10th International
Conference on
Teaching Statistics
8 – 13 July 2018
Kyoto, Japan

Focus on Posters

P1: Monday 9th, 17:30-18:30
P2: Tuesday 10th, 17:30-18:30

Note: only the presenter is
shown below



Poster session P1 (Monday 9th, 17:30-18:30, Foyer)

1The Impact of Homework Feedback on Statistical Literacy in Austrian Computer Science Students Thomas Forstner
This paper compares statistical literacy measured by a self-developed test between two types of feedback on homework assignments in a basic statistics course: (A) no immediate feedback from the lecturer and (B) immediate feedback from the lecturer. Group A consists of around 200 business computer science students and group B of around 200 computer science students. Each group was given a week to prepare a homework assignment (in total 10 assignments had to be prepared). The students of group A handed in the homework on paper and got back the corrected homework after a week. The students of group B presented the homework in class, so immediate feedback could be provided. At the end of the semester the statistical literacy of these two groups was compared.

3Learning Statistics from data Yuan-Chin Chang
Digging out information from data is essential for learning Statistics. Modern computing and communication technologies can make data collection very efficient. However, our ability to analyze and extract information from large data sets is hard-pressed to keep up with our capacity for data collection. It is know that all statistical methods are developed with some data analysis tasks in mind. However, when we teach these fabulous statistical methods, we usually apply a "topic-oriented" teaching scheme, which is useful to let students learn the skill of particular methods. However, it is hard to pass the original thoughts or reasons why we developed these methods. Here we like to share our experience with problem/data analysis oriented teaching arrangement for an introduction course in our data science program.

5Practice of Proactive Utilization of Mathematical Representation in Learning of Statistics Kunihiko Shimizu
By providing the children the trial and error training, I urge the children to make use of mathematical representation on their own initiative. So, as a developmental learning of statistics of the first year senior high school, I dealt with the transformation of the variable based on the roll of the dice. From the previous result, I asked the teacher to let the students discover the problem with the change of numbers/variables from the rolling of the dice. The students' trial and error development on the new data in the following cases, they analyzed and concluded: (1) new data obtained by adding constants all together, (2) new data obtained by multiplying data all. As a result of practice, I could encourage subjective use of mathematical representation.

7Context of statistical graphs in primary school textbooks in Costa Rica Pedro Arteaga
The aim of this research was to analyse the tasks including statistical graphs in primary education textbooks in Costa Rica. It is part of a more extensive project, where the way in which statistical graphs are presented in the textbooks in Costa Rica is analysed. In order to achieve this aim we performed a content analysis of all the activities (n=167) related to statistical graphs in the two books series (grades 1rst to 6th) which are most widely used in Costa Rica. Specifically we analyse the graph context, taking into account the categories proposed in the PISA tests, which have contributed to the current renewed interest in context-based education

9Learning of statistics in a Mexican telescondary school: impact of statistical enquiry in students outcomes Alberto Santana
Based on some guidelines of Wild and Pfannkuch (1999), Sovak (2003) and Makar (2008) for the implementation and evalution of statistical enquiry in the classroom, a didactic strategy was designed to improve the learning of statistics in third-grade students of a Mexican telesecondary -secondary school with televised classes-. We analyzed the impact of this strategy, for which a quasi-experimental study was carried out with three control groups and one experimental group. To evaluate the learning outcomes, a pretest and postest based on the CAOS-4 test were applied. The quantitative analysis performed shows an improvement in the learning of curricular statistical knowledge. This suggests that statistical enquiry is a method that helps, and motivates, students to learn the main subjects of statistics for the secondary level.

11The introduction of chemometrics course as a step forward to modernization of study curriculum at engineering faculties Sanja Podunavac-Kuzmanovic
Chemometrics, as a relatively young scientific discipline, has developed during 1970s. In 1980s Chemometrics was distinguished as a scientific discipline and multidisciplinary approach in data analysis and processing particularly in chemistry, applying integrated mathematical and statistical methods. Since then, the application of chemometrics overcomes the frames of chemistry. Nowadays, chemometrics is being successfully used in engineering, including food engineering, pharmaceutical engineering, chemical engineering, agriculture and biotechnology. Considering the high position that chemometrics has on many universities in the world, the main aim of the present paper is to emphasize the importance of studying chemometrics at engineering faculties, the possibilities that chemometrics offers in fast and efficient data processing, prediction of outcomes and in saving time and financial resources.

