But the y-axis does not start from 0 is very normal, even many videos from Khan AP Gov use them

Only if 0 is a relevant, or nearby data point should it be included in the scale. Otherwise it's just excess space. If you properly label the axis, the necessary information is there. If you assume the x-axis is always at 0 value, then you are making a faulty assumption.

Main content

Course: AP®︎/College US Government and Politics > Unit 4

Lesson 6: Evaluating public opinion data

Evaluating public opinion data: lesson summary

Google Classroom

Public opinion polls can only tell us so much about the outcome of elections and public policy.

Common problems with polls

Problem	Why it's misleading
improper sampling techniques	Samples should be random, otherwise poll results might not accurately represent the target population.
biased questions	It’s tough for questions to be truly unbiased, but questions framed to portray candidates or policies in a positive or negative light can strongly influence responses.
small sample size	The smaller the sample size, the larger the margin of error. A survey of ten people is unlikely to turn up results consistent with the preferences of the entire population of the United States. Mass surveys usually have a sample size of at least 1000 individuals.
large or unreported sampling error	The results from a sample won’t exactly represent a population. This is called sampling error, and the likely size of the error is called the margin of error. Researchers should report their poll’s margin of error. If a poll reports that 58% of Americans prefer a candidate, but the margin of error is 10%, researchers should not conclude that the majority of Americans prefer the candidate (since the true percentage could be as little as 48%).
lack of transparency in methods	Reliable polls publish the methods by which researchers collected the data. Without this information, it’s impossible to judge whether the researchers followed scientific procedures.

How much does public opinion data really tell us?

The media breathlessly reports the results of polls. During a presidential election, news outlets broadcast the result of opinion polls on a daily basis, showing one candidate ahead on Monday and the other candidate ahead on Tuesday. Did the public’s opinion really shift that much in just 24 hours?

Even more perplexing are notable differences between public opinion and the actions of government officials. Generally, politicians hoping for reelection want to please their constituents, so it seems logical that they would align their policies with the results of public opinion polls. But there are plenty of examples of mismatches between public opinion and public policy. For example, polls consistently reveal that the majority of Americans support stricter gun control measures, but few elected officials have proposed legislation restricting firearms. What accounts for this difference between opinion and action?

Public opinion polls can only tell us so much about the outcome of elections and public policy. In some cases, this is because the data itself was not reliable, or was presented in a misleading fashion. In other cases, the disconnect results from situations where public opinion was not the driving factor behind a political opinion.

Review questions: know your graphs

News outlets, campaigns, and corporations frequently publish graphs to illustrate the results of public opinion polls. But graphs can be misleading, even if the information they contain is technically correct!

1. Here's an unscientific graph. Can you identify two problems with it?

Why it’s unscientific: We can see that the amount of cheeseburgers consumed changes, but we don’t know whether the y-axis depicts the number of cheeseburgers or cheeseburgers as a percentage of the school’s diet. The x-axis does not specify the timespan over which cheeseburger consumption was measured. Is the number changing by the hour, day, week, month, or year? The title of the graph isn’t very descriptive, either. Who is eating all these cheeseburgers? Students, faculty, or three very hungry linebackers from each football team?

Problems with this graph:

This graph needs labels for its axes. A graph should have a clear label for both its x-axis (horizontal scale) and y-axis (vertical scale).
This graph needs a descriptive title. A graph should have a title that helps you understand what it depicts.

A better version of this graph:

Why it's better: This graph tells us who was eating the cheeseburgers and the time period over which they were eating them. Looks like Fulton High’s seniors are the real cheeseburger heroes.

2. Here's a misleading graph. Can you identify two ways it's misleading?

Why it’s misleading: A graph should have consistent measurements on both of its axes. In this graph, the vertical axis is truncated (shortened). It starts at 5 instead of 0. This means that the scale exaggerates differences in measurements. For example, how many more giraffes escaped from Porous Zoo in 2017 compared to 2016? The difference in height of the bars makes it seem as though 5 times as many giraffes escaped in 2017—which would be 30 giraffes, not 10! In fact, only 4 more giraffes escaped in 2017 than in 2016.

This graph also compares one zoo (Porous Zoo) to all other zoos worldwide. A casual glance would suggest that Porous Zoo is doing a better job keeping its giraffes contained than its competitors, but it’s not fair to compare one zoo to the total of all other zoos combined. A more accurate graph would compare Porous Zoo to one other zoo, or to the average number of giraffes that escaped from zoos each year.

Problems with this graph:

This graph needs a consistent scale. The truncated y-axis exaggerates small differences.
This graph compares one institution to all other institutions combined. This does not allow for an accurate comparison between the two sets of data presented.

A better version of this graph:

Why it's better: The y-axis has a consistent scale that starts at zero. This allows the viewer to see that 40% more giraffes escaped from Porous Zoo in 2017 than in each of the previous years, rather than implying that 400% more giraffes escaped. This graph also compares one zoo (Porous Zoo) to one other zoo (Adamant Zoo, which we will assume is a zoo of comparable size and resources).

Want to join the conversation?

Sort by:

Wenqi
Posted 3 years ago. Direct link to Wenqi's post “So if you have to start t...”
So if you have to start the y axis with a gap (like if the data is 2000, 2030, 2050; you will not want to make it start from 0 then 10,20,30 ...), you should make a special mark?
Button navigates to signup pageButton navigates to signup page
(2 votes)
Answer
Wenqi
Posted 3 years ago. Direct link to Wenqi's post “But the y-axis does not s...”
But the y-axis does not start from 0 is very normal, even many videos from Khan AP Gov use them
Button navigates to signup pageButton navigates to signup page
(0 votes)
Answer
- Davin V Jones
  Posted 3 years ago. Direct link to Davin V Jones's post “Only if 0 is a relevant, ...”
  Only if 0 is a relevant, or nearby data point should it be included in the scale. Otherwise it's just excess space. If you properly label the axis, the necessary information is there. If you assume the x-axis is always at 0 value, then you are making a faulty assumption.
  Button navigates to signup page
  (4 votes)