SCIENCE & ENGINEERING INDICATORS

Although this report tracks a set of questions aimed to assess knowledge of several basic scientific facts, substantial research has shown that general measures of science knowledge typically only have small—although meaningful—relationships with how people make decisions in their public and private lives (Allum et al. 2008). NASEM also recently highlighted that science literacy is largely a function of general (or foundational) literacy and that more focus should be directed toward the ability of groups to use science to make evidence-based decisions (NASEM 2016b). In this regard, it should be recognized that the science literacy of individuals is unequally distributed across social groups. This means that some groups or communities can use science when needed, whereas others cannot because they may not have access to resources such as local expertise (e.g., community members who are also scientists, engineers, or doctors).

The current GSS uses nine questions and therefore does not address the full range of scientific subjects that could be included. Further, these questions were selected several decades ago based on the likelihood that they would remain stable over time rather than as an effort to capture any specific body of scientific knowledge. Consequently, the survey data do not represent a deep or comprehensive measurement of scientific knowledge. These questions might instead be understood as a way to capture the degree to which people have paid attention to science over their life or might be expected to do so in the future (Kahan 2017). To address these types of concerns, the 2010 edition of Indicators included responses to an expanded list of questions about scientific ideas based on regular exams given to American students as part of the American Association for the Advancement of Science’s Project 2061. This research found that respondents who “answered the additional factual questions accurately also tended to provide correct answers to the trend factual knowledge questions included in the GSS” (NSB Indicators 2010: Science and Technology: Public Attitudes and Understanding). These nine items are presented as an indicator of people’s familiarity with scientific ideas or facts taught in school, as a means for evaluating trends or conducting group comparisons. Making generalizations about Americans’ overall knowledge of science should be made cautiously given that this indicator comprises a small number of questions on school-level knowledge of science.

In 2018, Americans correctly answered an average of 62% of the nine true-or-false or multiple-choice items from the long-running factual knowledge questions (Table 7-1; Table S7-24 and Table S7-25). The 2018 average is statistically similar to averages in recent years and the historical average since 1992 (Table S7-25). In terms of specific questions (Table S7-26 and Table S7-27), the overall average stability since 1992 hides some variation within individual questions. For example, the share of correct answers to questions on radioactivity and the fact that the Earth goes around the Sun has been relatively stable except for spikes of additional correct or incorrect responses in specific years. After increasing in the early period of the survey, the share of correct answers for several questions has been relatively stable. Examples include whether antibiotics kill viruses, whether electrons are smaller than atoms, and whether lasers work by focusing sound waves. The one question where there has been a small decline in correct answers over time is whether it is the “father’s gene that decides whether the baby is a boy or a girl.” The Pew Research Center has also collected data on this topic and found similar patterns of results (Kennedy and Hefferon 2019).

The GSS includes two additional true-or-false science questions that are not included in the trend data reported earlier because Americans’ responses to these questions appear to reflect factors beyond familiarity with scientific facts (e.g., beliefs related to specific religious teachings). One of these questions is about evolution, and the other is about the origins of the universe. The data presented in this section show that, in some specific cases, changes to question wording can produce substantially different responses.

In 2018, nearly half of Americans (49%) correctly indicated that “human beings, as we know them today, developed from earlier species of animals,” and 38% indicated that “the universe began with a big explosion” (Table S7-26). Both percentages are relatively low compared with scores on most of the other factual information questions in the survey. Some people may respond to the evolution question based on their religious beliefs rather than familiarity with scientific concepts (Maitland, Tourangeau, and Sun 2018). As such, half of 2018 GSS respondents received an evolution question that omitted reference to human evolution and instead read, “Elephants, as we know them today, developed from earlier species of animals.” With the question posed in this form, 66% gave the scientifically expected response (compared with 49% when the question focused on humans, as noted earlier).^{​Data source is SRI International (2020).}

Similarly, two alternate “origin of the universe” questions were also provided to random subsets of respondents in the 2018 GSS. First, simply adding the preface “according to astronomers” to the statement “the universe began with a huge explosion” resulted in about two-thirds (65%) of respondents providing the scientifically correct answer, considerably higher than the 38% who provided the correct answer without the preface. Another subset of respondents was given the statement “the universe has been expanding ever since it began” and was asked if this was true or false. Again, about two-thirds (68%) of respondents gave the correct scientifically understood response.^{​The nature and origins of this difference were discussed at length in the 2018 edition of Indicators using 2016 data that showed similar patterns. The analyses suggested that people who responded correctly to the other factual questions were also more likely to respond to the modified Big Bang and evolution questions. However, correct responses to the original Big Bang and evolution questions were not as closely connected to correct responses to the other questions (NSB Indicators 2018: Science and Technology: Public Attitudes and Understanding).}

Another indicator of the public’s understanding of science focuses on the public’s understanding of how the scientific process generates and assesses evidence. Data on three scientific process elements—probability, experimental design, and the scientific method—show some previous increases in Americans’ understanding of the scientific inquiry process but substantial stability in recent years.

