Reliability, Validity, and Generalizability of Results: Is this possible?
Social media is becoming a more and more vital role in our society to make connections between our friends and family even when we are miles away from those people. Many new studies are now emerging that are attempting to find out the amount of time being spent online and what the main uses of social media are. I was reading a study (College Students Social Networking Experiences on Facebook, Pempek et. Al, 2009) about this idea. As I was reading this study, I questioned the validity and reliability of the methods when it came to where the information was gathered. The researchers gathered a group of college individuals who were different levels of education and different ethnic backgrounds and made them complete questionnaires about their usage of the site. My main query comes from the selection of the sample and whether or not it is an actual representation of the amount of time the average person uses social media. The results of this study are probably very accurate for the general population of university students but can they be generalized across populations of non-students? My guess is probably no. I personally believe that being able to apply the results of one study to an entire population (even with random sampling) is impossible on account of the diversity of people in the world. Even though the sample in this study appears to be reliable at face value, when we look deeper into the study it leaves out a vast majority of social networking users who are not in school. Would this have a huge change on the outcome of the study? It is tough to tell but it may have a massive effect (positively or negatively) on the results of the research by leaving them out.
This idea raises the question if we can ever have a truly reliable sample population for any study that will give an accurate representation of a general population as a whole. I am interested in hearing all of your opinions on this subject. What are your takes on whether or not we can ever achieve an unbiased and equal sample size that will allow for the generalization of the results to span over a population. What ways have you seen this be attempted and what are your thoughts on the effectiveness of these methods used?