The aim of this study is to do a bank of elements, which allows to apply and evaluate in a precise and balanced way cognitive aptitudes (concepts, reasoning, identification, analysis) specific to the domain of the subject of probability and statistics, of the civil engineering degree in a public university. Three professors were invited as experts in the study area to validate the content of the proposed articles. A 34-item questionnaire was administered to a total number of 167 students in their 2nd year of Civil Engineering where the responses were coded dichotomously. The analyzes carried out are the reliability and validity of the questionnaire, the unidimensionality of the scale was also verified through exploratory analysis, and the ability of the students and the difficulty of the item analyzed by the Rasch model were evaluated. Within the results obtained, the elements that did not comply with the discrimination and with the model were eliminated; remaining 10 items. The separation index for the items was 4.35 while their reliability was 0.95. Therefore, these values are considered to be adequate. On the other hand, the separation indices of 1.06 people and the reliability index of 0.53 people are not very adequate. With the parameters obtained, it is highlighted that teaching strategies should be improved. Therefore, these values were considered appropriate. Additionally, the separation index for people was 1.06 while the reliability index was 0.53. These values are not really appropriate. The parameters that were obtained underline the importance of improving learning strategies.Este trabajo presenta la construcción de un banco de ítems, que permita aplicar y evaluar de manera precisa y equilibrada aptitudes cognitivas (conceptos, razonamiento, identificación, análisis) específicas del dominio de la asignatura de probabilidad y estadística, de la licenciatura de ingeniería civil en una universidad pública. Se invitó a tres profesores como expertos del área de estudio para validar el contenido de los ítems propuestos. Se aplico un cuestionario de 34 ítems a una muestra de 167 alumnos de tercer semestre en donde las respuestas se codificaron dicotómicamente. Los análisis realizados son la confiabilidad y validez del cuestionario, también se verificó la unidimensionalidad de la escala por medio del análisis exploratorio, y se evaluó la habilidad de los alumnos y la dificultad del ítem analizado por el modelo de Rasch. Dentro de los resultados obtenidos, se eliminaron los ítems que no cumplían con la discriminación y con el modelo; permaneciendo 10 ítems. Se evaluó el índice de separación de los  ítems es 4.35 y la fiabilidad de los ´ítems 0.95; se considera que estos valores son adecuados. Por otro lado, los índices de separación de las personas 1.06 y el índice de fiabilidad de las personas 0.53, no son muy adecuados. Con los parámetros obtenidos se destaca que se deben mejorar las estrategias de enseñanza.