13Modified Information Matrix Tests for Detecting Misspecification in the Random Effects of GLMMs Kuo-Chin Lin
Generalized linear mixed models (GLMMs) are commonly applied to regress a non-Gaussian clustered structure response for hierarchical data analysis and longitudinal studies. The normality assumption of the random-effects distribution in GLMMs is practically assumed, but it may be too restrictive to reveal the major feature of data. The test statistics are proposed based on a variety of modified information matrix tests introduced by White (1982), and their limiting chi-squared distributions are derived under the null hypothesis that the distribution of random-effects is corrected specified. Simulation results are presented under various configurations of practical relevance data generating mechanism with different modified matrices, and the power performance of the proposed tests are demonstrated. Furthermore, real longitudinal case studies are employed to illustrate the applications of proposed tests.

15Every learner is unique so why couldn’t their assignments be unique? The use of simulated datasets in inferential statistics courses. S. Jeanne Horst
Data simulation is a technique used for robustness studies, model estimation, and investigation of new analytic methods. We incorporated simulated data into inferential statistics course assignments. Throughout the course, students completed formal assignments to answer identical research questions, performing analyses on their own unique instructor-provided simulated dataset. Using simulated data allowed the instructor to control outliers, assumption violations, and results. Course evaluation ratings of the assignments showed that students exhibited greater passion compared to more traditional assignments. Additionally, compared to traditional assignments, the technique minimized honor code violations by eliminating the ability to copy results and interpretations. To expedite the grading of unique analyses, solution syntax is developed that automates the process. Example assignments, solution syntax, and student evaluation of the assignments are presented.

17Developing secondary preservice mathematics teachers’ statistical knowledge for teaching Stephanie Casey
The MODULE(S^2) project funded by the National Science Foundation in the United States has created teacher education curriculum materials that develop preservice secondary (grades 6-12) mathematics teachers’ statistical knowledge for teaching. This poster presentation will share the design principles of the project that detail our approach to developing teachers’ statistical content knowledge while at the same time developing their pedagogical content knowledge in ways that are closely tied to the practice of teaching statistics. Excerpts from the curriculum materials that illustrate these design principles will be included (for example, how we have teachers respond to students’ work), as will pilot research results on the materials’ effectiveness.

19Integration of Extensive Technology in a Canadian Service Statistics Course Wesley Burr
At Trent University (Ontario, Canada) we recently migrated our large first-year undergraduate service course in statistics to a randomization and simulation approach, with mandatory integration of R code and content for all students. In addition, we moved all homework to an online WeBWorK platform with weekly deliverables, and introduced a number of other technological changes, including videos of lectures and tutorials and an online persistent chat system. In this poster, we explore the results of this process, including student feedback, student performance, and general comments from the instructors and TAs for the two sections of the course. We conclude with some hard lessons learned, and some suggestions for any other instructor or institution thinking of implementing any or all of our recent changes.

25A study of practical statistics education using questionnaire survey, through the joint education of companies and universities Mie Fujiki
It is a study on a lecture method of "Innovation-Challenge Program" which started last year from Faculty of Business at Aichi Shukutoku University, a part of marketing and statistical education. This program implements jointly by companies and universities in five groups. We show the case of a lecture method about financial institutions. In this lecture, each groups of five or six students conducted market research and proposed services according to customer's needs. They created question items, gathered data using the Internet, and made suggestions from the results of the aggregation. Moreover, we show the results of the analysis related to student evaluation for group study.

27University Math and Stats Support Centre and how to reduce a rate of unsuccessful studies Maria Králová
Mathematics and Statistics underpin many university subjects in many disciplines (economics, engineering, biology, ...), but are often perceived as too difficult. This poses barriers to successful study. Furthermore, students often face difficulties with a data-driven approach in their final theses. Many higher education institutions throughout the world have established Mathematics and Statistics Support Centres (MSSC) to assist students with these issues. The first MSSC in the Czech Republic was established in 2016 at Masaryk University. Since then it has inspired other universities in the Czech Republic. It operates in a drop-in mode with support of staff-tutors and volunteering students. Since its establishment, there were more than 1000 visits related to study within the curriculum and for support on students' final theses.