Two probability questions are included in the GSS. Most (84%) Americans in 2018 correctly indicated that, faced with an inherited disease that affects 1 in 4 children, the fact that a couple’s first child has the illness does not affect whether three future children will have the illness.^{​Data source is SRI International (2020).} In addition, about three-quarters (74%) correctly responded that the odds of a genetic illness are equal for all of a couple’s children. Overall, 65% correctly answered both probability questions. The public’s understanding of probability as measured by these two questions has been stable for most of the last 20 years (Figure 7-11; Table S7-28).^{​Earlier NSF surveys used for Indicators employed additional questions to measure understanding of probability. Bann and Schwerin (2004) identified a smaller number of questions that could be administered to develop a comparable indicator. Starting in 2004, the NSF surveys used these questions for a trend factual knowledge scale. This scale does not include the questions aimed at studying scientific reasoning and understanding (e.g., questions about probability or the design of an experiment), and the current report attempts to avoid describing the combined questions as an overall knowledge scale. Instead, the report recognizes that the nine questions “are not understood to certify comprehension of any canonical set of facts or principles. Rather, they are conceptualized as observable (or manifest) indicators of an unobservable (latent) cognitive capacity that enables individuals to acquire and use scientific knowledge” (Kahan 2017:997).}

With regard to understanding experiments, nearly half (49%) of Americans in 2018 correctly answered a question about how to test a drug, then provided a correct response to an open-ended question that required them to explain the rationale for an experimental design (i.e., giving 500 people a drug while not giving the drug to 500 additional people, who then serve as a control group) (Table S7-29). On average, the percentage of correct responses rose over the previous 20 years (Table S7-28); this is despite the substantial year-to-year variation that may be partially explained by reliance on human coders to categorize responses.^{​Declines, such as those seen in 2012, need to be regarded with caution. In that case, the percentage of Americans who correctly answered the initial multiple-choice question about how to conduct a pharmaceutical trial remained stable between 2010 and 2012. It was the only follow-up question that asked respondents to use their own words to justify the use of a control group that saw a decline. For this question, interviewers recorded the response, then trained coders to use a standard set of rules to judge whether the response was correct. Although the instructions and training have remained the same in different survey years, small changes in survey administration practices can sometimes substantially affect such estimates.}

Similarly, respondents were asked whether they have “a clear understanding,” “a general sense,” or “little understanding” of the term scientific study. About 27% in 2018 said they have “a clear understanding,” whereas 51% said they have “a general sense.” These respondents were then asked to use their own words to describe what it meant to study something scientifically, and their responses were coded. Overall, about a quarter of respondents (24%) adequately described a scientific study as involving something to do with testing theories or hypotheses, conducting experiments, or making systematic comparisons, similar to results dating back to 1999 (Table S7-28, Table S7-30, and Table S7-31).

In general, those with the most education and those who answered the most factual questions correctly were more likely to respond that they had “a clear understanding,” but many respondents with relatively limited background in science also reported high levels of understanding (Table S7-30). For example, 15% of those with the lowest level of science education reported having “a clear understanding” of what constitutes a scientific study, compared with 51% of those with the most scientific education.

All scientific reasoning questions can be combined into an overall measure of scientific inquiry understanding. Using this combined measure, about 43% of Americans in 2018 could correctly respond to the two probability questions and provide a correct response to at least one of the open-ended questions about experimental design or what it means to study something scientifically (Table S7-28). In general, respondents with more education and respondents with higher incomes performed better on the scientific inquiry questions (Table S7-29 and Table S7-30).

Previous editions of Indicators have reported that people outside the United States generally do similarly (e.g., Canada) or less well than Americans on similar questions; however, few countries currently put substantial focus on public literacy surveys (Table 7-1). The 140-country survey for the Wellcome Trust (2019), however, included several questions about self-perceived knowledge and found that people in developed regions, such as North America and Europe, are more likely to say they know “a lot” or “some” about science; people in poorer regions and countries in Asia, Africa, South America, and the Middle East are much less likely to indicate science knowledge. Younger people (and those with more education) in all regions are also more likely to report higher knowledge levels.

Another indicator of public understanding about S&T comes from a measure focused on the public’s capacity to distinguish science from pseudoscience. One such measure has been included in Indicators because of the availability of data going back to the late 1970s: Americans’ views on whether astrology is scientific.^{​Some respondents might understandably argue that because astrology is based on systematic observation of planets and stars, it is “sort of scientific.” The fact that those with more formal education and higher factual science knowledge scores are consistently more likely to fully reject astrology suggests that this nuance has only a limited effect on results. Another problem is that some respondents may also confuse astrology with astronomy, and such confusion seems most likely to occur in some of the same groups (i.e., relatively lower education and factual knowledge) that might be predicted to get the question wrong. This could artificially inflate the number of incorrect responses. However, the question comes immediately after a question that asks respondents if they have ever “read a horoscope or personal astrology report,” which offers respondents a hint that astrology is not astronomy. Also noteworthy is the fact that a Pew Research Center study (2009) using a different question found that 25% of Americans believe in “astrology, or that the position of the stars and planets can affect people’s lives.” Gallup found the same result with the same question in 2005 (Lyons 2005). In contrast, the 2010 GSS found that 6% saw astrology as “very scientific,” and 28% saw astrology as “sort of scientific” (34% total). Pew Research Center found that 73% could distinguish between astrology and astronomy and that there were few demographic differences beyond education (Funk and Goo 2015).} Other examples of pseudoscience include the belief in lucky numbers, extrasensory perception, or magnetic therapy.

More Americans today than in the past see astrology as unscientific, although there has been some variation in recent years. In 2018, about 58% of Americans said astrology was “not at all scientific,” a value near the middle of the historical range and down somewhat from 65% in 2014 (Table S7-32). About a third of Americans thought astrology was “sort of scientific,” and the remainder thought astrology was “very scientific” or “didn’t know.” Men, older respondents, those with more education, and those with more correct answers on the factual science questions all tend to be more likely to see astrology as nonscientific.