Les informations fournies dans la section « Synopsis » peuvent faire référence à une autre édition de ce titre.
Acknowledgments,
Introduction: Critical Thinking,
1. Numeracy,
2. Sampling and Generalizability,
3. Probability and Coincidence,
4. Visual Thinking,
5. Correlation and Causation,
6. Scientific Thinking,
7. Fact, Opinion, and Logical Reasoning,
References,
Index,
Numeracy
Critical Thinking, Step One: What do the numbers and statistics tell you? When hearing or reading about research results, often the first things you notice are the data. Perhaps the initial step in critically thinking about any claims made is to understand the data and how accurate the numbers, averages, and percentages are. Learning to interpret data, read statistics, and make estimates of what may be reasonable about a set of numerical findings is a good place to begin. Understanding, representing, using, and analyzing numbers are important tools of critical thinking and numeracy.
When asked, most people claim they are better drivers or fairer or smarter than the average person. It is not atypical for people to overestimate their abilities: As Zuckerman and Jost state (2001: 209): "Desires for self-enhancement and positive self-presentation lead people to make self-serving comparisons between themselves and others. Thus most people believe that they are 'better than average' on a wide variety of traits, abilities, and outcomes." How is it possible for most of us all to be better than average? What makes this statement suspicious depends on knowing something about percentages, averages, and their mathematical properties. More important, it requires overcoming a block some people have when dealing with numbers — what has been termed "innumeracy" (Paulos, 1989).
Learning to interpret data and make estimates of what may be reasonable about a set of quantitative findings is essential to thinking critically. Just knowing how to perform basic mathematical functions is not the important thing; it's also crucial to develop a sense of understanding, representing, using, and analyzing numbers. Deciphering and interpreting numbers with confidence are central elements of "numeracy," "quantitative reasoning," or "mathematical literacy." Numeracy is a building block of critical thinking. The California Critical Thinking Skills Test-Numeracy (2016) defines it as
the ability to solve quantitative reasoning problems and to make well-reasoned judgments derived from quantitative information in a variety of contexts. More than being able to compute or calculate a solution to a mathematical equation, numeracy includes understanding how quantitative information is gathered, represented, and correctly interpreted using graphs, charts, tables and diagrams.
According to the Association of American Colleges & Universities (Rhodes, 2010b), quantitative literacy involves:
• interpretation (ability to explain information presented in mathematical forms — e.g., equations, graphs, diagrams, tables, words),
• representation (ability to convert relevant information into various mathematical forms — e.g., equations, graphs, diagrams, tables, words),
• calculation (successful and sufficiently comprehensive to solve the problem clearly),
• application/analysis (ability to make judgments and draw appropriate conclusions based on the quantitative analysis of data while recognizing the limits of this analysis),
• assumptions (ability to make and evaluate important assumptions in estimation, modeling, and data analysis), and
• communication (expressing quantitative evidence in support of the argument or purpose of the work).
An element of quantitative reasoning is developing confidence with statistics. The American Statistical Association (GAISE, 2015: 11–12) wants students to believe and understand why:
• data beat anecdotes;
• variability is natural, predictable, and quantifiable;
• random sampling allows results of surveys and experiments to be extended to the population from which the sample was taken;
• random assignment in comparative experiments allows cause-and-effect conclusions to be drawn;
• association is not causation;
• statistical significance does not necessarily imply practical importance, especially for studies with large sample sizes;
• finding no statistically significant difference or relationship does not necessarily mean there is no difference or relationship in the population, especially for studies with small sample sizes.
In addition (GAISE, 2015: 12), students should recognize:
• common sources of bias in surveys and experiments;
• how to determine the population to which the results of statistical inference can be extended, if any, based on how the data were collected;
• how to determine when a cause-and-effect inference can be drawn from an association based on how the data were collected (e.g., the design of the study);
• that words such as "normal," "random," and "correlation" have specific meanings in statistics that may differ from common usage.
These quantitative-reasoning and numeracy goals and skills are addressed in this chapter and subsequent ones, with the emphasis less on calculation and more on learning the basic tools required to interpret, represent, critically analyze, and communicate quantitatively.
Let's consider below several basic concepts relevant to developing a critical ability to evaluate common everyday reports that use or misuse numbers: percentages, margin of error, levels of measurement, central tendency measures (averages), and estimations.
