Mathematics, Physics, Biology, and Chemistry. Ingeneral, students choose one elective course at the beginning

45

of sixth grade; this course is one of French, Italian, Natu-ral Sciences (a combined course on Physics, Biology, andChemistry), or Employment Studies. Each of these coursesis taught for 120 mins. per week and fully counts towards thenal exams, i.e. failing such a course has the same eect asfailing Mathematics or English. In support of our research,a Computer Science course was added to the list of electives.In 2010/2011, 27 students enrolled in this course.

2.1 Overview of the CurriculumThe course reported upon was designed according to the

principles of braided teaching we proposed in an earlier arti-cle [7] and taught by the rst author, an in-service teacherwho has taught Computer Science, Mathematics, and Physicsat the Fritz-Steinho-Gesamtschule for over two decades.The course was organized along the following ve strands:

Programming: Statements, sequences, control structures,variables and assignment, procedures, functions, . . .

(Semi-)Structured Data: HTML, XHTML, SVG, XML,XSLT/XPath, . . .

Typed Systems: Organizing and dispatching by type, e.g.classes in programming languages, or le types in anoperating system.

Multimedia: Raster graphics, vector graphics, sound, . . .

Operating Systems: Elementary functions, GUI-based ap-plications, shell, remote access, . . .

Again in line with our original proposal, we used Tcl/Tk [6]as the programming language for this course see our pre-vious article for the rationale behind this. In accordancewith the schools policy regarding the use of free and opensoftware, the course was instantiated using a Linux environ-ment. It should be noted that this choice also helped estab-lishing equity since the vast majority of students had accessto computers at home which ran some version of a Windowsoperating system (see Section 3.1 for more comments).

2.2 Details of the ImplementationThe break-down of the curriculum for the rst year of the

braided teaching curriculum into teaching units and theirassociation with the strands is shown below:

Operating Systems

Multimedia

Typed Systems

(Semi-)Structured Data

Programming

654321UnitStrand

In the remainder of this section, we outline all teachingunits and verify that each strand is indeed a strand accord-ing to Denition 1. Criterion (1) (well-defined subject mat-ter) is veried easily by the above description of the strandsand since criterion (2) (easily recognizable by the students)also depends on the way the teacher is presenting the ma-terial, only criteria (3) and (4) remain to be veried. Theabove table shows that even in the rst year three of the vestrands appear in more than one block of teaching units, so,extrapolating to the second and third year of the course, wecan assume that the fourth criterion is fullled. This leavesus with criterion (3) (more than one point of view/context)

which is indeed crucial for ensuring that students do not fallfor the misconception of a many-to-one or even one-to-onerelationship between concepts and contexts.

Unit 1: Buttons and Labels (5 Weeks).An obvious requirement for contextualized teaching is that

students should be familiar with the context itself. A com-mon body of knowledge for all students in sixth grade is thecomputer and interaction with the computer, e.g. launchinga web browser, a word processor, or a game, and it seemsnatural to start with the computer itself as context.

While students experience the GUI as the primary (if notsole) interface between the user and the machine, they donot preceive the GUI as anything but a black box. Thus,their rst encouter with Computer Science is to open thisbox and to use Tcl/Tk to build a launch button.

button .opera text "start Opera" command operapack .opera

Ultimately, as part of the Programming strand, studentscome to realize that all commands executed inside a com-puter can be represented in textual form.

With moderate eort students can design (mockup) a GUIwhich resembles the GUI they see in their everyday lives.

The second part of this unit involves creating an arrayof buttons and labels used to display each students weeklytimetable. At this point, students learn that the GUI iscomposed of dierent types of components, i.e. the coursevisits the Typed Systems strand.

Unit 2: Creating a Web Page (5 Weeks).Following up on the rst units last topic, the goal of the

second unit is to enable each student to create a small webpage displaying his/her timetable (thus making contact withthe (Semi-)Structured Data strand). At this point, studentsrealize that it becomes necessary to convert the Tcl/Tk out-put into an image that can be put on the web page. Since,in contrast to most environments of other interpreted lan-guages, the Tcl/Tk-console also allows to execute commandsof the underlying operating system, the course can move tothe Operating Systems strand rather eortlessly.

