Knowledge, Literacy, and Power

Thomas G. Sticht, C. Richard Hofstetter, Carolyn H. Hofstetter

San Diego Consortium for Workforce Education & Lifelong Learning (CWELL)
March 1997

Support for this research was provided in part by the William and Flora Hewlett Foundation; the Lila Wallace Reader's Digest fund, the Department of Political Science, San Diego State University; and the Spencer Foundation. The opinions expressed are solely those of the authors, and no official endorsement by their institutional affiliations should be inferred.

Knowledge, Literacy and Power

Abstract

The importance of content knowledge and reading practices to the achievement of power was studied with adults. Relationships were examined among general, "mainstream" society knowledge, domain specific political knowledge, the amount of reading engaged in and three indicators of power, occupation, income and political activity. Care was taken to ensure that extraneous cognitive processing variance did not influence the results by using simple checklists of declarative knowledge that required listeners, on the telephone, to simply say "yes" if they thought they recognized a given factual stimulus. The results of two studies indicated that there were positive relationships among amount of content knowledge, reading and power, even when age, education and ethnicity were held constant. The latter is important because it indicates that regardless of one's cultural background, possession of large "banks" of declarative knowledge about the "mainstream" culture of the United States is associated with achieving and manifesting power.

Knowledge, Literacy and Power

In democratic nations that subscribe to meritocratic principles, it is generally assumed that "knowledge is power," and that, to a large extent, knowledge is based on literacy (Lerner, 1958). The traditional view has been that, as a general rule, the more literate a person is, the more knowledgeable the person will be and the more likely he or she is to gain access to socially privileged positions or to gain a status that carries with it the capacity for influencing the thoughts and behaviors of others in direct or indirect ways (Lipset, 1960; Verba & Nie, 1972; Wolfinger & Rosenstone, 1980).

Today, however, among educators and other citizens alike, the importance of the acquisition of content knowledge as an outcome of education and as an important contributor to social status and power is in a state of ambivalence. On the one hand, education reformers such as Freire (1970) have argued against the "banking" concept of education in which the culture of the "dominant" classes is transmitted to the "oppressed" by requiring the latter to learn the facts, concepts, famous personalities, etc. of the "dominant" classes. He and various sympathizers have argued for an approach to education in which processes of "critical thinking" are emphasized over the acquisition ("banking") of content knowledge.

Interestingly, Freire, heavily influenced by Marxist-socialist thinking, has recently been joined by businessmen who also argue against the "banking" approach to education. Robert W. Galvin, Chairman of the Executive Committee of the Motorola Corporation has stated, "While most descriptions of necessary skills for children do not list "learning to learn," this should be the capstone skill upon which all others depend. Memorized facts, which are the basis for most testing done in schools today, are of little use in the age in which information is doubling every two or three years. We have expert systems in computers and the Internet that can provide the facts we need when we need them. Our workforce needs to utilize facts to assist in developing solutions to problems. The worker needs to be able to utilize the systems which give him or her "just-in-time" access to information when it's required in the problem-solving process." (Galvin, in Murnane & Levy, 1996, p. vxii)

On the other hand, Hirsch (1987; 1996) has reviewed and argued against the position of many educators and researchers who call for the teaching of "critical thinking," "learning to learn," and other abstract cognitive processes over the teaching and learning of content knowledge. He argues for the importance of acquiring knowledge, that is, facts, names, concepts, and so forth, during the school years as the basis for being a knowledgeable, literate and productive adult who can continue to learn independently by reading. Hirsch has bolstered his arguments with references to the fairly large body of research that indicates the importance of "prior knowledge" for reading comprehension.

Both Hirsch's arguments and research on the importance of "prior" and "background" knowledge in reading comprehension are reviewed by Bruer (1993, pp. 173-213). While Bruer indicates the importance of learning various cognitive and metacognitive strategies for improving reading and learning from text, he also agrees to a large extent with Hirsch regarding the importance of knowledge in making it possible to acquire new knowledge from texts. He points out the importance of having a large body ("bank") of vocabulary knowledge, background knowledge about the world around us, knowledge about topics for creating "gists," domain-specific knowledge for understanding texts in special areas of content, knowledge of literary forms and genres to aid in understanding special forms of texts, and the importance of "prior" or "background" knowledge in reading to learn new knowledge.

Power, content knowledge, and cognitive processes

In the present study we view "power" as indicated by the achievement of a higher status occupation and/or the ability to earn an average or higher level of income. These achievements empower the person in the mainstream society. Additionally, the more knowledgeable person is more likely to exercise the rights (power) of citizenship and to engage in political activities, such as voting, to advance his or her causes (Neuman, Just, & Crigler, 1992; Zaller, 1992). Through these means the more highly knowledgeable person joins the ranks of the influential in the "dominant" or "mainstream" society.

Previous studies have confirmed relationships among literacy and the indicators of power outlined above; occupation, income and political activity, including voting (Kirsch, Jungeblut, Jenkins & Kolstad, 1993; Sticht & Armstrong, 1994; Kaplan & Venezky, 1994). However, in these studies of literacy, declarative knowledge (i.e., factual or content knowledge available to people for use) is confounded with process skills of a largely unknown admixture. Because of this commingling of declarative knowledge and skills, it is not possible to use such studies to evaluate the role of content knowledge in empowering people. For instance, the prose, document and quantitative literacy tasks of the National Adult Literacy Survey (NALS) incorporated a number of "search and locate" and other cognitive skills that placed heavy demands on working memory. It is well established that working memory becomes increasingly less efficient with advanced age (Bernstein, Roy, Srull, & Wickens, 1988; Meyer, Marsiske, & Willis, 1993) so it is possible that the load on working memory contributed to the decline in performance observed for adults over the age of fifty (Kirsch, Jungeblut, Jenkins & Kolstad, 1993, p. 31).

