Part 4 — GED ® Social Studies · Unit 4 — Citizenship and Civic Participation
Chapter 19 — Public Opinion, Polls, and Misleading Displays
GED® test targets: CG.e.3; SSP.7.a, SSP.10.a, SSP.5.b
Suppose a fictional survey asks only people already attending an education class whether more classes are needed. Their answers are worth hearing, but they may differ from people who cannot attend or chose not to. The problem is not that the people answering are dishonest. The question is whether they stand in for the whole group the conclusion talks about. A thousand answers from the wrong group do not magically become everybody’s opinion.
Who was counted matters
A population is the whole group a study seeks to describe. A sample is the group actually observed. A representative sample reflects the relevant population well enough for the intended estimate. A large sample can still be biased if it systematically misses important groups. Ten thousand responses from a campaign’s supporters still do not represent all voters.
Question wording, nonresponse, timing, and survey method can influence results. A margin of error describes specified sampling uncertainty under a survey’s assumptions; it does not repair every form of bias. A poll is an estimate at a time, not a guarantee of election results.
Percent and percentage points answer different questions. If support rises from 40 percent to 50 percent, the change is 10 percentage points. Relative to the original 40, the increase is 10 ÷ 40 × 100 = 25 percent. Just saying “up 10 percent” can blur that difference.
Completed graph reading
The following original bar chart uses fictional survey results. Both surveys use the same stated population and method. No margin of error is supplied.

Figure 1 — Support for a fictional transit proposal
The values are 40 percent in Survey 1 and 50 percent in Survey 2. The vertical axis begins at zero, so bar heights show the proportions without exaggerating their ratio. A second chart beginning at 39 percent could make the one-point visible height for 40 look tiny beside the eleven-point visible height for 50. The underlying difference would still be 10 percentage points.
Question: Which conclusion is justified by the chart alone? A. Every resident now supports the proposal. B. Measured support rose by 10 percentage points. C. A campaign advertisement caused the rise. D. The proposal will certainly pass.
Answer: B. A contradicts the 50 percent result. C claims causation without a design that establishes it. D turns a poll into a certainty. Because uncertainty information is missing, the chart alone also does not establish whether the difference is statistically distinguishable from sampling variation.
Facts, opinions, and reasoned judgments
“The chart reports 50 percent” is a checkable factual statement about the display. “The proposal is wonderful” is an opinion expressing evaluation. “Support appears higher in the later survey, but the evidence does not explain why” is a reasoned judgment: an interpretation tied to evidence and its limits.
Do not label every judgment “just opinion” to avoid examining its support. Some judgments are carefully grounded and others are speculation. Ask what evidence would back each one up, weaken it, or tell them apart. That is how you keep a pretty chart from selling you a load of nonsense.
Second Look — A Big Sample Can Still Be a Biased-Ass Sample
The population is the group you want to describe; the sample is the group actually observed. A sampling frame is the set from which participants can be selected. Selection bias occurs when the method systematically favors some kinds of participants. Nonresponse can create another difference if those who answer differ from those who do not. Bigger samples cut down some random error, but they do not fix a built-in bias.
Question wording also matters. “Should the city improve unsafe parks?” bundles safety and approval in a way that differs from a neutral question about a specified spending proposal. Timing, response options, and order can influence answers. Compare like questions before calling a difference a change in opinion. Otherwise a change in wording may be doing the work while the headline claims society had a revelation.
A reported margin of error describes a particular kind of sampling uncertainty under assumptions; it does not cover every flaw. A poll at 51% versus another at 49% may not establish a clear change when their uncertainty ranges are considered. Read the supplied interval and method. Do not act certain just because one number crossed the 50% line that feels so important.
On a GED chart or graph, check the axes, the denominator, and who was sampled. A rise from 40% to 50% is 10 percentage points or a 25% relative increase, not both called “10 percent.” An axis that does not start at zero can make the change look bigger without changing the math. The graph is allowed to be attractive; it is not allowed to think for you. Read the damn labels.
Your Turn — Guided and Independent Practice — Practice Check SS19
Guided: A poll of 1,000 volunteers on a partisan website reports 92% support for a proposal. Identify two threats to representativeness.
Guided answer: Voluntary response lets highly motivated people self-select, and a partisan website reaches a narrow audience. Sample size does not cure selection bias. Ask who could participate, who did, wording, timing, and margin of error.
Independent: A graph shows approval rising from 48% to 50% but starts the axis at 47%. Which statement is best? A. Approval rose 2 percentage points, though the display exaggerates the visual jump. B. Approval doubled. C. The graph proves fraud. D. Scale never matters.
Independent answer: A. B misreads the values, C makes an unsupported accusation, and D ignores graphical design. Common mistake: letting a dramatic bar do the thinking.