06.02.2013 Views

Research in Engineering Education Symposium 2011 - rees2009

Research in Engineering Education Symposium 2011 - rees2009

Research in Engineering Education Symposium 2011 - rees2009

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Universidad Politécnica de Madrid (UPM) Pág<strong>in</strong>a 309 de 957<br />

Tak<strong>in</strong>g <strong>in</strong>to account the previous analysis for the marks, it can be said that there were<br />

some factors or hidden variables that were affect<strong>in</strong>g significantly the academic results of<br />

the students and orig<strong>in</strong>at<strong>in</strong>g heterogeneity. That is to say, there were some factors that<br />

were produc<strong>in</strong>g a higher variability than the rest and, as a result, these factors were<br />

segment<strong>in</strong>g the population.<br />

Here, it is important to po<strong>in</strong>t out that for the case under analysis the only <strong>in</strong>formation<br />

available was the marks of AC, which was not enough to look for the qualitative factors<br />

that differentiated the students of the two groups. However, it was suspected that one<br />

important factor, among others, was the follow<strong>in</strong>g: how Mathematical tools are used <strong>in</strong> AC.<br />

The academic results <strong>in</strong> Math should have an <strong>in</strong>fluence <strong>in</strong> the ones <strong>in</strong> AC and could be one<br />

of the reasons for segment<strong>in</strong>g the population. Hence, <strong>in</strong> order to cont<strong>in</strong>ue improv<strong>in</strong>g the<br />

performance of the students <strong>in</strong> AC, it could be <strong>in</strong>terest<strong>in</strong>g to study whether the results <strong>in</strong><br />

Mathematics I (MATI) have <strong>in</strong>fluenced the ones <strong>in</strong> Analysis of Circuits I (ACI), and if the<br />

marks <strong>in</strong> MATI represent a factor that segment the mark <strong>in</strong> ACI.<br />

F<strong>in</strong>ite mixture of regression<br />

In the third stage, which is the current one, the marks and trajectories of the students that<br />

were major<strong>in</strong>g <strong>in</strong> Sound Systems (SM) and Telecommunication Systems (TM) <strong>in</strong> the<br />

subjects ACI and Math, have been collected (MATISM, ACISM, MATITM, ACITM) for six<br />

years and compared with each other. Instructors of both subjects (i.e., ACI and MATI) have<br />

worked together and new <strong>in</strong>terdiscipl<strong>in</strong>ary materials of study have been created<br />

[Hernandez (2010)].<br />

The study that has been carried out from the marks of the students has been the follow<strong>in</strong>g:<br />

First, from the dispersion diagrams shown <strong>in</strong> Fig.1 and the calculation of the correlation<br />

coefficient (0.5345 for SM and 0.4822 for TM), it is observed both that there exists a l<strong>in</strong>ear<br />

relation between the <strong>in</strong>dependent variable MATISM (variable x) and the dependent<br />

variable ACISM (variable y), and that there also exists a l<strong>in</strong>ear relation between the<br />

<strong>in</strong>dependent variable MATITM (variable x) and the dependent variable ACITM (variable<br />

y). Hence, a l<strong>in</strong>ear regression model could be adjusted to the data. Nevertheless, when<br />

observ<strong>in</strong>g Fig. 1 deeply it can be seen that the marks <strong>in</strong> ACI are not homogeneous among<br />

students with similar marks <strong>in</strong> MATI. Therefore, it could be suggested that there are<br />

several groups for which a l<strong>in</strong>ear regression model would represent a good<br />

approximation. That is to say, there exists a different l<strong>in</strong>ear relation for groups of students<br />

for the level, the slope and the variability [Justel (2001)].<br />

For the regression models it is assumed that the regression coefficients are the same for<br />

all the observations and it is also assumed that the sample (,)iiyx is a homogeneous group.<br />

In many cases, as it could be ours, the former assumption cannot be made if there are<br />

important variables that are not <strong>in</strong>cluded <strong>in</strong> the model; that is, there is non-observed<br />

heterogeneity.<br />

Proceed<strong>in</strong>gs of <strong>Research</strong> <strong>in</strong> Eng<strong>in</strong>eer<strong>in</strong>g <strong>Education</strong> <strong>Symposium</strong> <strong>2011</strong><br />

Madrid, 4 th - 7 th October <strong>2011</strong>

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!