Knowledge and literacy

Several lines of research have converged to suggest that people become highly knowledgeable and highly literate largely by engaging in numerous literacy practices, such as reading books, magazines, newspapers, and so forth (Krashen, 1993; Reder, 1994; Kaplan & Venezky, 1994). A review of the major assessments of adult literacy in the United States revealed that, since 1937 it has repeatedly been found that for adults, as years of education increases there are corresponding increases in both the number of literacy practices in which adults engage and the amount of knowledge and skill displayed in the assessments (Sticht & Armstrong, 1994; Smith, 1996, p. 196).

The importance of the development of knowledge as both an outcome of and contributor to adult literacy was expressed in the definition of literacy that was adopted by the advisory panel of experts for the 1993 National Adult Literacy Survey (Kirsch, Jungeblut, Jenkins & Kolstad, 1993). The definition of literacy agreed to was: "Using printed and written information to function in society, to achieve one's goals, and to develop one's knowledge and potential (italics added)."

This definition notes the important role that literacy plays in helping people develop their store of knowledge. In contrast, the role that one's prior store of knowledge plays in helping people use their literacy skills was also acknowledged by the advisory panel for the NALS in its acceptance of the definitions of the three different literacy scales that were developed. These included: prose literacy, the knowledge and skills needed to understand and use information from texts, document literacy, the knowledge and skills required to locate and use information contained in materials, and quantitative literacy, the knowledge and skills required to apply arithmetic operations embedded in printed materials (Kirsch, Jungeblut, Jenkins & Kolstad, 1993).

It should be recognized that "skills" are, themselves, forms of procedural knowledge. Skills may be regarded as procedural knowledge that is acquired along with declarative knowledge. From these definitions, it is clear that the advisory panel for the National Adult Literacy Survey understood that the use of printed and written information to accomplish tasks requires, as a prerequisite, prior knowledge and skills to make such use possible. In this sense, knowledge is both a prerequisite for and an outcome of the use of literacy.

Despite the acknowledged importance of declarative knowledge as a component of all literacy practices, a review of every large-scale assessment of adult literacy since 1917, both military and civilian, revealed that there has never been an attempt to determine the validity of the maxim, "knowledge is power" by identifying the contribution to power (jobs, income, voting) of the declarative knowledge component of literacy, separate from the many other demands of more or less complex cognitive tasks that introduce unknown process variance in the study of relationships among literacy and power (Sticht & Armstrong, 1994; Messick, 1989). The research reported here aimed at determining empirically the value of "banking" declarative knowledge of the "dominant" or "mainstream" society in the United States of the mid-1990's as a contributor to achieving power.

Theoretical basis for the present research

The present research used telephone survey and checklist methodologies to study relationships of declarative knowledge to power. The general approach is based on the "modal model of memory" that has been examined in over thirty years of research (Healy & McNamara, 1996, p. 143). Simply stated, this model conceives of a human cognitive system that includes both a knowledge base in long term memory consisting largely, though not exclusively, of language-based declarative and procedural knowledge, and a working memory in which active information processing takes place using the knowledge from long term memory and information picked-up from the external world through a perceptual system. Generally speaking, highly literate individuals possess large bodies of knowledge and information processing capacity and efficiency in working memory to process information in complex graphic documents (Kyllonen & Christal, 1990).

Stanovich (1993) and associates have conducted research to explore how engagement in literacy practices by children and adults contributes to their declarative knowledge base in the human cognitive system. In this research, Stanovich and associates developed an innovative method for assessing declarative knowledge with checklists that reduce task demands to a simple "yes" or "no" judgment on the part of the reader. Performance on these checklists correlates significantly with a variety of literacy activities and cognitive assessments (Stanovich, 1993; Stanovich & Cunningham, 1993; Allen, Cipielewski, & Stanovich, 1992; West, Stanovich, & Mitchell, 1993). From a causal perspective, the argument by Stanovich and associates is that those who read a lot acquire a large knowledge base containing the names of authors, magazines, and persons known for their contributions to film, theatre, music and other cultural activities, and a large vocabulary of words that are typically not encountered with high frequency in day-to-day oral communication nor on television or radio. The declarative knowledge base made-up of authors, magazines, famous people and vocabulary is an indicator of both the amount of reading in which individuals have engaged and of the cognitive outcomes of that reading in terms of the growth in the individual's declarative knowledge base.

In the present research, two studies were conducted that examined how declarative knowledge relates to power. Both studies used the checklist methodology developed by Stanovich and associates to obtain information about adult's declarative knowledge. Lists containing discrete items, such as names or single vocabulary words, that require only a yes/no decision for each item are particularly suitable for presentation by telephone because they do not overload working memory and introduce irrelevant task variance (Messick, 1989).

Study 1 looks at relationships among various demographic variables, engagement in reading and electronic media usage, general, "mainstream," or "dominant" cultural declarative knowledge, and two indicators of power, occupational status and income. Study 2 focuses on domain specific political knowledge and looks at relationships among demographic variables, newspaper reading and electronic media usage, political knowledge and three power indicators, household income, voting, and engagement in political activities.

Study 1: General Declarative Knowledge and Power

In Study 1, subjects' declarative knowledge was assessed using shortened versions of the Stanovich checklists for declarative knowledge of famous authors, magazine titles, famous people, and vocabulary words (West, Stanovich, & Mitchell, 1993). It provided information about respondent's general declarative knowledge taken from samples of the knowledge of the "mainstream" or "dominant" society. Study 1 examines for the first time the relationships of declarative knowledge to two indicators of power, occupational status and income, when age, education, and ethnicity are held constant.

Method

Subjects

Data for this study were derived from telephone interviews with 538 adults residing in households that could be reached by listed or unlisted telephone in the larger San Diego, California metropolitan area. This included approximately 96 percent of all households. Sampling was conducted by using a random-digit-dialing procedure designed to reach households without numbers listed in the telephone directory, due to unlisted numbers or newly listed numbers not yet printed, as well as households listed (Dillman, 1978, pp. 232-281; Frey, 1989, pp. 79-116). Respondents who agreed to participate further and were willing to provide their name and address were mailed a written questionnaire as an alternative modality follow-up.

In the telephone interview, the subjects' mean length of residence in San Diego was 20.6 years (SD=15.5), mean educational level was 14.5 years (SD=2.6), mean age was 41.0 years (SD=16.0), mean total household income was $34, 340 (SD=$12,240).