PERCENTAGES
In the summer of 2015, the California Highway Patrol (CHP) "released a new study showing a 39 percent increase in the percentage of California drivers seen using a cell phone while driving." That figure is based on 9.2 percent of drivers on a cell phone in 2015 compared with 6.6 percent in 2014. CHP also said "law enforcement wrote 35 percent more tickets for texting-while-driving compared to 2014" (California Highway Patrol, 2015).
The short press release received much attention in the media. Details about the study, however, were not in the brief announcement, thus illustrating a problematic way of reporting the results of research. To find out the relevant details left out of the announcement or news broadcast, you had to make an effort to go online to read the complete report. What more information would you want in order to make sense out of this too-brief media report of survey findings? How do you decide whether these 39 percent,9.2 percent, and 6.6 percent figures are meaningful or not?
In the CHP study (Ewald and Waserman Research Consultants, 2015), we are told that the researchers saw 9.2 percent of drivers on a cell phone in 2015. The first question to ask is, How many drivers were studied? We need to know the denominator. Remember, a percentage is a proportion calculated by dividing a subset of items (numerator) by the total number of items (denominator) and then multiplied by 100 (per centum is Latin meaning "per one hundred"). Sometimes you will read findings "per capita" (Latin for "per head": that is, per person) and "per 1,000" as the World Bank (2014) does in reporting the number of physicians for every 1,000 people in countries around the world (such as Australia, 3.3; Canada, 2.1; Kenya, 0.2). Sometimes you may even see "per 10,000" which the World Health Organization (2013: 122, 125) uses in reporting the number of hospital beds globally (like Ethiopia, 63; New Zealand, 23). These kinds of standardization (percent, per 1,000, etc.) allow us to compare findings among samples of different sizes.
Consider how important this ability to standardize is when deciding whether to report data in terms of absolute numbers as opposed to percentages. For example, it would be accurate to report that Californians hold the most passports among all Americans, averaging close to 2.5 million per year. As the most populous state, with nearly 40 million residents, this shouldn't be a surprising finding. But using raw numbers when comparing states of different populations distorts the message. It turns out that California is actually behind six other states, including less populous states like Alaska and Delaware, when taking into account the relative population of the states: that is, when calculating the percentage of citizens with passports (Stabile, 2016).
Understanding how to read or calculate an accurate percentage is one step in learning numeracy and thinking critically. It's also important to understand what is meant when reports discuss a percentage increase or decrease over time. Recall the CHP cell-phone-use-while-driving study: 9.2 percent of drivers on a cell phone is 39 percent higher than the previous year's 6.6 percent. How can 9.2 be 39 percent higher than 6.6?
At the simplest level, a percentage point increase or decrease is a difference between two percentages, so you could say that cell phone usage went up 2.6 percent in one year (9.2 minus 6.6 equals 2.6 percentage point increase). Some may erroneously misinterpret the "39 percent higher" to mean that driving while on a cell phone increased from 6.6 percent to 45.6 percent. This isn't the case, yet it's easy to see how some misunderstanding or distorted findings can be communicated when not critically asking what the numbers represent.
To arrive at the 39 percent increase, you take the percentage point difference and then divide it by the original percentage (2.6/6.6) to get 0.394 or 39.4 percent (multiply by 100 to get a percentage). Note that the percentage point difference is 2.6, and the percentage increase over time is 39.4. The fact that those are two different numbers can create confusion in readers of data, and can sometimes purposely be used unethically to inflate or deflate actual change.
For example, imagine a shift from 20 percent of a sample in an election political poll supporting Candidate A to 30 percent a few months later. This ten percentage point increase indicates a modest shift in support for the struggling politician, who still does not have a majority. Simply take the percentage point difference and divide it by the original percentage (10/20 = 0.50). Yet, it would be easy to distort the findings to suggest a surge of interest in the candidate by reporting a "tremendous 50 percent improvement" in popularity. Although the statement would be accurate, those without numeracy skills might mistakenly think the candidate went up 50 points in the polls, from 20 percent to 70 percent, focusing on the number 50 rather than the much smaller 30.
Look at this paragraph from a New York Times article (Saulny, 2011) about an increase in multiracial people in the 2010 U.S. Census: "In North Carolina, the mixed-race population doubled. In Georgia, it expanded by more than 80 percent, and by nearly as much in Kentucky and Tennessee. In Indiana, Iowa and South Dakota, the multiracial population increased by about 70 percent." Note the word "doubled" and the large numbers, 80 and 70. Sounds impressive. A few paragraphs later the article reports a possible national multiracial growth rate of 35 percent, maybe even a 50 percent increase from the last census, in 2000. Again, these are large and impressive numbers. What kinds of missing information do you need to better understand these percentages?
