SoylentNews is people
SoylentNews
from the was-not-paying-attention-to-begin-with dept.
A small study into electronic device usage during lectures found that there was minimal difference in scores between those who were distracted while listening to the lecture and those who weren't when there was a quiz afterwards.
Results. The sample was comprised of 26 students. Of these, 17 were distracted in some form (either checking email, sending email, checking Facebook, or sending texts). The overall mean score on the test was 9.85 (9.53 for distracted students and 10.44 for non-distracted students). There were no significant differences in test scores between distracted and non-distracted students (p = 0.652). Gender and types of distractions were not significantly associated with test scores (p > 0.05). All students believed that they understood all the important points from the lecture.
Conclusions. Every class member felt that they acquired the important learning points during the lecture. Those who were distracted by electronic devices during the lecture performed similarly to those who were not. However, results should be interpreted with caution as this study was a small quasi-experimental design and further research should examine the influence of different types of distraction on different types of learning.
(Score: 2) by melikamp on Wednesday September 17 2014, @04:16PM
I wouldn't say that sample size 26 is "nothing", but it looks like investigators made some choices that rendered the study effectively meanigless. First, there really are 2 samples here: one of size 17, and the other of size 9, and 9 is a really small sample size. Second, a quick search of TFA fails to bring up the word "population", and the sampling process is not described, so it looks like it wasn't really a sample at all (samples are taken out of populations they are supposed to represent). In other words, this was properly a survey of the population of size 26, and no conclusion of this study can apply to any other student population. Lastly, what about people being distracted by someone else surfing? A more interesting result could be obtained by splitting 26 people into 2 groups of 13 randomly, and then giving them the same lecture, with one group being allowed to surf, and the other one forbidden. It would be a different kind of conclusion qualitatively, but at least it would be meaningful.
(Score: 2) by TGV on Wednesday September 17 2014, @06:35PM
Actually, if all of the population was there, it is a fact that the distracted students scored worse. It doesn't generalize, though.
Anyway, 26 data points is really nothing in this kind of test. The variance is too high. Suppose you want to distinguish a false coin from a true coin (the false one with probability p for heads, and the true one with probability 0.5). With 26 drawings, the 95% interval (leaving both tails at 5%) is from 8 to 17. So to have a 95% chance (a priori) that your false coin throws less than 8 (or more than 17) heads in 26 samples, it would need to be something like 0.2 (or 0.8). If it's inside the range 0.2-0.8, you don't have enough samples to be relatively sure to find a difference before starting the experiment.
In psycholinguistic experiments of this nature, the number of subjects there is usually around 30, and that's not considered high. Each subject usually does 10 to 20 (or more) samples in the same condition, and hopefully across all conditions. In this case, that would mean at least 30x10x2 = 600 samples instead of 26.