Table 1.
Comparison of the San Diego telephone samples in studies 1 and 2 with the 1990 U. S. Census figures for San Diego County.
Variables Study 1 Study 2 Census
Household Income (478)b (562)    
Under $10,000 9.0 9.6 6.8
$10,000-$49,000 59.6 59.6 56.5
Over $50,000 31.4 30.8 36.7
Age (512) (676)   
18-24 15.4 15.2 17.8
25-64 72.4 70.4 67.6
65+ 12.1 14.4 14.4
Gender (522) (676)   
Male 47.9 49.1 50.9
Female 52.1 50.9 49.1
Education (517) (676)    
0-12 25.5 28.7 40.9
13-16 53.4 55.7 50.3
17+ 21.1 15.5 8.8
Ethnicity (520) (663)    
White 72.9 74.8 65.6
Latino 13.7 10.7 20.0
Black 4.6 6.9 6.0
Asian 5.4 5.1 7.5
Other 3.5 2.4 0.0

a = Numbers are percentages with the characteristics listed.

b = Numbers in brackets are total numbers in samples with data for a given characteristic.

The survey procedures yielded a sample that matched the 1990 U.S. Census data closely, with some notable exceptions. Table 1 shows statistics for the telephone and U. S. Census population for the San Diego region. The telephone sample's gender, age, and income were similar to census distributions. Minorities were somewhat underrepresented and the telephone data were skewed upward in educational attainment, underrepresenting the lowest level of educational attainment and overrepresenting the highest level, in comparison to census data. Although moderate upward bias in the education parameters was present in the sample we do not believe that this undermines the main findings of the study. First, all English-speaking groups were represented in the telephone survey. Second, the thrust of our argument is based on correlational analyses. These kinds of sample biases present would reduce, not increase, variance. This attentuates the correlations making the findings overly conservative.

Interview procedures

Interviewing was conducted by university students trained for telephone interviewing for the project during the late spring and early summer, 1994. Subjects were called between 4:30 p.m. and 9:30 p.m. weekdays and between 9:00 a.m. and 9:30 p.m. on weekends. Interviewers introduced the survey to the person who answered the telephone, gained informed consent, and asked to speak to the adult (18 years of age or older) who had "the most recent birthday" as a random method of selection among adults in the household. No substitutions were allowed so interviewers frequently were required to call the household back in order to complete an interview with the appropriate respondent. Up to four callbacks were made to households and a response rate of approximately 50 percent was attained. Due to resource constraints, interviews were conducted only in English, a procedure that eliminated approximately four percent of households.

The telephone interviews provided an oral presentation of information requiring the respondents to listen and respond from what they heard. Interviewers followed a protocol containing 63 questions, some with multiple items. About half of the questions concerned the assessment of knowledge. These questions were interspersed among other questions that were part of another on-going research project conducted in the area of political behavior. The interviews required a mean of 27.7 (SD= 7.6) minutes to complete.

Instrumentation

Literacy knowledge checklists. For the sake of time, four abbreviated versions of the checklists used by West, Stanovich and Mitchell (1993) were used in the telephone survey. The appendix presents the items used in the Author Recognition Test , the Magazine Recognition Test, the Cultural Knowledge Test and the Vocabulary Knowledge Test. Generally, items were chosen to represent mainstream cultural knowledge (some items came from the work of Hirsch (1987), though some names were chosen to reflect multicultural knowledge within the United States (e.g., Steve Biko, Rosa Parks). Five of the vocabulary items included words typically known by students in the 8th through 12th grades, and the rest are familiar to adults with some college education. The four declarative knowledge checklists used in the telephone survey are given in Appendix A, along with the interview question number, the question asked, the means, standard deviations, and numbers of adults responding to the particular item, and the percentage of adults in each of five knowledge levels that knew the item. In the actual interview, the foils were mixed randomly among the genuine items.

For each checklist, an adjusted percent correct score was calculated. The adjusted percent correct score for an individual was the proportion of correctly identified real names or words minus the proportion of foils incorrectly identified as real names or words. For instance, if a person said "yes" to 10 of the 17 names of famous people on the Cultural Knowledge Test and to 2 of the 6 foils, the person's score for the Cultural Knowledge Test was 25.5 (10/17 minus 2/6, or 58.8 minus 33.3). The correction for guessing prevented the subjects from simply responding "yes" to all items. The signal detection rationale for the scoring procedures are given in West, Stanovich & Mitchell (1993, p. 38).

Split-half, internal consistency (Spearman-Brown) reliabilities of the checklists ranged from .80 (Magazine Recognition Test) to .88 (Cultural Knowledge Test). To increase the reliability of the checklists as measures of the knowledge component of literacy, a Total Knowledge score was calculated made up of the full number of 50 actual and 24 foil names and words. The internal consistency reliability for Total Knowledge was .91. For a sub-set of 140 respondents who completed the checklists by telephone and then later by reading and responding to mailed-out copies of the same checklists, test-retest ("alternate modality" or stability) reliabilities were obtained as: Total Knowledge score (r=.80), Author Recognition Test (r=.71), Magazine Recognition Test (r=.67), Cultural Knowledge Test (r=.73), and Vocabulary Knowledge Test (r=.63). Thus, strong evidence of reliability (p<.05), both in terms of internal consistency and test-retest, was present for each of the four scales and the Total Knowledge scale.

Engagement in media practices. Subjects were asked the number of times in an average week he or she engaged in various media practices such as reading for pleasure newspapers, books, and newsmagazines or reading job-related materials for work, watching television, listening to the radio, etc. Table A-5 in the Appendix shows the questions used to explore engagement in literacy and other media practices. For each respondent, a total literacy practices score was obtained as the average of questions 25a-k in Table A-5. The literacy practice variable is a composite indicator of "print exposure" used to relate average frequency of weekly reading of different materials for various purposes to demographic variables and the knowledge checklists.

