Tom W. Smith
National Opinion Research Center
A Grant Report Submitted to the Security Awareness Area of the Personnel Security Research Institutional Award Program
Defense Personnel Security Research Center (PERSEREC)
While less than two percent of the American population, at any given time, actually holds a security clearance, most adult Americans have formed opinions about security, clearances, and government secrecy. We should be able to discern what the American public thinks and provide useful public opinion information to security policy-makers and practitioners. To accomplish this the Defense Personnel Security Research Center has sponsored a series of studies designed to assess the state of various security-related opinions and the likely direction of these opinions in the future. In 1991 we published Security Awareness and the Climate of Public Opinion: With Special Attention to Financial and Credit Issues. In 1993 we issued the report Security Awareness and the Climate of Public Opinion: An Analysis of Recent Trends. In both of these studies we analyzed survey data collected by other researchers around the nation.
This report documents the results of our latest effort in which we developed our own survey ites to tap into issues we thought particularly relevant to Department of Defense and the Intelligence Community. These items were included in the 1994 General Social Survey fielded by the National Opinion Research Center. This survey was administered to a national sample of the adult household population of the United States. Fifteen hundred respondents were asked their opinions regarding the need for secrecy, the extent to which secrecy is over-used, reporting adverse information about co-workers, and the appropriateness of collecting various types of personal background information before granting access to classified information.
While the data reveal concern that the government classifies too many documents as secret and top secret, US citizens favor maintaining a high level of secrecy surrounding technology with military uses. Importantly, respondents generally expressed a high level of support for the government's right to ask personal background questions. It appears that current government personnel security policy in connection with granting access to classified information is in synch with the climate of public opinion. It seems unlikely that major public outcries against current background question procedures will occur.
Roger P. Denk
To examine these issues the National Opinion Research Center (NORC) included a battery of 11 items on its 1994 General Social Survey (GSS). The items were drafted by researchers at the Defense Personnel Security Research and Education Center in collaboration with NORC. The GSS is a nationally representative, full-probability sample of adults living in households in the United States. The 1994 GSS had a response rate of 78% and a sample size of 2992. The security items were administered to a random half of the total sample (the "B" sample), a total of 1474 respondents. Data were collected in February-April, 1994. Full technical details on the GSS are found in Davis and Smith, 1994.
There is even a stronger pro-security consensus when it comes to screening applicants for secret or top secret clearance (Table 1 - Q.4). Over 90% agree that the government should "have the right to ask ... detailed, personal questions" about criminal arrests and convictions (98%), illegal drug use (96%), mental health history (95%), and alcohol use (93%). Somewhat smaller majorities endorse questioning about financial and credit history (82%) and foreign relatives and friends (78%). For all of these reasons except for the case of foreign connection, the majority definitely thinks the government has the right to question applicants. Only on the topic of sexual orientation in the public divided: 47% support inquiring about this topic, 50% oppose, and 2% are uncertain. Similarly, when asked to balance an applicant's right to privacy vs. the government's need to collect personal background information, the public overwhelmingly favors the government 80% to 15% with 5% unsure (Table I - Q. 5). At least when it concerns those who will be handling secrets, the public clearly and strongly favors national security over individual privacy and backs extensive background chocks.
The public also places the protection of secrets above the protection of co-workers. People were asked to consider a situation in which there is "a conflict between loyalty to one's employer and loyalty to a co-worker vho is violating rules protecting secret and top secret information." In that circumstance 41% said the person should "report tho co-worker to a company official" and another 41% favored asking the co-vorker to stop, but reporting him or her if the misbehavior did not stop. Only 14% favor a weaker response. While people are evenly split between immediately reporting the errant co-worker and giving him or her a chance to stop the violation, the relative leniency of the latter group probably reflects the fact that the co-worker is not described as acting traitorously or even actually compromising secrets, but only as "violating rules protecting" secret information.