Using the import command provided by ImageMagick,students can interactively capture the section of the screendisplaying their timetable and save it as a .jpg-le.

import pause 10 timetable.jpg

The outcome of this rst encounter with the (Semi-)Struc-tured Data strand is an understanding of the (hierarchi-cal) constructs used in dening a simple web page and themarkup languages syntax. A typical markup of a web pagecreated by the students looks as follows:

Lisa s Timetable

Lisa s Timetable (Grade 6)

Unit 3: Files, Directories, and Trees (3 Weeks).As the students create more web pages (possibly using

their own images) the need for organizing the correspond-ing les becomes evident. Students realize that their les

46

should be easy to access, even if the class takes place in adierent lab. The concepts of file, directory, and tree areintroduced. In particular, students create dierent repre-sentations of directory trees and learn how to translate eachof these representations into commands for the operatingsystem level. Students learn how to create directories andto navigate in a directory tree, reacting to situational con-ditions and teacher inputs. The dierent types of entries ina directory (les and folders) and the dierent subtypes ofles present a new view on the Typed Systems strand.

Unit 4: Drawing in Tcl/Tk (8 Weeks).In the fourth unit, the Programming strand is touched

upon for the second time. According to Denition 1(3), itis mandatory to present this strand embedded in a dier-ent context. The course does so by having the studentscreate simple vector graphics using Tcl/Tk-commands thusalso embarking on the Multimedia strand.

The graphics are created either from instructions verballygiven by the teacher or trying to reproduce simple drawingshanded out to the students. In either situtation, students areasked to write a Tcl/Tk-script that executes the appropriatedrawing on the canvas. To do this, it is necessary to analyzewhich components in the beginning merely lines makeup a picture and then to translate them into a sequence ofcorresponding instructions.

Finally, students move on to (describe and) create theirown graphics which allows them to practice modelling skillsrequired in Computer Science as well as in Geometry.

Unit 5: Raster and Vector Graphics (6 Weeks).In the previous unit, students created, described, and pro-

grammed line drawings using Tcl/Tk. In the second unit,they created a screenshot from a Tcl/Tk-window and usedthe resulting .jpg-le as part of their web page. At thispoint, the students merely used the .jpg-le without hav-ing been taught about the concept of raster graphics.

Following a spiral curriculum and continuing along theMultimedia strand, the fth unit revisits the second andfourth unit, formally introduces the concepts of raster andvector graphics and points out the dierences between thesetwo types. In this unit, students work with .xpm-les in-stead of .jpg-les for a variety of reasons: First and fore-most, the X PixMap format is a textual representation of animage, and thus an .xpm-le can be opened and modiedusing any text editor. Secondly, research on misconceptionshas shown that students tend to conate a concept and itsinstantiation if only one (type of) instantiation is considered,and thus presenting a second le format helps preventing aconation between raster graphics and file format used forrepresenting a raster graphic. Finally, concepts in operatingsystems (such as extensions and associated actions) presentthemselves quite naturally.

During this unit, students work to produce a (vector)graphic in Tcl/Tk (see Unit 4), export this into an .xpm-le,modify it using a text editor, and nally view the result us-ing a graphics viewer. In addition to the strands covered inUnit 4, this unit also revisits the Operating Systems strandand presents as discussed above a new perspective.

Unit 6: Variables (5 Weeks).Piaget observes that the building up of formal relations

begins at about 11 or 12 years [8, p. 162]. Thus it is ad-

equate to introduce the concept of variable in sixth grade.To facilitate the transgression from (according to Piaget)the concrete operational phase to the formal operationalphase, it is helpful to introduce the formal construct ofvariables by using them for storing (representations of)concrete, real-world objects.

As a result of the previous unit(s), students have a varietyof images created by themselves, and in addition to this,students are provided with a collection of images. Thesepictures are stored in a directory on the le server. Studentsare taught how to load a picture into a variable and to workwith such variables, e.g. how to draw (on the Tcl/Tk canvas)multiple copies of the picture stored in a single variable.Again, the playfulness of sixth graders is build upon, andstudents create pictures of, e.g. villages with streets, cars,trees, and houses.

This unit lies in both the Multimedia and Programmingstrand. As a follow-up, students will learn to use loops forscripts which produce animated lms on the Tcl/Tk canvas.