Defining knowledge levels

The National Adult Literacy Survey cast distributions of scores on literacy assessments into five levels of proficiency to identify groups, ranging from low to high (Sticht & Armstrong, 1994). Similarly, to illustrate the feasibility of that approach in the present case, the results of the telephone survey were cast into five levels of declarative knowledge using Total Knowledge scores.

To obtain the five levels of knowledge, each person's adjusted percent correct score on each checklist was added together to give a Total Knowledge score. Then the mean adjusted, percent correct score for Total Knowledge for the combined sample was calculated. This score was used to divide the sample into five groups or levels using the mean (45 adjusted percent correct) and the standard deviation (SD, 25 adjusted percent correct) for the total sample. Knowledge levels were defined in adjusted percent correct scores from low to high proficiency as: Level 1= scores at -1.0 SD below the mean or lower (0-20 adjusted percent correct), Level 2 = scores between -.5 to -.1.0 SD (21-32 adjusted percent correct), Level 3 = scores between _ .5 SD (35-58 adjusted percent correct), Level 4 = scores between +.5 to +1.0 SD (59-70 adjusted percent correct), and Level 5 = scores from +1.0 SD and above (71-100 adjusted percent correct).

The five knowledge levels are used in presenting the findings for various demographic and media practice variables in two ways. First, the data are analyzed to find out what proportion of a given demographic or media practices sample is in each of the five levels. For instance, what percentage of all the males in the telephone interview sample are in level 1, what percentage are in level 2, and so on for levels 3, 4 and 5 on the Total Knowledge scale. In a second use, the data are analyzed to find out what proportion of people in each level of knowledge are in a given demographic or media practices group. For instance, what percentage of people in knowledge level 1 are males, what percentage in knowledge level 2 are males, and so forth for each knowledge level.

Results

Table 2 presents the correlations among key demographic variables, the four checklists, a "practice" variable (e.g., How often during an average week do you read a local or national newspaper?) computed as the average of questions 25a-k (see appendix), and two indicators of power, occupational status and income. The practice variable is a composite indicator of "print exposure" and relates average frequency of weekly reading of different materials for various purposes to education, age, and the knowledge checklists.

Table 2 shows positive correlations among demographic, knowledge, practice and power variables consistently found in adult literacy assessments for over 75 years (Sticht & Armstrong, 1994). Generally, better educated subjects scored higher than less well educated subjects, older adults scored better than younger, the majority group (whites) scored better than minorities (Hispanics, Blacks, Asians, others), managers and professionals performed better than clerical and sales persons, who, in turn, performed better than unskilled workers and laborers, those who earned more scored higher than those who earned less, and those who spent more time per week reading scored higher than those who read less.

Women tended to be somewhat less well educated, to engage in fewer literacy practices, to hold somewhat higher level jobs but to earn less than men.

Table 2.
Study 1: Correlations among knowledge and demographic variables.
Variables
1
2
3
4
5
6
7
8
9
10
11
12
Education 1.00 .07* 0.23 0.31 0.31 0.36 0.37 0.34 0.14 -.10a 0.34 0.34
Age   1.00 0.25 0.16 0.25 0.19 0.27 .08* 0.30 .05* 0.12 .04*
ART     1.00 0.61 0.61 0.53 0.81 0.23 0.27 -.10a 0.24 0.23
MRT       1.00 0.58 0.54 0.82 0.22 0.30 .05* 0.25 0.23
CKT         1.00 0.62 0.84 0.23 0.32 .05* 0.23 0.18
VKT           1.00 0.82 0.25 0.29 -.04* 0.21 0.26
Total Knowledge             1.00 0.28 0.36 .05* 0.29 0.27
Practice               1.00 .09a -0.16 0.16 0.26
Ethnicity                 1.00 .06* 0.22 0.13
Gender                   1.00 0.21 -0.13
                      1.00 0.17
Annual Income                       1.00

ART=Author Recognition Test; MRT=Magazine Recognition Test; CKT= Cultural Knowledge Test; VKT=Vocabulary Knowledge Test. Total Knowledge= scores summed over the four checklists; Practice = mean scores on questions 25a through k (see appendix) for different reading practices. *= Not significant, a= p < .05, all others significant beyond p < .01. Underlined r's are part-whole correlations. Ethnicity= nonwhites (0) and whites (1); gender= males (1) and females (2).

Engagement in literacy practices/print exposure

Table A-5 in the appendix presents mean scores and SD's related to the respondent's estimates of the frequency they engaged in various literacy practices during a typical week. Overall, subjects reported reading a newspaper an average 4.4 times a week (SD=2.8) (Q's 7 & 25g). Reading for pleasure (Q25a) was the most frequent reading practice (Mean=4.68; SD=2.50) while listening to someone read aloud was the least frequently engaged in weekly literacy practice (Mean=0.52; SD=0.74).

Generally, the trends for practice follow those for Total Knowledge and are significant at p<.05. As educational attainment (r=.34), age (r=.08), occupational status (r=.17), and income (r=.26) increase, the average frequency of weekly literacy practices increases. Whites were slightly more likely to engage in literacy practices than nonwhites (r=.09).

Questions 5A and 6 in Table A-5 of the appendix present the mean hours per day the subjects reported spending on either watching television or listening to the radio. There was a significant, negative (r=-.14, p<.001) relation between the number of hours of television watched and the average weekly literacy practice score. No relation between radio listening and literacy practices was found. Neither television viewing nor radio listening was related to any of the declarative knowledge checklist scores.

Analyses by knowledge levels

Correlational analyses reveal the overall trends in relationships among the variables under investigation. But they do not reveal what goes on within the distributions of scores. For that reason analyses are presented here that show how different variables are distributed in the five different Total Knowledge categories defined above.

Table 3.
Study 1.

Percentage of respondents in each demographic group falling into each of five levels of Total Knowledge. For instance, 9.2 percent of those with 17+ years of education were in Level 1 while 43.1 percent were in Level 5.

