A three-arm cohort study with negative and positive control groups was conducted at four Danish Agriculture colleges. Requirements for conducting the study were to conflict minimal with the curriculum at the schools and implement the study into the schedule. Small changes in the schedule were necessary to ensure teaching in farrowing management at the start of the course. Students were divided into three groups at each school. Each group received instruction covering farrowing management but with three different didactic approaches. All groups received traditional classroom instruction. To ensure that classroom instruction had the same content between schools, a list of learning goals were shared between responsible teachers. These learning goals were all about management around farrowing. The subject farrowing management was for all schools integrated in a larger course about general husbandry. Farrowing management was taught in classroom between two lectures (90 min) to four lectures (180 min) between schools. Group one (classroom, negative control) did not receive any further instruction. Group two (simulation, exposure group) used virtual simulation of farrowing management for two hours. Group three (hands-on, positive control) had a half-day expedition to a farm with hands-on instruction. The farm visits had the objective to introduce students to see and try tasks in a farrowing unit. Teachers encourage students to train the skills on the visit, but this was not mandatory for the students. One week after exposure, students were given a multiple-choice test and a self-efficacy questionnaire to test the hypothesis. To measure knowledge acquisition in farrowing management amongst the groups, a multiple-choice test was conducted. The book “Developing and Validating Multiple-choice Items”  was used to construct the test. The test was a 20-question multiple-choice test with three possible answers per question. Two full-time swine veterinarians were chosen to peer-review the questions. Furthermore, 16 students from the Agriculture College with experience from swine production validated the test. To reduce guesswork, the scoring in the test awarded 2 points for a correct answer, − 1 points for an incorrect answer and 0 points for no answer. The total score from all 20 questions was summarized in a final score as a continuous variable. If the respondent had given two answers for one question it was registered as incorrect. A questionnaire was constructed to measure students’ perceived self-efficacy. Perceived self-efficacy evaluation is a method to measure how students judge their own competencies. The methodology relies on the theory of Albert Bandura. To summarize, he claims that there is an association between the level of perceived self-efficacy and the performance of a certain skill . Students were given statements about specific skills in farrowing management. They were asked to rate each statement on a scale from zero to seven. Zero was “cannot do at all” and seven was “highly confident can do.” Finally, the simulation group completed a satisfaction survey after the gaming session.
Danish agriculture colleges (17 in Denmark) are independent private institutions. Most of the Danish agriculture colleges (13 represented) are organized under “Danske Landbrugsskoler” . This organization takes care of the educational policy interests of agricultural schools, and strengthens networking and cooperation between schools in the educational, academic, and institutional areas. However, teachers in swine husbandry are also organized under the so-called “Hyo-academy.” The Danish Agriculture & Food Council  facilitates this academy with 17 agricultural colleges represented. Four schools from the “Hyo-academy” volunteered to participate in this study. The study was conducted from January 2017 to March 2017. During this period, new students began the 20 weeks “Basic Course Two” in general husbandry at all colleges. All schools follow the same program over the year. For the majority of students, this was their first experience with farming animals and agriculture in general.
Multiple choice test scores were used to calculate the sample size. The difference in means was set to two points and the standard deviation was set to four. Using a confidence level of 95% and power of 80% the necessary size of each group was calculated to 63 students, giving a total of 189 students in the study. The inclusion criterion for the colleges was a minimum of 21 students enrolled in the class. The college should also be able to facilitate partitioning of students into three groups throughout the study period. Partitioning was done with respect to the existing curriculum, in collaboration with the course coordinator from each school, to maximize randomization of the groups. In two of the schools, students were partitioned into three independent classes. In the two other schools, the students were partitioned within the class. Each college was requested to facilitate classroom instruction and hands-on instruction. The investigator administered the virtual simulation experience at all schools. One week after exposure, all students were tested with the multiple-choice test and the self-efficacy questionnaire. All answers were obtained with all students gathered in one room. Oral instruction was given in how to fill out the questionnaire. A written description was also provided. The questionnaire was administered in paper forms. Students returned the questionnaire after answering and left the room. The questionnaire was constructed in three modules. The first module determined demographical information by means of closed questions: gender, age, experience in swine production, and gaming behaviour. Finally, students were asked in an open question to state their favourite digital game. The second module was the multiple-choice test, and the third module was the perceived self-efficacy evaluation.
The data was anonymized and entered into Microsoft® Excel® for Mac 2011 and exported into statistical software R-studio version 1.1.383. A descriptive analysis was initially performed using different plot and summary statistics. The summarized average test score from the multiple-choice test was tested for statistical differences between study-groups using a univariable ANOVA-test. Differences between standard deviation (SD) were tested using an F test. Next, a multivariable model was constructed with the outcome test score and tested against all predictive variables (school, study group, age, sex, experience, background, play behaviour, farm simulator). Backwards elimination was performed using Akaike’s Information Criterion (AIC) test. Each skill in the perceived self-efficacy questionnaire was tested for statistical difference between study groups using the Kruskal-Wallis test. This test has the limitation of being unable to distinguish between groups that differ. Therefore, if statistical differences were found, the Wilcoxon signed-rank test was used to identify which groups differed from others. If respondents had marked two numbers in the self-efficacy questionnaire, the score was rounded down to the closest whole number. The level of significance was set to 95% in all analyses.