Then the article states that in 2000 only 2.4 percent of Americans selected more than one race on the census form. It's one thing to claim that the multiracial population may increase 50 percent, but when the original figure is only 2.4 percent of Americans, a 50 percent increase simply means that the 2010 multiracial population could end up around 3.6 percent of the population. (50% of 2.4 = 1.2, and 2.4 + 1.2 = 3.6.) The number 50 surely sounds more impressive than the smaller figure, 3.6. Manipulating these numbers can create misleading impressions, sometimes unethically done with intention.
By the way, in the 2010 United States Census, 2.9 percent identified as two or more races, while the Pew Research Center (2015a) estimates that 6.9 percent of the U.S. population could be considered multiracial when including "how adults describe their own race as well as the racial backgrounds of their parents and grandparents." Understanding how researchers calculated the number and how it was measured are important questions to ask when interpreting data. Which percentage makes more sense to you, 2.9 or 6.9? What questions would you ask to determine if respondents are multiracial?
Being able to interpret numbers correctly often comes down to reporting all the relevant data. In the New York Times article, we really do not have enough information to know exactly what the multiracial population is in the states listed. That they doubled in North Carolina and increased 80 percent in Georgia tells us little about the actual 2000 or 2010 census figures in those states.
MARGIN OF ERROR
When political polls show that Candidate A has 46 percent of the popular vote and Candidate B has 42 percent, we cannot conclude that Candidate A is winning, despite announcements that A is ahead. Before making this statement, critical thinkers need to ask what the margin of error for the poll is. If the margin of error is plus or minus 4 percent, then it would suggest that Candidate A has between 42 percent and 50 percent support in the total population from which the sample was selected, and Candidate B has between 38 percent and 46 percent. There is a lot of overlap in support, and it could well be that A's true support percentage is, for example, 43 percent, whereas B's is 46 percent. Both those figures are within each candidate's margin of error.
The margin of error is based on sampling issues (see chapter 2) assuming that randomly selected respondents reflect the characteristics of a population with some, but not perfect, accuracy. Usually, the margin of error specifies how confident the researcher is in generalizing a statistical finding to the total population based on a random sample of a particular size drawn from that population. Here is the Pew Research Center's (2016a) statement about their survey practices:
The sampling error for a typical Pew Research Center national survey of 1,500 completed interviews is plus or minus approximately 3 percentage points with a 95% confidence interval. This means that in 95 out of every 100 samples of the same size and type, the results we would obtain will vary by no more than plus or minus 3 percentage points from the result we would get if we could interview every member of the population. Thus, the chances are very high (95 out of 100) that any sample we draw will be within 3 points of the true population value.
Consider the margin of error in Pew Research Center's (2015b) study about teenagers' friendships and romantic relationships in the digital age. As table 1 indicates, any results from the parents surveyed should be viewed as accurate 95 percent of the time within plus or minus 3.4 percentage points of the figure reported. Notice how size of sample impacts the margin of error: With only 101 black teenagers surveyed, any responses to individual survey questions should be interpreted as being within plus or minus 13.3 percentage points of the reported result.
In the survey, 79 percent of all teens said that they instant-messaged their friends. With the margin of error for all teens at 3.7 percent, this means that if Pew were to complete 100 surveys with a similar random sample of 1,060 teens, 95 of the surveys would report a finding between 75.3 percent and 82.7 percent of teens texting their friends (79 – 3.7 and 79 + 3.7). That is, there is a high probability (95%) that the Pew Research Center sample finding of 79 percent is within plus or minus 3.7 percent of the true value in the total population.
Here's a situation where margin of error resulted in some arbitrary decisions with important political ramifications. For the Fox News Channel's (FNC's) first Republican Party U.S. presidential candidates' debate, in August 2015, FNC decided to use an average of several polls to determine the ten most popular participants. What's evident in table 2 is that the margins of error for each candidate in the average of five opinion polls made deciding who should be in and who should be out a difficult choice among those in the tenth through fourteenth positions (Kurtzleben, 2015). How would you interpret these findings using the concept of the margin of error?
LEVELS OF MEASUREMENT
Calculating percentages and margins of error, however, depends on the level of measurement used in a study. Level of measurement refers to the type of values that characterize the elements of a variable. For example, imagine a survey asking, "What type of music do you listen to?" In this case, "music type" is a variable (a concept that varies); "folk," "rap," "classical," and "show tunes" could be the values used to measure that variable. The method used to determine the values is called the level of measurement.
