SQuID: developing data simulation tools for research and teaching of multi-level statistical analyses

Nesten alle vitenskapelige disipliner krever nå kunnskap om kvantitative metoder, noe som har stor samfunnsmessig betydning for naturfagundervisning og forskning. I biovitenskap forbedres datainnsamling og analyse raskt, fra DNA-sekvensering med høy gjennomstrømning til automatiske sporingssystemer for atferd og fysiologi. Effektiv bruk av statistiske metoder er derfor avgjørende for biologisk undervisning og forskning. Som svar på dette har vi utviklet SQuID (Statistical Quantification of Individual Differences) – et internasjonalt samarbeid mellom forskere og lærere som bruker skreddersydde datasimuleringer for å forstå komplekse statistiske modeller og undersøke mangfold i biologiske systemer. INTPART-utvidelsen av SQuID omfatter over 12 akademiske institusjoner i 9 land (Norge, Canada, Frankrike, Tyskland, USA, Ungarn, Brasil, Australia og Japan) og kombinerer samarbeidsforskning i verdensklasse med fremragende naturvitenskapelig utdanning. Som en del av SQuID INTPART-bevilgningen har vi ferdigstilt vår nettbaserte datasimuleringsprogramvare med åpen tilgang, samt statistiske analyseressurser og veiledninger som er fritt tilgjengelige for studenter og forskere (http://squidgroup.org/). I løpet av de siste tre årene har vi gjennomført seks fullfinansierte internasjonale forsknings- og undervisningsverksteder for over 215 doktorgradsstudenter verden over (Norge, Frankrike, USA, Brasil, Japan og Canada), inkludert ytterligere tre mindre satellittverksteder i Brasil og ett eksternt finansiert verksted i Mexico. Disse workshopene har resultert i to internasjonale publikasjoner om statistiske metoder, med fire flere under arbeid, samt en kommende pedagogisk forskningsartikkel som evaluerer våre interaktive undervisningsmetoder. Vi har også finansiert og veiledet ni forskningsopphold for PhD- og postdoktorale forskere fra Storbritannia, India (3), Sveits, USA, Uruguay, Spania og Frankrike ved SQuID-medlemmenes forskningsgrupper, hvorav de fleste vil resultere i ytterligere forskningspublikasjoner.

A major outcome and impact of this SQuID INTPART grant has been the deliberate expansion and development of the SQuID international cooperative network, especially amongst female researchers and those in developing countries (e.g. Brazil). All new research collaborators and workshop teachers have been added to the SQuID network, plus the nine recipients of our travelling research fellowships. These new younger members are now actively involved in workshops, research publications and future plans for SQuID. The employed two post-doctoral fellows also became major parts of the SQuID network, and their careers continue to benefit greatly from this. Most importantly, they completed the SQuID open-access online data simulation software (squidsim) and statistical analyses resources and tutorials as shiney apps. These are now freely available to students and researchers worldwide alongside SQuIDs active social media presence (see http://squidgroup.org/), and are positively impacting users around the world with and without access to formal courses. The largest impact of this grant must be the teaching of more than 215 graduate students at workshops in Norway, France, USA, Brazil, Japan and Canada, as well as Mexico. This positive academic impact is evidenced by improvements in answers to our anonymous online questionnaires immediately before versus after each workshop regarding student ability and confidence in using advanced statistical models. A follow-up questionnaire recently sent around to all students has confirmed this long-term positive impact of our teaching methods on knowledge and ongoing use of advanced statistics by these early career researchers worldwide. Our SQuID workshops also had a sustained impact on the effectiveness of statistics teaching by local educators involved, plus many of the students now have access to wider international research, including the SQuID network. A major impact has been the nine travelling research (non-salary) fellowship visits to the research groups of SQuID members by PhD and post-doctoral researchers from the UK, India (3), Switzerland, USA, Uruguay, Spain and France. These have allowed a unique opportunity for individually customized training and career development, especially for the graduate students from developing countries. An important outcome here is the scientific publications arising from our travelling fellowships, again with substantial positive professional impacts for such early career researchers. SQuID members themselves have thus far published in two high-quality international statistical methods papers, with at least four more to come either submitted or late stages of preparation, plus plans for quite a few more arising from all of our workshop discussions, etc. An additional educational research publication is in preparation assessing SQuID’s interactive teaching methods used in our workshops, which should impact the methods used in future statistical teaching far beyond SQuID.

SQuID (Statistical Quantification of Individual Differences) is an international network using customised data simulations of variability at multiple hierarchical levels to improve the use of generalized linear mixed models (GLMMs). This INTPART extension of SQuID will formalize and thus facilitate world-class collaborative research with excellence in education in Norway by creating a long-term international cooperation with 12 academic institutions in 9 countries. SQuID will accomplish this via three linked activities at the SFF Centre for Biodiversity Dynamics (CBD), NTNU, that will further extend its research excellence in quantitative methods and develop robust future teaching programs at NTNU. First, using workshops at CBD and 5 focal institutions in target countries, we will formally trial various innovative teaching methods as part of short graduate courses for local students and faculty. Second, original research programs will be promoted in 4 subtopics in which GLMMs could have significant impact within biology. Third, the SQuID opensource data simulation and interactive tutorial software will be further developed for use in research, formal statistics courses and individual-based learning at NTNU and elsewhere. The expanded SQuID collaborative network of institutes will allow us to focus on key research questions with specific relevance to local and national strategy, which will in turn drive improvements by CBD staff in the SQuID data simulations and online tutorials. Research exchanges for promising researchers and students will enhance these activities and identify future members of the SQuID network. The effectiveness of SQuID interactive teaching, use of social media and online tutorials for both classes and distance learning will be assessed with the help of NTNU educational professionals, allowing for greater improvement and dissemination of best-practice in these educational approaches at NTNU, across Norway and beyond.