29A novel way to teach biostatistics to public health students Qi Zheng
Public health students need conceptual knowledge to correctly apply biostatistical procedures. Categorical data analysis particularly presents conceptual hurdles to many students, but an unrealistic emphasis on conceptual knowledge allows unfamiliar mathematics to supplant logical reasoning. This presentation gives concrete examples of how I impart conceptual knowledge to public health students in a categorical data analysis course. By relying on high school algebra and first principles, I first demonstrate the algebraic connection between probabilities and logits. Then, via SAS software, I show students how to code the likelihood functions for several important models. This paves the way for students to appreciate abstract concepts like the deviance and the likelihood ratio statistic. The presentation illustrates the new teaching method by classroom-tested problems and their solutions by students.

31Information Request of the Society as a Condition for the Development of Statistical Literacy Elena Zarova
Information society as a new stage of social and economic systems development determines a special type of interaction between the official statistics and its users: the population, business, authority. The creation of official statistical information due to current users' requests raises the need, on the one hand, of adequate statistical literacy of users of information, and, on the other hand, the adaptation of official statistical methodology to user requests. Authors demonstrate the results of the interaction of the Moscow Analytical Centre with the Federal Service of State Statistics of Russia (Rosstat) of the methodology improvement for calculating earnings and income indicators. The conclusions include proposals for methodological development of official statistics in order to improve its effectiveness, taking into account the development of statistical literacy.

33Modelling and statistics in food product design - modernization of food engineering courses Gordana Cetkovic
Modelling of food processes allows food engineers not only to understand these processes more clearly but also to control them more closely and make predictions about them. The application of predictive models in foods has emerged significantly in the last two decades due to development of computer science and statistical packages. The aim of defining predictive models in food science is to ensure safety and quality of food. Modelling could be a contemporary tool for designing food with higher and consistent quality. Applying modelling and statistical methods to food processing enables engineers to predict behavior of food products under different combinations of factors. This study presents the range of modelling techniques and their applications across the food chain for students at undergraduate food engineering courses.

35C-SOMAS: Measuring Classroom Characteristics Marjorie Bond
The family of instruments known as SOMAS (Survey of Motivational Attitudes towards Statistics) will measure classroom characteristics in C-SOMAS. While the student and instructor instruments (S-SOMAS and I-SOMAS) are based on Expectancy-Value Theory (Eccles, 1983, Eccles &Wigfield, 2002), our model for the classroom characteristic instrument is broken into two factors with three elements in each. The two factors are split by the instructor’s locus of control. While influenced by the instructor, the first factor, institutional structures and characteristics, is not fully within the instructor’s control and will vary between instructors. The elements are institutional characteristics, course characteristics, and learning environment. Our link between I-SOMAS and C-SOMAS is the second factor, enacted classroom behaviors, which consists of general pedagogy practices, statistics-specific pedagogy practices and teacher-student relations.

37Basic statistics starts with bivariate for academic degree program Lita Pertiwi
In official statistics, bivariate statistics is used to start long-term statistical education such as covariance matrix. We propose that variance and covariance be taught simultaneously. Iterative proportional fitting (raking) is another form of statistics from two different sources or two different periods. Statistics can be introduced by using at minimum two sources of data or bivariate. Ever since computer time be affordable (inexpensive processing cost), statistics can be introduced in dimensionality reduction instead of merely generalization from univariate. Example is to start with structural equation modelling (SEM) concept then proceed with multiple regression. Example is to start with trivariate projection pursuit then apply principal component analysis (PCA). The views here are that of author and co-authors and not necessarily reflect official views.

39A web-based learning system for junior high school students and high school students Takatsugu Yoshioka
A web-based learning system "Data-oriented Statistical System" or called DoSS@d (http://mo161.soci.ous.ac.jp/@d/index.html) has been developed mainly for educational use, which archives a large number of datasets and the corresponding analysis stories so that students can find suitable datasets for their purposes and learn practical data analysis based on the analysis stories. DoLStat@d, one of modules in DoSS@d, supplies several courses consisting of analysis stories suitable for learning purposes, e.g., marketing, visualization, and classification. In this study, new courses with "PPDAC cycle" for junior high school and high school students are developed. The contents are basic statistics and data visualization that junior high and high school students ordinally learn in mathematics and are based on PPDAC cycle.

41A case study on the development of statistical thinking using multivariate data: Based on Olympic Decathlon Koki Hosoda
In highly-networked information society, statistical thinking to judgment and prediction based on data is required. To develop high school student's statistical thinking, we focused on teaching materials using multivariate data. The purpose of study is to clarify the students' process of using multivariate data and to examine the way of guidance to develop statistical thinking. To achieve them, firstly, we developed the teaching material using Olympic Decathlon data for Grade 10. Secondly, we analyzed small lesson about it for 7 students. The result, we revealed difficulty of focusing on the relevance of the data variables and statistical evidence. As conclusion, it is necessary to teach analyzing data qualitatively and quantitatively and data variables difference.