In sum, although the public believes the goverrment classifies too many documents, people overwhelmingly back the protection of military-related technology, the detailed vetting of applicants for security clearance, and the reporting of employees who are violating security procedures.
However, a factor analysis with a varimax rotation does indicate that there is not a simple and general security dimension, but three somewhat distinct factors (Table 3). The first factor consists of the four asking items with the hiqhest level of support (drugs, crime, alcohol, and mental health). The second factor is made up of the three asking items with lower support (sexual orientation, foreign connections, and finances). Askinq about alcohol use, which is in the middle in terms of approval, loads fairly hiqhly on both factors. The third factor consists of the remaining items (protecting military-related technology, classifying documents, reporting co-workers, and privacy vs. security checks). It is the weakest of the three factors (top loadings of only .65) and the item on individual rights to privacy has a low loadinq of -.43.
Table 4 shows the scales that we constructed from the 11 security items. The first two items are two scales made from the seven items on background checks. The first excludes Don't Knows and the second retains then by recoding them to the middle. We used the latter scale in subsequent analysis since it preserves more cases and had marginally hiqher associations with independent variables. The third scale consists of the remaining security items. For both scales low values represent pro-security responses and high scores indicate those opposed to protecting secrets and background checks.
Support for the military is strongly associated with supporting security measures. Favoring more defense spending and having more confidence in the military are related to both more support for background checks and pro-security positions.
There is only a weak and irregular association between confidence in governmental institutions and security measures. Confidence in the executive branch of the federal government is not related to either security scale. Confidence in the legislature has non-linear and inconsistent relationships with security. Background checks are most supported by those with a great deal of confidence, but those with only some confidence show the least support. On the protecting scale support is highest for those with only some confidence and lowest for those with a great deal of confidence.
Generalized patriotism or national pride is strongly associated with favoring security measures. Those who are proud of being an American are much more supportive of both extensive vetting and the other security measures. Those who think America is better than other countries also are more supportive of background checks. However, for the other security measures the relationship is more complex. Pro-security positions are stronger among those who think America is both better and worse-off than other advanced countries and lowest for those who see America and others as equally well-off. Those who see America as ahead may want to maintain that lead by protecting our secrets and security and those who believe we are behind may see these measures an necessary to improve our position against foreign competition in general and espionage in particular.
Nativist patriotism has a weaker and less consistent association with security measures. Those who favor assimilation over pluralism and decreased immigration are more supportive of background checks, but attitudes towards government either assisting pluralism or assimilation are not related to approving of rigorous clearance procedures. Support for other protective security measures is also higher among those favoring assimilation and decreased immigration and somewhat greater among those who think groups should assimilate with goverrmental assistance.
Similarly, the four English vs. bilingualism items show that those who favor English are more supportive of both security scales.
Political conservatives, Republicans, and Bush voters in 1992 are all more in favor of both security measures than those with centrist or liberal political leanings.
Religious conservatives (those who belong to Fundamentalist denominations and those who personally believe in Bible inerrancy) and those who attend church frequently are more supportive of background checks. The protective security scale has a weaker relationship with religion. Fundamentalists are more pro-security, but beliefs about the Bible has an irregular association and church attendance is not significantly associated with protective security measures.
Religious conservativism and involvement are probably related to strict vetting because Fundamentalists believe in personal sin and think that rigorous steps have to be taken to identify and weed out sinners. Since the element of moral weakness is less apparent in the protective security scale, these religion variables do not clearly differentiate in this case.
Crime and Punishment
Those who favor the tough punishment of criminals and assisting law enforcement efforts generally back stronger security measures. For background checks the association is less pronounced since only in the case of favoring capital punishment is the relationship statistically significant. For the protective security scale, more stringent measures are backed by those who want tougher courts, capital punishment, and more spending for law enforcement.
Those who value obedience highly as a value to instil in children tend to favor both extensive background checks and strong counter-espionage measures. The strongest association comes for the measure rating "obedience and respect for authority" as the top child values. It is likely the "respect for authority" dimension, which does not appear in the tvo other obedience items, de-emphasizes parental obedience and focuses attention more on obedience in general. This is supported by the fact that the obedience/respect item is more highly associated vith general patriotism than are the two other obedience items.
