6.24 Assignment 3 - A Survey with Bias

6.24 Assignment 3 - A Survey with Bias
CREATE A SURVEY WITH BIAS
Create and a questionnaire and conduct a survey bias to create a favorable result that would be difficult to conclude
without bias.
Using online questionnaires will be looked on favorably.
www.surveymonkey.com
www.zoomerang.com
http://www.freeonlinesurveys.com/
Decide on an issue that people have different opinions on and write a biased questionnaire and conduct a survey so
that you can prove that people generally agree with what you want them agree with.
For example, if I wanted to create a survey that showed Americans love British Culture I might ask questions like this:
Who do you think most influenced modern rock music culture?
a. Led Zepplin
b. The Beatles
c. Pink Floyd
d. Bon Jovi
e. Kiss
Alternatively:
Considering the influence and leadership Winston Churchill showed during World War II, would you consider him to
be the greatest person of the 20th Century?
a. Yes
b. No
Choose a topic that is opinion based rather than something that cannot be measured. For example, it would not be a
good idea to show that ‘House’ is the most popular show on television because viewing figures quantify that it is not.
DO NOT choose a topic that is too controversial, that might upset subjects or your ‘sensitive’ Statistics Teacher.
Criteria:
1. Make the questionnaire 8 – 10 questions long
2. Ask a minimum of 10 subjects
3. Consider who you send your survey to (and how well you know them)?
4. Create your bias through the wording of the questions and sampling technique. . .
To be collected
1. Explain what your aim was
2. A copy of your questionnaire and data analysis of your survey – this can be simple: just percentage breakdown
for the results of each question. Visualizations are always nice.
3. Review each question; why was each question created and how did each question help you achieve you aim?
4. Critique your work – in light of your results, did you achieve your aim, how did your questions and/or your
interviewing technique assist you (or not)?
5. Use your class notes to give statistical context (consider survey method, bias and error).
6. What would you improve next time to achieve your aim?
IMPORTANT NOTE: I will not grade you on how biased your questions were or how well you achieved your aim. Your
write-up is where the grade is. The class notes should be the starting point, not afterthought. Simple studies (e.g.
convenience sample, collected verbally with selection bias) will struggle to receive top grades.
6. Experimental Design I Page 1
Cat Newman
A Survey with Bias
The purpose of this experiment was to view student opinion on the problem's involved in
destructive decisions and what should be done to prevent these problems. The survey was purposefully
biased in an attempt to get students to agree to unusually strict and extensive prevention of these .
Here is the survey and the responses.
Question 1:
This question started off the survey to a biased note by presenting a startling statistic before
asking a yes or no question. Most high school students hate classes on alcohol because students who
don’t drink feel it is a waste of time and students who do drink don’t want to be changed. However, 8 of
the 11 responders to the survey said yes, which means that either the wording of the question
persuaded the subjects from answering differently, or the survey sample was already inclined to answer
this way. Three of the subjects said no though, which means that the question is not as biased as it
could be. Since answering no was an option, the subjects, especially ones who had strong opinions,
could ignore the biased question.
Question 2:
This question was similar to question one, in that it used a statistic that was shocking to change
the responses. It also served to extend on the first question, and to have the subjects agree to an even
more. However, the difference was that this question had a more surprising statistic. Also, it did not
affect the subjects of the experiment, but younger students. The results were that all 11 subjects
agreed that it was necessary to teach younger students. More people agreed with this one than the first
one, most likely because they went into the question with less of an opinion and were thus more open
to being swayed. Also, a more distressing statistic would do a better job of changing opinions.
Question 3:
The purpose of this question was too make students agree to keep the drinking age. This
question, which was extremely biased, was almost accusatory and made it hard for the subjects to say
yes. The results, with only one person saying yes, shows that either almost all the subjects already liked
having a drinking age or they were persuaded to say no by the question. Making it hard for the subjects
to give a certain answer thus seems to be effective, although as questions later will show it is more
effective to eliminate entirely an answer you don’t want.
Question 4:
This question had a very interesting result. The purpose of this question was to extend on the
previous question and to get subjects to agree that the drinking age was not only needed, but it should
remain at this age or even be raised. The question was not as biased as others though; it only gave
negative language. This caused a greater variety in answers, and even some answers that didn’t follow
along with the purpose of the question. With a question that is less biased, the effects of biased
questions and limited answers are demonstrated.
Question 5:
Question 5 had some really unexpected results. The question itself was relatively less
influential. The answers however, were more limited and forced the subjects to agree to a harsher
punishment no matter what. However, what is surprising is that most of the subjects chose the most
extreme answer. Either the subjects came into the survey with strong opinions, or the fact that this
question had three similar answers and one more extreme answer caused the subjects to agree to the
more extreme answer, since they didn’t see a difference in the other ones.
Question 6:
This question proved to the most effective way to skew results in the data: not making an
answer choice possible. No matter what, the subjects had to agree to motor vehicle fatalities being the
greatest factor that affected teenagers. Without other options, subject’s opinions and the wording of
the question does not matter. Another effect of this question is that it is impossible to get even a slight
insight into the opinions of the subjects.
Question 7:
This question is directed at the guys surveyed, because they would not feel that they should
have extra driving school. Though this sample of subjects was even, a skewed sample could affect the
data. In this question the sample doesn’t matter though, because like question 6 there is no option but
to say that men need more time being educated about destructive decisions. It builds upon question 6
though, because like question 5, the answers tended towards the extremes. It seems that either a
subject is biased and will pick an extreme answer, or a subject picks a more extreme, different answer
when they are not sure.
Question 8:
This was another surprising question. The purpose was to try to make students agree to wear
seat belts as often as possible. However, the subjects were still given the option of saying that it was
okay to not wear a seatbelt while in the passenger’s seat. Though a statistic was given in the question
to create bias, the extreme results of this question beg for more of an explanation. This is especially
true since the second and third answer are basically the same, so most would expect to see a slight
variety. The reason for the result must be a sampling error, because all the subjects seemed to have
come into the survey with biased opinions on this question.
Question 9:
This final served to wrap up the survey. The question itself is not biased, but the answers force
the reader to agree to at least some extra time being spent on spreading awareness. What is surprising
is that a majority of those surveyed said they wanted as much time as possible. Part of this might be
due to the fact that, after answering so many biased questions, their opinions actually changed.
However, another possibility is that, if they couldn’t find an answer they liked, they picked a more
extreme answer that seemed different from the other answers.
Most of the bias and error in this experiment comes from the response bias. The questions
either convinced the subjects of an answer, or they forced the subjects to answer a certain way. Some
sampling method error occurred too, probably due to the small sample size and lack of randomization.
What was interesting to see was that biased questions were not always able to completely eliminate the
opinions of the subjects, but a small range of answers would of course always force the subjects to
answer a certain way. Looking at different survey methods that create errors allows us to see what
creates error, and avoid it.