3. EVALUATIONThe main goal of our evaluation was to investigate whether

students were able to understand the Computer Science con-cepts taught during the course of the rst year. We alsoevaluated the students knowledge of ICT topics that werenot part of the course and investigated possible changes inattitude towards topics in Computer Science and ICT.

3.1 Exams and Homework ExercisesA standard indicator of students knowledge of subjects

taught in class is their performance in exams. As pointedout before, the course in question was designed from scratchand thus no exams from previous courses could be reused.Over the duration of one year, four written exams of sixtyminutes each were administered. The distribution of thegrades is given in Figure 1.

Exam 1

0

2

4

6

8

10Exam 2

0

2

4

6

8

10

Exam 3

0

2

4

6

8

10Exam 4

0

2

4

6

8

10

A B C D F A B C D F

A B C D F A B C D F

Figure 1: Histogram of the grades for all exams.

The rst exams histogram exhibits a slight imbalance to-wards less perfect grades but is not unusual for rst exams inSciences where students still have to struggle with the eectof the rst contact with a new Science subject. The sec-ond exam, however, shows a rather dramatic shift towardsa bimodal distribution. At this point, we realized that theschools general policy not to require extensive homeworkexercises in Sciences should be overruled in favor of givinga student more opportunities to recapitulate and practice.Since even as of today, one cannot assume every student tohave access to a computer at home that provides an environ-ment comparable to the one in school, we decided to providemore but only paper-and-pencil homework exercises.

47

Not unexpectedly, the results from the third exam showa boost for almost all students which can be directly at-tributed to more practice activities due to the increased vol-ume of homework activities. The distribution of the fourthexam is almost exactly what one would expect.

Examples.The fact that students in sixth grade are relatively young

poses a particular problem for exams in Computer Science:on the one hand, students work with texts in some (program-ming or markup) language in our case Tcl/Tk, HTML, orshell commands on the other hand, they often still strug-gle with the language of instruction (both in speaking andin writing). While these problems have been found to oc-cur more often with non-native speakers (roughly a thirdof the students), another, even more challenging problem isthat ICT and Computer Science terminology has permeatedeveryday language in a way that causes ambiguities and im-precisions that cannot be (detected and resolved) until onehas obtained a full understanding of the (scientic) mean-ing. Thus, it is necessary to ban test items which demand atranslation from one language level to another.

Humans think in images real images or concepts whilethe computer manipulates (character) strings. Often thelast step of running a program is to visualize the results ofa computation. Conversely, modelling in Computer Scienceultimatly can be seen as transforming ideas (with interimstages) into a textual form. Thus, one type of test item wasto let students draw a picture of the result from running aTcl/Tk-script or rendering a page described in HTML:

Joe has created the following three images:

me.jpg bike.jpg train.jpg

He now uses HTML for a markup of his webpage:

Joe s Webpage

Joe s gorgeous webpageHere s my new webpage. Do you like it?That s me:

Things I like:

Biking and trains.

How does this page look like in a web browser?

As children at this age like to paint and draw the inversetype of test items is possible (and highly popular): Write aTcl/Tk-script for drawing a given picture or use HTML fora markup of a given web site.

1. Using paper and pencil, draw a picture of a chair anda desk using three straight lines for each.

2. Write a Tcl/Tk-script to have the computer draw yourpicture.

3.2 Final GradesFinally, we compare the distribution of the students nal

grades (which are computed from both the written examsand in-class assessment) for Computer Science, Mathemat-ics, and Natural Sciences. The distribution (Figure 2) showsthat the grades in Mathematics follow an almost perfectGaussian distribution (again, this is not surprising given thelong tradition of Mathematics Education) while the gradesin both Computer Science and Natural Sciences exhibit a bi-modality. What is interesting, though, is that there are veryfew borderline pass cases in Computer Science, i.e. thatthe fail grades were clearly separate from pass grades.This rebuts some of the concerns raised by parents prior tothe course, namely that a course in a new subject wouldaect their childrens chances of passing the nal exam.

0.0

0.1

0.2

0.3

0.4

0.5

A B C D F

CSNSMA

Figure 2: Grade distributions: Mathematics (MA)Natural Sciences (NS), and Computer Science (CS).