Total Knowledge Levels*

(*Data are percentages in each knowledge level. See text for definition of levels)
Variables N 1 2 3 4 5
Total Sample 538 19.2 14.1 31.4 16.7 18.6
Normal Curve   16.0 15.0 38.0 15.0 16.0
Gender
Male 250 21.2 19.2 22.0 18.8 18.8
Female 272 18.8 19.5 19.1 21.3 21.3
Education
0-12 132 31.0 30.1 21.1 11.9 06.8
13-14 146 22.6 22.6 21.9 18.5 14.4
15-16 130 13.1 13.1 25.4 26.9 21.5
17* 109 09.2 11.0 12.8 23.9 43.1
Age
16-18 14 35.7 42.9 21.4 00.0 00.0
19-24 65 41.5 23.1 16.9 13.8 04.6
25-39 198 21.7 21.2 23.2 21.7 12.1
40-54 141 09.9 13.5 15.6 24.1 36.9
55-64 32 06.3 12.5 28.1 18.8 34.4
65+ 62 17.7 19.4 22.6 21.0 19.4
Ethnicity
White 379 11.9 17.4 22.4 23.2 25.1
African-Amer. 24 25.0 37.5 16.7 20.8 --
Hispanic 71 45.1 19.7 18.3 08.5 08.5
Asian 28 42.9 21.4 14.3 14.3 07.1
Other 18 33.3 38.9 05.6 11.1 11.1
Occupation
Labor /Operator 50 42.0 26.0 14.0 12.0 06.0
Semi/Skill 103 21.4 22.3 25.2 20.4 10.7
Clerk/Sales 97 21.6 23.7 18.6 16.5 19.6
Tech/Engr 60 10.0 11.7 31.7 23.3 23.3
Mn/Ex/Prf 165 12.1 15.8 15.2 25.5 31.5
Hourly Pay
0-$5.99 56 26.8 26.8 17.9 10.7 17.9
$6-10.99 103 29.1 26.2 17.5 20.4 06.8
$11-15.99 82 19.5 12.2 20.7 25.6 22.0
$16-20.99 55 09.1 20.0 27.3 20.0 23.6
$21+ 82 13.4 12.2 22.0 24.4 28.0
Household Income
Under $10,000 43 34.9 20.9 18.6 18.6 07.0
$10,000-19,999 54 29.6 22.2 11.1 16.7 20.4
$20,000-29,999 98 29.6 20.4 20.4 19.4 10.2
$30,000-39,999 69 20.3 15.9 23.2 21.7 18.8
$40,000-59,999 64 10.9 15.6 28.1 31.3 14.1
$60,000+ 150 10.0 20.7 19.3 19.3 30.7

First, as given in Table 3, we take a demographically defined group of adults, such as females, and show the percentage that fall into each of the five levels of knowledge. Looking at Table 3 it is clear that the less well educated, the younger, minorities, less occupationally skilled, and lower income respondents tend to fall with higher frequencies into the lower levels of Total Knowledge, confirming the data of Table 2. However, it is also clear that in all of these groups, there is a wide range of knowledge. For instance, about one in five Blacks fell in the next to the highest category of knowledge, and about one in eight managers and professionals fell in the least knowledgeable category.

The importance of Total Knowledge in relation to the power indicators of occupation and income is revealed again in Table 4 where each of the five Total Knowledge categories is analyzed to see how adults in a given category of knowledge are distributed across the demographic variables. As indicated, about one in five of the adults in the lowest knowledge category are managers/professionals, whereas over half of those in the highest category of knowledge are managers/professionals. Similarly, only one in six of those in knowledge category 1 earn over $50,000 a year while half of those in category 5 earn that much.

Table 4.
Study 1.

Percentage of respondents in each level of Total Knowledge who are in each of the variable categories. For instance, 51 percent of those in Level 1 are males, 49 percent are females; in Level 5, 44.8 percent are males and 55.2 females.

Total Knowledge Levels
Variable
1
2
3
4
5
Total
Gender
Male
51.0
47.5
51.4
44.8
44.8
47.9
Female
49.0
52.5
48.6
55.2
55.2
52.1
 
100
100
100
100
100
100
Education
0-8
02.0
01.0
00.9
00.0
01.8
01.2
9-12
38.6
37.6
25.2
14.6
06.7
24.4
13-14
32.7
32.7
29.9
26.2
20.0
28.2
15-16
16.8
16.8
30.8
34.0
26.7
25.1
17+
09.9
11.9
13.1
25.2
44.8
21.1
 
100
100
100
100
100
100
Age
16-18
04.9
06.1
02.8
00.0
00.0
02.7
19-24
26.5
15.3
10.5
08.6
02.9
12.7
25-39
42.2
42.9
43.8
41.0
23.5
38.7
40-54
13.6
19.4
21.0
32.4
51.0
27.5
55-64
02.0
04.1
08.6
05.6
10.8
06.3
65+
10.8
12.2
13.3
12.4
11.8
12.1
 
100
100
100
100
100
100
Ethnicity
Caucasian
44.6
64.7
79.4
83.8
90.5
72.9
African-Amer.
05.9
08.8
03.9
04.8
00.0
04.6
Hispanic
31.7
13.7
12.1
05.7
05.7
13.7
Asian
11.9
05.9
03.7
03.8
01.9
05.3
Other
05.9
06.9
00.9
01.9
01.9
03.5
 
100
100
100
100
100
100
Occupation
Unemployed/Student
01.1
01.1
00.0
00.0
00.0
00.4
Homemaker
01.1
02.1
01.0
00.0
00.0
00.8
Laborer/Operator  
22.5
13.7
08.2
06.1
03.0
Skilled/Semi-skilled
23.6
24.2
26.8
21.2
11.1
21.3
Clerical/Sales
23.7
24.2
18.6
16.2
19.2
20.3
Technical/Engineers
06.5
07.3
19.6
14.1
14.2
12.4
Managers/Professionals
21.5
27.4
25.8
42.4
52.5
34.2
 
100
100
100
100
100
100
Hourly Pay
$0-5.99
19.5
20.5
12.8
07.6
14.1
14.8
$6-10.99
39.0
37.0
23.1
26.6
09.8
27.2
$11-15.99
20.7
13.7
21.8
26.6
25.4
21.7
$16-20.99
06.5
15.1
19.2
13.9
18.3
14.6
$21+
14.3
13.7
23.1
25.3
32.4
21.7
 