43Mastery-Based Grading in Introductory Statistics Katherine Kinnaird
Mastery-based grading values iterative improvement and deep learning of course material over high-stakes cumulative assessments. This poster shares a method of adopting this grading method to a large introductory statistics college course with an audience of largely non-majors. Central to this method is outlining 15 concepts and tasks (called `standards) that, by the end of term, students should be able to articulate and appropriately apply. Each standard receives a level of Gold (meaning excellent), Silver (meaning mastery of the concept), Bronze (progressing towards understanding), or cannot be assessed. The level-based scoring system transparently signals to students where their understanding can be deepened. Evaluating each standard individually and giving students many opportunities to demonstrate their knowledge together incentivizes students to review their understanding of concept.

45Students competences in dealing with the classical and frequentist approach to probability Tobias Rolfes
The concept of probability is a key concept. The classical approach was defined by Laplace and prevailed in teaching probability at the beginning. The focus on the classical approach raised criticism and nowadays the frequentist approach is widely promoted for teaching probability. The aim of the project was to identify secondary students’ competences in dealing with the classical and the frequentist approach to probability. Therefore, approximately 500 students in Germany from grade 8, 9 and 10 performed a paper and pencil test. The analysis of the item difficulties was conducted with item response theory. The results showed that students had basic competences in dealing with the classical concept of probability but struggled to apply and comprehend the frequentist approach to probability.

47The construction of the central ideas of the sampling through technology in university students Amable Moreno
In this paper we present an instructional proposal for the understanding of some ideas related to sampling in university students, by applying the R software. These ideas are the sampling representativeness, the sampling variability and the distribution concept in four levels. The students in order to achieve their understanding, must establish differences between samples and populations; recognize differences and similarities between samples of a population. Then, they must distinguish between four levels of the data: distribution of the population (level 1), distribution of a sample (level 2), distribution of the random sample as a random n-dimensional variable(level 3) and distribution of sample statistics (level 4). The use of simulations and visualizations will help students to achieve an understanding of these ideas.

49Do standardized tests contribute to statistical education? Adriana D’Amelio
In every country to measure student performance there are different national tests in different disciplines including mathematics. The statistic that is part of the Mathematics curriculum in standardized tests represents 25% or less of the test. The disaggregated results in general are not available for decision-making in the statistics as a discipline. Its availability would compare the performance of the students with the results of the international tests. This helps to define policies of education and training in the statistical discipline in order to be able to make decisions at national level. Such education and training should be aimed at teachers and students. A concrete example is presented with a Provincial and National test of Argentina,taken by students at the primary and secondary levels.

51What Would Fisher Do? A Useful Tool for Teaching the Model Construction Process Julie Garai
Students often struggle converting study descriptions into plausible statistical models. What Would Fisher Do (WWFD) first appeared in Stroup (2013) as a tool to motivate the model construction process for generalized linear mixed models. Inspired by Fisher’s comments following Yates (1935) “Complex Experiments,” WWFD is a technique that organizes the design and treatment structures of the study. The result is a re-envisioned ANOVA table that provides a basis for constructing appropriate models. WWFD aids in understanding the difference between fixed and random effects, the impact of the response variable distribution on the model, and the role of the residual in context of different distributions. Examples of WWFD implementation in both undergraduate and graduate courses in statistical modeling and design of experiments are presented.

53Statistical Practices: What do Statisticians do? Layla Guyot
The demand for statistical skills is growing in many different fields and sectors, and the employment of statisticians is expected to increase drastically. However, the transition from learning to practicing statistics is challenging (Gibbons & MacGillivray, 2014) because professional statisticians are developing ways of reasoning and practices that grow out of experience (Pfannkuch & Wild, 2000). In order to investigate statistical practices that are developed at the workplace, statisticians were solicited to reflect on their experience. Each statistician sorted and identified prevailing practices performed at the workplace. The analysis revealed predominant and newly developed statistical skills. This identification of skills from the perspective of the statistical practitioners helps inform how to better promote an authentic experience of statistical practices throughout education.

