Statistics for Dummies (42 page)
The
p
-value is the percentage chance of being at or beyond (in this case to the right of) 1.5, which is 100%
−
93.32% = 6.68%, which is written as a probability as 0.0668. (See
Table 8-1
of
Chapter 8
.) This
p
-value is just a shade over 0.05, so, technically, you don't have quite enough evidence to reject H
o
. That means vomiting is not experienced any more by those taking this drug when compared to a placebo (although this result is one a statistician would call
marginal
).
| HEADS UP | A |
| HEADS UP | Most hypothesis tests comparing two separate population proportions are done using samples that are quite large, given that they're most often based on surveys, so you won't likely run into a case that uses small samples. |
Statistical Studies—The Inside Scoop
Chapter List
- Chapter 16:
Polls, Polls, and More Polls - Chapter 17:
Experiments—Medical Breakthroughs or Misleading Results? - Chapter 18:
Looking for Links—Correlations and Associations - Chapter 19:
Statistics and Toothpaste—Quality Control
Many statistics that you hear and see each day are based on the results of surveys, experiments, and observational studies. Unfortunately, you can't believe everything you read or hear.
In this part, you look at what actually happens behind the scenes of these studies: how they are designed and conducted; how the data is (supposed to be) collected; and how to spot misleading results.
Polls, Polls, and More Polls
Surveys seem to be all the rage in today's information explosion. Everyone wants to know how the public feels about issues, from prescription drug prices and methods of disciplining children to approval ratings of the president and reality TV. Polls and surveys are a big part of American life; they're a vehicle for quickly getting information about how you feel, what you think, and how you live your life, and are a means of quickly disseminating information about important issues. Surveys are used to highlight controversial topics, raise awareness, make political points, stress the importance of an issue, and educate or persuade the public.
Survey results can be powerful, because when many people hear that "such and such percentage of the American people do this or that", they accept these results as the truth, and then make decisions and form opinions based on that information. In fact, many surveys don't provide correct, complete, or even fair and balanced information. In this chapter, I discuss the impact of surveys and how they're used, and I take you behind the scenes of how surveys are designed and conducted so that you know what to watch for when examining survey results. I also talk about how to interpret survey results and how to spot biased and inaccurate information, so that you can determine for yourself which results to believe and which to ignore.
A survey is an instrument that collects data through questions and answers and is used to gather information about the opinions, behaviors, demographics, lifestyles, and other reportable characteristics of the population of interest. What's the difference between a poll and a survey? Statisticians don't
make a clear distinction between the two, but I've noticed that what people call a poll is typically a short survey containing only a few questions (maybe that's how researchers get more people to respond — they call it a poll rather than a survey!). But for all intents and purposes, surveys and polls are the same thing.
You come into contact with surveys and their results on a daily basis. Surveys even have their own television program: The game show
Family Feud
is completely based on surveys and the ability of the contestants to list the top answers that people provided on a survey. Contestants on this show must correctly identify the answers provided by respondents to survey questions such as, "Name an animal you may see at the zoo" or "Name a famous person named John."
Compared to other types of studies, such as medical experiments, surveys are relatively easy to conduct and aren't as expensive to carry out. They provide quick results that can often make interesting headlines in newspapers or eyecatching stories in magazines. People connect with surveys because they feel that survey results represent the opinions of people just like themselves (even though they may never have been asked to participate in a survey). And many people enjoy seeing how other people feel, what they do, where they go, and what they care about. Looking at survey results makes people feel connected with a bigger group, somehow. That's what
pollsters
(the people who conduct surveys) bank on, and that's why they spend so much time doing surveys and polls and reporting the results of this research.
Who conducts surveys these days? Pretty much anyone and everyone who has a question to ask. Some of the groups that conduct polls and report the results include:
News organizations (for example, ABC News, CNN, Reuters)
Political parties (those in office and those trying to get into office)
Professional polling organizations (such as The Gallup Organization, The Harris Poll, Zogby International, and so on)
Representatives of magazines, TV shows, and radio programs
Professional organizations (such as the American Medical Association, which often conducts surveys of its membership)
Special-interest groups (such as the National Rifle Association)
Academic researchers (who conduct studies on a huge range of topics)
The U.S. government (which conducts the American Community Survey, the Crime Victimization Survey, and numerous other surveys through the Census Bureau)
Joe Public (who can easily conduct his own survey on the Internet)
| HEADS UP | Not everyone who conducts a poll is legitimate and trustworthy, so be sure to check the source of any survey in which you're asked to participate and for which you're given results. Groups that have a special interest in the results should either hire an independent organization to conduct (or at least to review) the survey, or they should offer copies of the survey questions to the public. Groups should also discuss in detail how the survey was designed and conducted, so that you can make an informed decision about the credibility of the results. |
The topics of many surveys are driven by current events, issues, and areas of interest; after all, timeliness and relevance to the public are two of the most attractive qualities of any survey. Here are just a few examples of some of the subjects being brought to the surface by today's surveys, along with some of the results being reported:
Does celebrity activism influence the political opinions of the American public? (Over 90% of the American public says no, according to CBS News.)
What percentage of Americans have dated someone online? (Only 6% of unmarried Internet users, according to CBS News.)
Is pain something that lots of Americans have to deal with? (According to CBS News, three-quarters of people under 50 suffer pain often or at least some of the time.)
How many people surf the Web to find health-related information? (About 98 million, according to The Harris Poll.)