3.3 Quantitative and Qualitative AnalysisThe evaluation of the students knowledge of factual and

procedural knowledge in Computer Science was measured bypre- and post-test adminstration of Likert-type scale ques-tionnaires and supported by semi-structured interviews us-ing think-aloud surveys. Additionally, we gathered dataabout the students attitudes towards Computer Science.

Control Groups.For all questionnaires and surveys, we used two control

groups. The external control group (CGext, n = 117) con-sisted of students from ICT courses at two other schools inthe same school district and age group. Their teachers wereinformed of the braided teaching concept, the courses de-sign, and the fact that questionnaires and surveys were tobe administered well in advance of the beginning of the re-spective courses. Due to the repeated adminstration of thesame questionnaires, a teaching to the test-eect couldnot be precluded. To prevent such an eect to occur for therst author who was teaching the study group (BT, n = 27),the items for which interviews were performed were selectedby the second author at the end of the course based uponthe curricula of both the BT and the CGext group.

At Fritz-Steinho-Gesamtschule, students in fth gradeare required to take a 360 mins. teaching unit on text pro-cessing, spreadsheets, and the schools intranet. To measurethe eects of this teaching unit and to verify that the stu-dent population was not dierent across the three schools,we used an internal control group (CGint, n = 27) in sixthgrade that did not select Computer Science as an elective.

Questionnaires.Due to the dierent foci of the courses and the fact that

the CGint control group did not receive either Computer

48

Yes/No/Dont Know Item Pre-Test Post-TestBT/ CGint BT/ CGext BT/ CGint BT/ CGext

I know what a variable is. 0.33 0.54 0.000 0.000I know what a vector graphic is. 0.29 0.40 0.001 0.000I know what a directory tree is. 0.19 0.15 0.01 0.001I know what a directory is. 0.21 0.21 1.0 0.25One can use a variable to store a value. 0.84 0.47 0.005 0.28

Figure 3: Self-assessment of the study group relative to each of the control groups (p-values for the 2-test).

Science or (additional) ICT training, all items in the ques-tionnaires allowed for an I dont know what this question isabout answer. Each questionnaire consisted of three parts:

(i) 29 yes/no/dont know-type items (including cross-validating items) that aimed at the self-assessment offactual knowledge (e.g. I know what an operatingsystem is or Variables can be used to store values).

(ii) 20 Likert-type items that aimed at the self-assessmentof abilities (e.g. I know how to use a word processingsystem or I know my way around in the WWW)and attitudes towards programming (e.g. Program-ming is important in Computer Science or If a pro-gram works, it is not important how it was programmed).

(iii) 21 Likert-type items that aimed at personal experi-ences and attitudes towards Computer Science (e.g.Im using a computer many times a week, I enrolledfor this course because my parents made me to, Boysare more successful in Computer Science than girls).

Due to space constraints, we only report on the evaluationof a few of the items related to factual knowledge, i.e. takenfrom the rst part of the questionnaire. These items askedthe students to assess whether they knew what a certainconcept in Computer Science or ICT is.

Figure 3 summarizes the statistical evaluation of the self-assessment in pre- and post-test questionnaires. For eachsetting, we used a 2-test for the hypothesis that the BTgroup and either of the two control groups responded ac-cording to the same distribution. For none of the items, theevaluation of the pre-test signicantly substantiated a rejec-tion of this hypothesis, i.e., the groups responses could notbe discriminated. The evaluation of the post-test, however,showed that seven out of ten times, the hypothesis has to berejected with a very high level of signicance (p 0.01). Indecreasing order of signicance, these items pertain to vari-ables, vector graphics, and directory trees. Since the resultsare statistically signicant for both the comparison withCGint and CGext, we can conclude that the students in theBT group have a higher condence in their level of knowl-edge (for instance, the distribution of yes:no:dont know(in percent) for the vector graphics item was 78:11:11 forthe BT group while it was 31:54:15 for CGint and 22:39:39for CGext). The items for which no statistically signicantdierence could be observed pertain to directories (i.e. aconcept students might be acquainted with from their ev-eryday use of a computer) and the question of whether it ispossible to use a variable to store a value (see next section).