Consider for a moment the Billboard Top 200 music chart. How do you create a list of the top best-selling music when the concept of an album has changed? For years, the Top 200 entailed counting sales of physical vinyl records and CDs by scanning bar codes. Then along came purchasing digital music online (such as from iTunes and Amazon Prime) and streaming music (such as with Spotify and Pandora). Billboard (2014), based on Nielsen Entertainment's measurements, counts "10 digital track sales from an album to one equivalent album sale, and 1,500 song streams from an album to one equivalent album sale."
Excerpted from Critical Thinking by Peter M. Nardi. Copyright © 2017 The Regents of the University of California. Excerpted by permission of UNIVERSITY OF CALIFORNIA PRESS.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.
Les informations fournies dans la section « A propos du livre » peuvent faire référence à une autre édition de ce titre.
Vos droits au titre de la garantie légale : Chaque consommateur de l'UE bénéficie d'au moins deux ans de protection au titre de la garantie légale pour tout bien de consommation vendu dans l'UE. Voir plus
Vendeur : BooksRun, Philadelphia, PA, Etats-Unis
Paperback. Etat : Very Good. First Edition. It's a well-cared-for item that has seen limited use. The item may show minor signs of wear. All the text is legible, with all pages included. It may have slight markings and/or highlighting. N° de réf. du vendeur 0520291840-8-1
Quantité disponible : 1 disponible(s)
Vendeur : ThriftBooks-Dallas, Dallas, TX, Etats-Unis
Paperback. Etat : Good. No Jacket. Pages can have notes/highlighting. Spine may show signs of wear. ~ ThriftBooks: Read More, Spend Less. N° de réf. du vendeur G0520291840I3N00
Quantité disponible : 1 disponible(s)
Vendeur : HPB-Red, Dallas, TX, Etats-Unis
Paperback. Etat : Good. Connecting readers with great books since 1972! Used textbooks may not include companion materials such as access codes, etc. May have some wear or writing/highlighting. We ship orders daily and Customer Service is our top priority! N° de réf. du vendeur S_447915907
Quantité disponible : 1 disponible(s)
Vendeur : Evergreen Goodwill, Seattle, WA, Etats-Unis
paperback. Etat : Good. N° de réf. du vendeur mon0000730256
Quantité disponible : 1 disponible(s)
Vendeur : St Vincent de Paul of Lane County, Eugene, OR, Etats-Unis
Etat : Good. EXTERIOR PAGE EDGES HAVE SOME STAINING AND/OR SMUDGING ON THEM BUT BOOK REMAINS IN GOOD READABLE CONDITION. Paperback 100% of proceeds go to charity! Good condition with all pages in tact. Item shows signs of use and may have cosmetic defects. N° de réf. du vendeur L-06-5366
Quantité disponible : 1 disponible(s)
Vendeur : One Planet Books, Columbia, MO, Etats-Unis
paperback. Etat : Good. First Edition. Ships in a BOX from Central Missouri! May not include working access code. Will not include dust jacket. Has used sticker(s) and some writing and/or highlighting. UPS shipping for most packages, (Priority Mail for AK/HI/APO/PO Boxes). N° de réf. du vendeur 002204415U
Quantité disponible : 20 disponible(s)
Vendeur : Textbooks_Source, Columbia, MO, Etats-Unis
paperback. Etat : Good. First Edition. Ships same day or next business day! UPS shipping available (Priority Mail for AK/HI/APO/PO Boxes). Used sticker and some writing and/or highlighting. Used books may not include working access code or dust jacket. N° de réf. du vendeur 002204415U
Quantité disponible : 20 disponible(s)
Vendeur : HPB-Emerald, Dallas, TX, Etats-Unis
paperback. Etat : Very Good. Connecting readers with great books since 1972! Used books may not include companion materials, and may have some shelf wear or limited writing. We ship orders daily and Customer Service is our top priority! N° de réf. du vendeur S_484959492
Quantité disponible : 1 disponible(s)
Vendeur : GreatBookPrices, Columbia, MD, Etats-Unis
Etat : good. May show signs of wear, highlighting, writing, and previous use. This item may be a former library book with typical markings. No guarantee on products that contain supplements Your satisfaction is 100% guaranteed. Twenty-five year bookseller with shipments to over fifty million happy customers. N° de réf. du vendeur 29179217-5
Quantité disponible : 15 disponible(s)
Vendeur : GreatBookPrices, Columbia, MD, Etats-Unis
Etat : As New. Unread book in perfect condition. N° de réf. du vendeur 29179217
Quantité disponible : 12 disponible(s)