On Monday, November 4th, the UB Spectrum ran its annual Political Issue. In it, they featured a poll of issues important to UB students, a community made up of over 31,000 people. We were saddened to see that only 100 students participated in the poll; this sample is merely 0.3% of the total population whom the survey supposedly represents, meaning the margin of error is more than triple what most published pollsters allow for (+/-3%).
Further, most reliable surveys include information on who the survey actually represents. The article states they collected results from 100 students, but it is not clear whether they are extrapolating this data to be representative of the entire UB population, or just of the sampled students. Without this vital piece of information, the reader of the published poll has no context for what this data means and how to interpret it.
This small sample size is not representative of the UB population for a number of reasons: Number one is the sheer size of the university, as mentioned above.
Secondly, the article claims the sample is “random,” however it’s likely that only students who follow the Spectrum on social media or visited the website where the link to the survey was published would have been able to participate. Alternatively, they may have heard of it through word of mouth. This inherently creates a bias in the sample, and does not equally target all kinds of students on campus. Students who completed the poll therefore engaged in a type of self-selection, taking it upon themselves to participate, which is intrinsically unscientific.
For example, the sample disproportionately represents first-year students, ringing in at more than 30% of those polled (based on our visual interpretation of an unlabeled pie chart). In fact, first-year only make up about 13% of the university community. If an email was sent to all UB students giving them equal opportunity to participate, not only would the sample size be larger (therefore more representative of the population), but it would be more scientifically reliable, ridding itself of the aforementioned biases.
Another important issue we take with this poll was the framing of the only inherently political question included, which asks respondents to rank the importance of the issues listed. The issues included the following: Marijuana Legalization, Abortion, Gun Control, Climate Change, Health Care, and College Tuition and Debt.
The problem here is that the question is too ambiguous; “importance” in this question is not attributed to any subject. One might instead ask “Rate these issues in order of importance to you”, or “to the university community,” “to the state,” “to the country” or otherwise. The framing of this question can drastically change how the respondents answer the question, as well as how the readers of the paper interpret the meaning of these results.
In addition, we think it would have been interesting to include more questions about political opinions and views, rather than just demographics of the students, which by themselves are not telling. We know that in the actual poll there were questions about which party students belong to, and who they will vote for in the 2020 presidential election, but the Spectrum chose not to publish this information. These questions, as well as others regarding “hot topics,” would have been more informative and interesting for the reader to understand the political climate on campus.
The core reason for this letter is to share the value and the significant implications that polling can have. Not only does polling share the views of those sampled, but it can also have an effect on those observing the published results and what their views may be. We live in a polarized and fast changing society in which journalism and data science can have a great effect on whether or not correct information is universally understood.
For future surveys, we hope that the Spectrum takes into account proper polling and survey methods in order to more accurately collect and distribute data to the community.
Sincerely,
Jaycee Miller and Alexa Federice
EDITOR’S NOTE: The Spectrum is independent (not funded by the university) and we do not have access to a student-body email listserv. This survey was funded only by editors on staff and, due to financial restrictions, we were not able to access all survey results.