100
100
100
100
100
100
Household Income
Under $10,000
15.6
09.7
08.2
08.0
03.3
09.0
$10-20,000
16.7
12.9
06.2
09.0
12.0
11.3
$20-30,000
30.2
21.5
20.6
19.0
10.9
20.5
$30-40,000
14.6
11.8
16.5
15.0
14.1
14.4
$40-50,000
07.3
10.8
18.6
20.0
09.8
13.4
>$50,000
15.6
33.3
29.9
29.0
50.0
31.4
 
100
100
100
100
100
100

Analyses holding age, education and ethnicity constant

Table 5 shows relationships among knowledge and literacy practices when knowledge scores were computed after having removed variation in knowledge due to age, education, and ethnicity. This procedure reduced the numbers falling in the lowest and highest categories of knowledge and so the bottom two levels of knowledge and the top two categories of knowledge have been combined in Table 5. The data show a consistent, positive relationship between the number of literacy practices respondents reported engaging in and their scores on the knowledge checklists even when the latter are adjusted for age, education and ethnicity. All were in the predicted direction, with only "reading books or manuals" failing to attain statistical significance (P<.05)

Table 5.
Study 1.

Age, Education & Ethnicity held constant. Relationships of Knowledge Levels to various reading practices that respondents reported engaging in 6-7 times a week. For instance, 46.8 percent of those in Knowledge Levels (1+2) combined reported reading newspapers 6-7 times a week, while 61.4 percent of those in Knowledge Levels (4+5) combined reported reading newspapers that often.(a)

Knowledge Levels
Variable (1+2) 3 (4+5)
Read For Pleasure 39.7 53.1 63.8
Read For Job 31.4 33.6 46.5
Read Book for Pleasure 31.7 39.8 43.0
Read Books or Manuals 17.5 24.8 25.7
Read Newspapers 46.8 56.8 61.4
(a) All relationships are statistically significant by Kendall's taub, P<.05, except reading books or manuals.

The relationships of Total Knowledge to the power indicators of occupation and income, after having removed variation due to age, education, and ethnicity, are shown in Table 6. The proportion of those in the lower two knowledge categories who are laborers was twice that of those in the upper two categories of knowledge. About three out of ten of those in the lowest two categories of knowledge are managers/professionals, while four out of ten of those in the highest two categories of knowledge are in these manager/professional jobs. Similar findings hold for income for those earning less than $10,0000 and those earning $40,000 or more.

Table 6.
Study 1.

Age, Education & Ethnicity held constant. Relationships among Knowledge Levels, Occupational Status, and Annual Income. Data are percentage of people in each Knowledge level who are in each of the demographic categories. For instance, 29.9 percent of those in Knowledge Levels (1+2) combined were Managers/Professionals, while 41.3 percent of those in Knowledge Levels (3+4) combined were in that occupational category.(a)

Knowledge Levels
Variable (1+2) 3 (4+5)
Occupation
Laborer/Operator 13.0 13.5 05.8
Skilled/Semi-skilled 23.4 20.9 20.0
Clerical/Sales 13.0 12.2 14.2
Technical/Engineers 20.8 20.9 18.7
Managers/Professionals 29.9 32.4 41.3
  100 100 100
Annual Income
Under $10,000 13.8 06.9 05.8
$10-20,000 15.1 07.5 11.6
$20-30,000 22.4 18.9 19.4
$30-40,000 13.8 13.8 16.1
$40-50,000 07.2 18.2 14.8
>$50,000 27.6 34.6 32.3
  100 100 100
(a) Relationships are statistically significant, P<.01, by Kendall's taub. Not in labor force categories (unemployed, student, homemaker) were removed from calculations due to very few cases (less than 2 in cells) from occupation cross-tabulation.

Together, Tables 5 and 6 indicate that more generally knowledgeable adults engage in greater amounts of reading, they hold higher status occupations and they earn higher levels of income, even when general knowledge scores are adjusted for differences in age, education and ethnicity.

Study 2: Domain Specific Declarative Knowledge and Power

In Study 1, subject's general declarative knowledge was assessed using shortened versions of the Stanovich checklists for knowledge of famous authors, magazine titles, famous people, and vocabulary words (West, Stanovich, & Mitchell, 1993). It provided information about respondent's general declarative knowledge taken from samples of the knowledge of the "mainstream" or "dominant" society.

One critic of the Stanovich checklists has referred to them as being similar to the game of "trivial pursuit," and that there is essentially no "real world" value to showing that people possess such culturally "biased" knowledge (Taylor, 1994). However, the results of Study 1 counter this argument by showing "real world" relationships among so-called "mainstream," "dominant" culture knowledge and the acquisition of power.

While some may question the utility of cultural knowledge as defined in Study 1, there is no questioning the fact that citizens in a democracy need to possess considerable political knowledge to make informed choices among political candidates to represent them and to pursue their vital interests through political activities. Political knowledge is not "trivial." Therefore, to further examine the role of declarative knowledge in achieving and exercising power, Study 2 examines for the first time the relationships of domain specific, declarative, political knowledge to two indicators of power, income and political activities, when age, education, and ethnicity are held constant.

To establish construct validity of the declarative knowledge measures as indicators of political knowledge, conventional measures of political knowledge used in earlier studies of political activity were administered (Delli Carpini & Keeter, 1993; Newman, Just, & Crigler, 1992). If positive correlations of the checklist and traditional measures of political knowledge are obtained, this offers convergent validity (Messick, 1989, p. 5) for the checklist knowledge measures. Additionally, checklist measures of declarative knowledge were related to a four-item rating scale that attempted to directly gauge subject's sense of power. If positive correlations are obtained between the checklist measures of declarative political knowledge and perceived power, this provides additional convergent evidence for the validity of the knowledge checklists and the relationship of knowledge to power.