What's the current level of investor optimism? (According to a survey by The Gallup Organization, it should be called investor pessimism.)
What was the worst car of the millennium? (The Yugo, according to listeners of the NPR radio show
Car Talk.
)
| HEADS UP | When you read the preceding survey results, do you find yourself thinking about what the results mean to you, rather than first asking yourself whether the results are valid? Some of the preceding survey results are more valid and accurate than others, and you should think about whether to believe the results first, before accepting them without question. |
 |
You may be familiar with a radio show called
Car Talk
that's typically aired Saturday mornings on National Public Radio, and is hosted by Click and Clack, two brothers who offer wise and wacky advice to callers with strange car problems. The show's Web site regularly offers surveys on a wide range of car-related topics, such as, "Who has bumper stickers on their cars, and what do they say?" One of their recent surveys asked the question, "What do you think was the worst car of the millennium?" Thousands upon thousands of folks responded with their votes, but, of course, these folks don't represent all car owners. They represent only those who listen to the radio show, logged on to the Web site, and answered the survey question.
Just so you won't be left hanging (and I know you're dying to find out!) the results of the survey are shown in the following table. Before you look at what others reported, you may want to cast your own vote! (Remember, though, that these results represent only the opinions of
Car Talk
fans who took the time to get to the Web site and take the survey.) Notice that the percentages won't add up to 100% because the results in the table represent only the top ten vote-getters.
Rank | Type of Car | Percentage of Votes |
|---|---|---|
1 | Yugo | 33.7% |
2 | Chevy Vega | 15.8% |
3 | Ford Pinto | 12.6% |
4 | AMC Gremlin | 8.5% |
5 | Chevy Chevette | 7.0% |
6 | Renault LeCar | 4.3% |
7 | Dodge Aspen / Plymouth Volare | 4.1% |
8 | Cadillac Cimarron | 4.0% |
9 | Renault Dauphine | 3.6% |
10 | Volkswagen (VW) Bus | 2.7% |
|
Whereas some surveys are fun to look at and think about, other surveys can have a direct impact on your life or your workplace. These life-decision surveys need to be closely scrutinized before action is taken or important decisions are made. Surveys at this level can cause politicians to change or
create new laws, motivate researchers to work on the latest problems, encourage manufacturers to invent new products or change business policies and practices, and influence people's behavior and ways of thinking. The following are some examples of recent survey results that can impact you:
Teens drive under the influence:
A recent Reuters survey of 1,119 teenagers in Ontario, Canada, from grades 7 through 13 found that, at some point during the previous year, 15% of them had driven a car after consuming at least two drinks.Children's health care suffers:
A survey of 400 pediatricians by the Children's National Medical Center in Washington, D.C., reported that pediatricians spend, on average, only 8 to 12 minutes with each patient.Crimes go unreported:
According to the U.S. Bureau of Justice 2001 Crime Victimization Survey, only 49.4% of violent crimes were reported to police. The reasons victims gave for not reporting crimes to the police are listed in
Table 16-1
.Table 16-1:
Reasons Victims Didn't Report Violent Crimes to the PoliceReason For Not Reporting
Percentage
Considered it to be a personal matter
19.2
The offender was not successful/didn't complete the crime
15.9
Reported the crime to another official
14.7
Didn't consider the crime to be important enough
5.5
Didn't think police would want to be bothered
5.3
Lack of proof
5.0
Fear of reprisal
4.6
Too inconvenient/time consuming to report it
3.9
Thought police would be biased/ineffective
2.7
Property stolen had no ID number
0.5
Not aware that a crime occurred until later
0.4
Other reasons
22.3
The most frequently given reason for not reporting a violent crime to the police was that the victim considered it to be a personal matter (19.2%). Note that almost 12% of the reasons relate to perception of the reporting process itself (for example, that it would take too much time or that the police would be bothered, biased, or ineffective).
Breast cancer treatment challenged:
Which treatment should a woman with breast cancer choose, lumpectomy (where the tumor is removed but most of the breast is not) or mastectomy (where the entire breast is removed)? The most popular belief has been that a lumpectomy is the best option for most women. However, in a recent survey asking surgeons what they would opt for if they themselves had early stage breast cancer, 50% would prefer a mastectomy.Cellphones and driving create hazard:
A recent
Consumer Reports
survey of new car buyers and leasers looked at the driving-related issues that concerned these consumers the most. The survey found that the driving-related issue that concerned the largest percentage of respondents (53%) was drivers who were distracted by the use of cellphones. This beat out high gas prices (of concern to 50% of the respondents) and road rage (which was a concern for 38% of the respondents).Cyber crime takes a toll on business:
The Computer Security Institute (CSI) conducted a recent survey of U.S. corporations to assess the extent to which cyber crime has affected their businesses. Ninety percent of respondents detected computer security breaches within the last year, and 80% of those acknowledged financial losses due to these breaches. Seventy-eight percent of respondents reported employee abuse of Internet-access privileges (for example, viewing pornography, downloading pirated software, and abusing e-mail privileges).Sexual harassment presents workplace problem:
A recent survey of the workplace conducted by the Equal Employment Opportunity Commission found that between 40% and 70% of women and 10% to 20% of men reported that they had been victims of sexual harassment in the workplace. In recent years, the number of complaints filed by men has tripled.
| HEADS UP | The preceding examples address some very important issues, but you have to decide whether you can believe in or act on those results. You have to be able to sort out what's credible and reliable from what isn't. Rule number one, don't automatically believe everything you read! |