Interviews.To validate the students self-assessments, a total of 54

interviews was conducted during the last week of the school-

year. 19 interviews were conducted in the BT course and35 interviews were conducted in the CGext and CGint group.The participants from the CGext and CGint groups were se-lected by their respective teachers, thus we can safely assumethat the level of knowledge represented in these interviewsdoes not underestimate the average level of the group.

The interviews touched upon the following topics: (i) lesand directories, (ii) text processing, (iii) variables, and (iv)graphics. Of these, topics (i), (iii), and (iv) are part of thebraided teaching curriculum, and topics (i), (ii), and (iv)are part of the ICT Literacy part of the Informationand Communication Technology in Secondary Educationcurriculum of the UNESCO [3, Units A2, A3, and A6].

We classied the interviews addressing a certain topicw.r.t. whether the topic was unknown or known. If it wasknown, follow-up questions were evaluated to classify howmuch (no/partial/full) understanding was evident.

For the variable concept (for which the dierence inself-assessment was strongest), the interviews conrmed astrong dierence between BT and the control groups

variable n unknown no partial fullBT 18 17% 28% 22% 33%CGint + CGext 18 100% 0% 0% 0%A similar conrmation was made for the vector graph-

ics concept; this item, however, also revealed an even moreprominent dierence in self-assessement (where 78% statedthey knew the concept) and external evaluation (where 56%were found not to know the concept by name). This as-tonishing dierence in combination with the fact that thestudents were rather well able to work with vector graphicsin the exams (see above) prompts us to revisit the respectiveteaching unit to ensure that the concept as such is presentedand contrasted with raster graphics much prominently.

To examine the eects of the BT course on the knowledgeof ICT concepts, we included a question related to directorytrees; also, we asked the students how they would (in aword processing systems) transform a left-aligned text intoa centered text. For this question, we provided printoutsof the text in left-aligned and centered form. Both items(subdirectories/centering) are explicitly mentioned inthe UNECSO ICT curriculum [3, p. 59/61].

directory tree n unknown no partial fullBT 13 38% 15% 23% 23%CGint 12 58% 33% 8% 0%CGext 4 75% 0% 25% 0%The responses to the rst question showed that students

from the BT group showed a much deeper understandingof the hierarchical concept represented by a directory tree an indication that Computer Science indeed augments theunderstanding of an ICT concept.

As mentioned in Section 2.2, one goal of our course was toopen the black box as which a computer and the software

49

is perceived. To our (pleasant) surprise, the success w.r.t.this goal became evident when evaluating the responses tothe text centering question which had been designed toverify the hypothesis that Computer Science and ICT couldbe distinguished by, e.g., the procedural knowledge w.r.t.using a word processing system. We classied the responsesaccording to whether students suggested to manually insertwhite spaces, to change the justication to centered, or both.

centering n no idea white space justify bothBT 14 0% 21% 43% 36%CGint 12 34% 25% 41% 0%CGext 19 21% 42% 31% 4%While a signicant percentage of students in each of the

groups suggested the appropriate solution, it is striking toobserve that the BT group was the only group where everystudent interviewed could solve the problem and almost 80%chose the appropriate answer. The answers to this itemindicate that students were in fact able to transfer a concept(justication) from a plain-text markup language (HTML)to the use of GUI in a word processing system.

Gender Issues.Reports on in-class and outreach activities imply that one

crucial factor inuencing the recruitment of female studentsis to debunk (negative) stereotypes and to do this as earlyas possible in the students careers see, e.g., [4]. We fol-lowed up on (part of) a recent study by Taub et al. [9] andincluded one item in the third part of the questionnaire thatasked for an assessment of the hypothesis that boys are moresuccessful in Computer Science than girls.

Group n Pre-Test Post-Test p-value

BT (female) 8 2.38 (1.77) 2.11 (1.76) 0.019CGint (female) 13 3.08 (1.16) 2.17 (1.03) 0.014BT (male) 19 3.00 (1.38) 3.14 (1.74) 0.003CGint (male) 14 3.11 (1.27) 2.21 (1.48) 0.005

Figure 4: Boys are more successful in ComputerScience than girls.: mean/std. deviation (six-levelscale) and p-value for Wilcoxons signed rank test.