Method

Subjects

Data for Study 2 followed the same general random-digit-dialing telephone survey procedures as used in Study 1. Structured telephone interviews were conducted with 644 English speaking persons selected to represent a cross-section of the population 18 or older who could be reached by residential telephone in the greater San Diego, California, area during late spring and early summer, 1995. Up to four call-backs were made resulting in an overall completion rate of 50 percent, a rate comparable to or surpassing that for the better survey research firms in the area. Fewer than five percent of respondents contacted were eliminated due to inability of the subject to respond to the protocol in English. Respondents were generally well educated and affluent, reporting 14.5 (SD=3.2) years of formal schooling completed and mean household income of $34,380 (SD=$12,244). Mean age was 41.8 (SD=17.2). As indicated in Table 1, the sample generally corresponded to the sample of Study 1 and the 1990 U. S. Census data for San Diego, although minorities and less well educated were slightly underrepresented.

Instrumentation

Using the same type of checklist approach as used in Study 1, a series of political knowledge checklists with foils was developed drawing on traditional bodies of political knowledge domains used by political scientists in studying political activism and voting (Delli Carpini & Keeter, 1993; Neuman, Just & Crigler, 1992). The measures of political knowledge were designed to tap those aspects of politics that are relevant for meaningful personal political action. For the present study, five political domains were identified including (1) Political Leaders, that is, actors or activists engaged in political processes, (2) Political Policies, policies produced by various political systems, (3) Political Groups, i.e., groups such as the National Organization for Women who are active in political movements, (4) Government Organizations, i.e., domestic or foreign government organizations such as the Bureau of Indian Affairs, and (5) Political Events such as the Three Mile Island or Exxon Valdez incidents.

As in Study 1, the approach presented subjects with a series of declarative knowledge stimuli. The Appendix, Tables A6-A10, shows the five domains of political knowledge, the questions asked to elicit responses from subjects, mean percent correct and standard deviations for each item, and the percentage of subjects for each item falling into each of five levels of total political knowledge (see below).

Reliability data for each of the subscales and the Total Political Knowledge scale were computed using Cronbach's Alpha. The reliability coefficient for the Political Leaders scale was .73; for Political Policies, .63, Political Groups, .61, Government Groups, .58, and Political Events, .62. For the Total Political Knowledge scale the reliability coefficient was .88.

Defining political knowledge levels

As in Study 1, the Total Political Knowledge checklist scores of subjects were used to define five levels of Total Political Knowledge. Scores for each subscale and for the Total Political Knowledge scale (all items from all subscales) were computed by subtracting the percentage of foils misidentified as real from the percentage of real items correctly identified as real. The computation was designed to correct for guessing (West, Stanovich, & Mitchell, 1993). To obtain the five levels, each person's adjusted percent correct score on each checklist was added together to give a Total Political Knowledge score. Then the mean adjusted, percent correct score for Total Political Knowledge for the combined sample was calculated. This score was used to divide the sample into five groups or levels using the mean (53.36 adjusted percent correct) and the standard deviation (SD, 21.97 adjusted percent correct) for the total sample. Political Knowledge levels were defined in adjusted percent correct scores from low to high proficiency as: Level 1= scores at -1.0 SD below the mean or lower (0-31 adjusted percent correct), Level 2 = scores between -.5 to -.1.0 SD (32-42 adjusted percent correct), Level 3 = scores between .5 SD (43-64 adjusted percent correct), Level 4 = scores between +.5 to +1.0 SD (65-75 adjusted percent correct), and Level 5 = scores from +1.0 SD and above (75-100 adjusted percent correct).

Conventional measures of political knowledge. To validate the political knowledge checklists as measures of political knowledge, political knowledge was also measured using a series of questions from political science studies (Delli Carpini & Keeter, 1993; Neuman, Just & Crigler, 1992). Subjects were asked which of the two parties "...is usually regarded as the most conservative" (Republican), "...currently has a majority in the U.S. House of Representatives" (Republican), "...had a majority ... before the last election" (Democratic), and "...currently has a majority in the U.S. Senate" (Republican). They were also asked the name of the first ten amendments to the Constitution (Bill of Rights), the number of times a person can be elected President (2), and the length of terms for the U. S. President (4 years), a U.S. Senator (6 years), and a U.S. Representative (2 years). For this nine item scale, called Conventional Political Knowledge, the Alpha reliability was .60.

Reading and media practices. To determine the role of media in developing political knowledge subjects were asked to indicate how many days a week they read a newspaper, how many hours a day they listened to the radio, and how many hours a day they watched television. They were asked to indicate about how many national network news programs, such as CBS, NBC, or ABC news they saw in an average week, how many local news programs they watched in a week, how many network news magazines such as 60 Minutes, 20/20, Frontline, Dateline, or Eye to Eye they watched during a week, and how many Public Broadcasting news programs such as the McNeil Lehrer News Hour, Washington Week in Review, or National Business Review they watched in an average week.

Political interests and activities. To determine relationships of political knowledge to political interests, subjects were asked to rate on a four point scale how interested they were in politics and public affairs. They were also asked to rate on a four point scale the amount of attention (high to low) they paid to politics or political issues when they watch television or read the newspaper.

Two questions were asked to determine the frequency on a four point scale (very often to never) with which subjects voted in local or national elections (combined into one voting score, Alpha reliability = .91), and thirteen questions were asked about various activities (encouraged others to vote for one of the parties or candidates, worked for one of the campaigns, talked about politics with family members, etc.) using the same four point scale. The mean score for the thirteen questions was used to form a political activities score for each subject that could be related to political knowledge (Alpha reliability = .83).

Measures of perceived power. To assess subject's sense of power directly, a scale of "powerlessness" taken from Kohn (1976) was used. Two items asked for respondents to state whether they 1-agree strongly, 2-agree, 3-disagree, 4-disagree strongly or 9-don't know. Scores of 9 were excluded from analyses. One of these items said, "I generally have confidence that when I make plans I will be able to carry them out." The second said, " There are things I can do that might influence national policy." A third item asked, "Do you feel that most of the things that happen to you are the result of your own decisions or of things over which you have no control?" and were scored 1, meaning that things happened due to their own control, or 2, meaning things happen due to decisions over which they had no control. The fourth and final item asked, "How often do you feel powerless to get what you want out of life?" It was scored 4-very often, 3-often, 2-sometimes, 1-rarely/never. Summed over the four items, low scores indicate a feeling of power, high scores feelings of powerlessness. For the present analyses, signs of correlations were reversed to show that increments in knowledge correlate positively with increments in perceived power. Cronbach's alpha for the power scale was . 39, a low but usable degree of reliability given that this is only a four item scale.