55Dealing with Symbols with Multiple Meanings in Inferential Statistics Von Bing Yap
A mathematical symbol can change meaning without warning, like in solving an optimization problem. The derivation of a maximum likelihood (ML) estimate inherits this issue, resulting in the parameter symbol θ acquiring multiple meanings in close proximity to each other. It is proposed that the instructor ought to explicitly indicate key points where the meaning of θ changes, and to introduce the hat notation θ-hat for the ML estimator carefully. In particular, the use of θ-hat; for a realization, i.e., an estimate, should be avoided. It will be demonstrated that words are powerful tools for untangling the overworked symbols, hence can be valuable for students to master the central ideas of inferential statistics.

59Effects and characteristics of representative values for the 6th grade in elementary school Naoki Ohta
In light of the progress of our advanced information society, statistical education using ICT equipment based on the PPDAC cycle is taught in many countries. Recently, Japan has designated “representative value” as its educational content for 6th grade of elementary school, for example mean value, mode, and median. This research aims to identify the effect of learning and its characteristics by teaching representative values for histogram as educational content. As a result of this investigation, it was easy for 6th grade children to understand the mean value and mode. However, an incorrect recognition of the median was confirmed.

61Data Science Education using Statistical Graphic Concour Chitose Nishiyama
It has been becoming important to consider effective Data Science education. To develop the student’s ability to process collected data and to explain obtained results so that everyone easily understands them, it is necessary to make students process real data. A higher motivation is very important for such education. One of successful approaches is to challenge to data competitions such as "Statistical Graphic Concour" in the classroom. The important aspects here are to collect suitable data for the purpose; to select contents to be resulted; to make several graphics have a story; and to visualize them understandable. In this study, a trial of Data Science education with the challenge to "Nation-wide Statistical Graphic Concour" designed for university students is introduced and the results are discussed.

65Comparison of assessment methods in an undergraduate biostatistics course Jennifer Daddysman
Improving instructional methodologies often requires evaluation of assessment strategies. With the variety of assessment techniques in education, it can be difficult to determine the best one(s) for any particular course. Closed-book exams have traditionally been used to assess knowledge in biostatistics courses; recently, project-based learning has been incorporated to evaluate student learning. As biostatistics is an applied field, project-based learning may be better suited as an evaluation tool than exams. An initial sample of approximately 50 undergraduate biostatistics students had their baseline knowledge measured using a pre-test, and their post-course knowledge was assessed using a post-test. The 20 multiple-choice questions on the pre- and post-tests were repeated; comparison of change scores are used as the primary determination in effectiveness of assessment methods.

67Comparison of Campus and Online Sections of a Flipped Introductory Statistics Course Craig Johnson
Brigham Young University-Idaho (BYU-Idaho) embraces a Learning Model, where students prepare in advance of class meetings, teach one another, and reflect on their learning. Under the Learning Model, flipped courses are the standard. Students can enroll in either face-to-face or online sections of three different types of introductory statistics courses: Business Statistics, Biostatistics, and Social Science Statistics. Comparisons of exam scores and surveys illustrate differences between three different groups: on-campus students in face-to-face sections, on-campus students in online sections, and remote online students. Student performance and perception is compared for the various sub-populations.

69Validity Evidence Claims and Plan for the SOMAS Instruments Douglas Whitaker
The Surveys Of Motivational Attitudes towards Statistics (SOMAS) is a family of instruments being developed to measure what has previously been conceptualized as attitudes in statistics education. For the Student and Instructor instruments (S-SOMAS and I-SOMAS, respectively), the SOMAS project uses Eccles’s Expectancy-Value Theory (EVT). While other attitude surveys have used EVT in some capacity, EVT is central to the development of the S-SOMAS and I-SOMAS instruments. As part of the development process, a robust view of validity evidence has been adopted. This poster will articulate explicit claims and evidence to be gathered to support these claims using the validity framework proposed in the Standards for Educational and Psychological Testing (AERA, APA, & NCME, 2014).

71Adapting statistics education to a cognitively heterogeneous student population Hilde Vinje
Historically, the introductory course in statistics at the Norwegian University of Life Sciences(NMBU), has been lecture based. Previous study at NMBU concluded that the course structure apparently disfavored certain cognitive types. Therefore the course was restructured into a student active learning course using flipped classroom. Output variables like exam scores, colloquium attendance and student evaluations were analyzed in light of cognitive information on the students as collected by an education test provided by the National Centre for Science Recruitment. One of the main findings shows that the previously negative effect of extraversion (E) disappeared in the flipped classroom course. In this paper we present several findings that indicate that additional adaptations should be made to reach an even wider group of the heterogeneous student mass.