The statistical evaluation (Figure 4) of the students re-sponses supports the following observations and hypotheses:(1) The pre-test shows that female students who chose theBT course revealed a stronger disagreement with the hy-pothesis than any of the other groups, i.e. a non-negativeattitude towards Computer Science can exist even in fthgrade and prior to any contact with Computer Science and this attitude can be capitalized on. (2) Over the courseof one year, all groups signicantly changed their attitudetowards the gender issue; this supports the hypothesis thatoutreach activities should target middle schoolers. (3) Ex-cept for the male students in the BT group, the results inthe post-test were almost identically distributed. This al-lows for the following interpretations: (3.1) Female studentsin the BT course positively experienced their abilities andvalidated their already strong disagreement. (3.2) Male stu-dents in the BT course also positively experienced their abil-ities and were led to a slightly stronger agreement. (3.3) Allstudents in the CGint group did not have any contact withComputer Science, and thus external factors such als grow-ing up and the inuence in other subjects, e.g. Mathemat-ics, lead to a uniform shift towards disagreement.

The disagreement with the hypothesis is stronger thanreported by Taub et al. [9]. Whether this is due to the coursedesign (Computer Science vs. Computer Science Outreach)or to external factors (see above) remains to be investigated.In fact, the result that all groups except the male students inthe BT group were led to stronger disagreement raises thequestion whether such external factors (possibly includingthe male(!) teacher as a role model) are predominant.

Summary.The evaluation of the rst year of our braided teaching

course is encouraging: we have demonstrated that a Com-puter Science course worth full curriculum credit can besuccessfully implemented in sixth grade and that statisti-cally signicant results related to both factual and procedu-ral knowledge as well as to attitude can be observed. Be-sides continuing the course described, future work will focuson long-term eects and on revisiting the teaching materialwith a stronger focus on conceptual knowledge.

AcknowledgmentsThe support of the colleagues teaching the CGext courses aswell as of the administration of the Fritz-Steinho-Gesamt-schule was essential for this project. The authors thankHolger Danielsiek, Rebecca Doherty, Jorn Godel, and Wolf-gang Paul for their help with evaluating the questionnairesand with conducting and transcribing the interviews.

4. REFERENCES[1] T. Bell, P. Andreae, and L. Lambert. Computer science in

New Zealand high schools. InProc. 12th AustralasianComputing Educ. Conf., pp. 1522, 2010.

[2] J. S. Bruner. The Process of Education. HarvardUniversity Press, Cambridge, MA, 1960.

[3] Y. Buettner, C. Duchateau, C. Fulford, P. Hogenbirk,M. Kendall, R. Morel, and T. van Weert. Information andcommunication technology in secondary education: Acurriculum for schools. UNESCO, Paris, 2000.

[4] S. Graham and C. Latulipe. CS girls rock: Sparkinginterest in computer science and debunking thestereotypes. In Proc. 34th SIGCSE Symp. ComputerScience Education, pp 322326, 2003.

[5] O. Hazzan, J. Gal-Ezer, and L. Blum. A model for highschool computer science education: the four key elementsthat make it! In Proc. 39th SIGCSE Symp. ComputerScience Education, pp. 281285. 2008.

[6] J. K. Ousterhout. Tcl and the Tk Toolkit. Addison-Wesley,Reading, MA, 1994.

[7] A. Pasternak and J. Vahrenhold. Braided teaching insecondary CS education: Contexts, continuity, and the roleof programming. In Proc. 41st SIGCSE Symp. ComputerScience Education, pp. 204208. 2010.

[8] J. Piaget. The Psychology of Intelligence. RoutledgeChapman & Hall, London/New York, 2nd ed., 2001.

[9] R. Taub, M. Ben-Ari, and M. Armoni. The eect of CSunplugged on middle-school students views of CS. In Proc.14th SIGCSE Conf. Innovation and Technology inComputer Science Education, pp. 99103. 2009.

[10] C. Wilson, L. A. Sudol, C. Stephenson, and M. Stehlik.Running on empty: The failure to teach K-12 computerscience in the digital age. ACM/CSTA, 2010.

[11] J. M. Wing. Computational thinking. Communications ofthe ACM, 49(3):3335, Mar. 2006.

[12] A. H. Yahya. The interaction between high schoolcurriculum and rst year college courses: The case ofcomputing. In Proc. 41st SIGCSE Symp. ComputerScience Education, pp. 406410. 2010.

50