Results

Validity indicators. Convergent evidence for the validity of the Total Political Knowledge checklist method as a measure of political knowledge was obtained by the finding of a significant, positive correlation of Total Political Knowledge with Total Conventional Political Knowledge (r =.47 ; p <.001). Convergent evidence that the Total Political Knowledge checklist scale is an indicator of perceived power was obtained by the significant positive correlation between the total perceived power scale and Total Political Knowledge checklist scores (r = .22; (p <.001).

Knowledge checklists. Tables A6-A10 in the appendix show each of the items in each of the knowledge checklists along with the mean percentage correct and standard deviations for each item as well as the average correct and standard deviations for the sum of each checklist. These tables show that the Political Leaders checklist had the highest average correct scores (74 percent) and the Political Policies checklist had the lowest scores (31 percent). The remaining scales were about equal in average difficulty.

Tables A6-A-10 also show the percentage of subjects in each of the five Total Political Knowledge levels that got each item correct. For Political Leaders in Table A6, 25 percent of those in Level 1 knew of John Major, the Prime Minister of Great Britain, while almost 90 percent of those in Level 5 knew of him.

Correlational analyses. There were significant positive relationships of Total Political Knowledge checklist scores to political interests (r =.28, p <.001) and to the amount of attention adults said they paid to politics or political issues when they watch television or read the newspaper (r =.11, p <.002). There were no significant relationships of Total Political Knowledge to the frequency of listening to the radio or watching television, nor to the types of news programs watched on television, with one exception. The number of Public Broadcasting news programs watched during an average week was significantly correlated with the Total Political Knowledge checklist scores (r =.14, p <.001

Interrelationships of demographic variables, political knowledge checklist scores, literacy practice and power indicators (household income, voting and political activism) are given in Table 7. These data are consistent with those of Study 1 in showing significant, positive correlations among the knowledge scores and the power indicators.

Table 7.
Study 2:

Correlations among demographic variables, knowledge, literacy practice and power indicators.

Variables 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Education
1.00
.06*
.27
.21
.25
.25
.25
.36
.11
.16
-.06*
.18
.27
.15
Age  
1.00
.18
.20
.15
.31
.19
.27
.29
          
Leaders    
1.00
.41
.52
.48
.46
.74
.19
.24
-.07*
.19
.29
.06*
Policies      
1.00
.46
.46
.41
.72
.19
.16
-.07*
.16
.27
.10
Groups        
1.00
.5
.44
.8
.13
.29
-.15
.18
.24
.06*
Organizations          
1.00
.49
.79
.18
.20
-.06*
.19
.34
.18
Events            
1.00
.76
.16
.2
-.10
.14
.24
.14
Total Political Knowledge            
1.00
.22
.22
-.13
.23
.369
.14
Literacy Practicea              
1.00
.11
-.11
.17
.25
.16
Ethnicity                  
1.00
.08
.17
.28
.06*
Gender                    
1.00
-.03*
.09
.04*
Annual Household Income                
1.00
.19
.09
Voting                    
1.00
.40
Political Activism                        
1.00
*Not significant at p <.05 or lower; all others statistically significant. a Literacy Practice is frequency of reading newspaper in a week.

To sum up these correlational data, better educated, older, caucasion adults who read newspapers and watch Public Broadcasting news programs more frequently, tend to have greater interests in politics, they are more knowledgeable about politics, they have higher household incomes, they vote more and they are more politically active.

Analyses by knowledge levels

Following the method of presenting of results in Study 1, analyses are presented here that show how different variables are distributed in the five different Total Political Knowledge categories defined above.

Table 8.
Study 2.

Percentage of respondents in each demographic group falling into each of five levels of Total Political Knowledge. For instance, 5.7 percent of those with 17+ years of education were in Level 1 while 28.6 percent were in Level 5.

Total Political Knowledge Levels*
Variables
N
1
2
3
4
5
Total Sample 644 16.1 12.6 35.9 19.7 15.7
Normal Curve   16.0 15.0 38.0 15.0 16.0
Education
1-8 22 22.7 09.1 45.5 04.5 18.2
9-12 160 26.9 15.0 37.5 16.9 03.8
13-14 129 24.0 15.5 33.3 17.1 10.1
15-16 228 08.3 12.7 34.6 23.2 21.1
17+ 105 05.7 05,7 37.1 22.9 28.6
Age
16-18 16 56.3 12.5 31.3 00.0 00.0
19-24 82 35.4 17.1 31.7 11.0 04.9
25-39 250 16.0 13.6 40.8 16.0 13.6
40-54 132 04.5 07.6 33.3 26.5 28.0
55-64 63 12.7 11.1 34.9 25.4 15.9
65+ 91 08.8 14.3 31.9 27.5 17.6
Ethnicity
Caucasian 477 10.7 11.7 36.5 22.6 18.4
African-Amer. 44 29.5 15.9 29.5 15.9 09.1
Hispanic 65 35.4 13.8 36.9 07.7 06.2
Asian 32 34.4 18.8 37.5 06.3 03.1
Other 14 21.4 14.3 35.7 14.3 14.3
Household Income
Under $10,000 51 21.6 19.6 35.3 13.7 09.8
$10,000-19,999 77 32.5 16.9 22.1 15.6 13.0
$20,000-29,999 92 16.3 17.4 35.9 19.6 10.9
  93 09.7 12.9 45.2 16.1 16.1
$40,000-49,999 60 11.7 10.0 31.7 28.3 18.3
$50,000+ 169 09.5 07.7 39.1 23.1 20.7
Voting Behaviora
Low 152 32.9 16.4 34.9 12.5 03.3
$30,000-39,999 198 10.6 14.1</