Those who oppose civil liberties (free speech, employment as a college teacher, and allowing a book in the public library) for political extremists and social outgroups (those who are Communists, militarists, racists, anti-religionists, or homosexuals) tend to favor extensive background checks. Those who consider free speech a more important goal than maintaining public order and those with less confidence in the press are more supportive of other security measures. The other relationships tend to lean in a similar direction, but are not statistically significant.
Personal Freedom and Deviance
Those who oppose exressive freedom and deviant or non-traditional life styles tend to favor background checks and other protective security measures. People who disagree that "it is wonderful that young people today have greater freedom to protest against things they don't like and to 'do their own thing'," those opposing the legalization of marijuana, and those who think homosexuality is wrong and is a matter of individual choice are more in favor of background checks. Support for other protective security measures lean in the same directions, but only marijuana legalization and morality of homosexuality have statistically significant relationships.
Counter to expectations judgments on whether people are trustworthy, fair, and helpful are not related to support for extensive vetting. Moreover, the two weak, but statistically significant, associations between trust and helpfulness and the protective security scale are the opposite of the hypothesized direction. Those who believe most people are trustworthy and helpful are more supportive of security measures.
Work and Finances
Low satisfaction with one's job or finances and recent financial downturns do not relate to either security scale. While those in security jobs who are dissatisfied with their employer or in a financial bind may well be greater security risks, generalized occupational or financial dissatisfaction does not undermine support for security measures.
Of the nine demographics examined only gender is consistently related to both security scales. Men are more supportive of background checks and protective security measures than are women. Background checks are more supported by older adults, those from the South, and those vith high school degrees or some college. It is unrelated to race, nativity, community type, or vocabulary score. Other protective measures are favored by Whites, college graduates, those with more verbal ability, and the native born. It in unrelated to age, region, community type, or parental nationality.
Education in particular has a very complex relationship to security items. Examining each of the 11 security items individually shows that education is unrelated to attitudes towards the classification of documents, the protection of technology, and asking about drug use and foreign connections. The better educated are more supportive of reporting errant co-workers, screening vs. personal privacy, and asking about crime. For asking about finances, mental health, and drinking support increases from among those with less than a high school education to a peak among those with some college, but then declines slightly among those with a college degree. Finally, support for asking about sexual orientation declines with education.
With multivariate controls support for extensive background checks is greater among males, Southerners, older adults, and political and religious conservatives. Rigorous vetting is more favored by those who are for the military, law and order, obedience and respect for authority, and patriotism. It is less supported by those who are more in favor of civil liberties, personal freedom, and deviant and permissive life styles.
Favoring other protective security measures is greater among the better educated, Whites, and political and religious conservatives. Support for security is higher among those who are pro-military, patriotic (both the proud and monolingual), and anti-crime. Other bivariate predictors discussed above (e.g. gender and obedience) do not appear as independent predictors in multivariate analysis.
The two security scales share a number of attitudinal correlates in common (being conservative, pro-military, anti-crime, patriotic), but notably differ in their overall pattern of association. First, demographic associations are quite different. Gender, region, and age are related to background checks, but education and race correlate with other security measures. Second, even vhen there is general consistency, the details are often divergent. For example, support for law and order is related to pro-security positions, but capital punishment is associated with background checks and favoring tougher courts is related to protective security measures. Similarly, religiousness is associated with favoring security efforts, but more church attendance is associated with background checks, while being a member of a Fundamentalist church is related to other security measures. As a result, it is unwise to make many sweeping statements about the attributes and correlates of pro-security positions. Instead, more limited and nuanced assessments are needed.
This pattern indicates that most members of the public value security highly, favor prudent steps to maintain security, and even back intrusive and restrictive government policies where secrets and national security are involved. However, in most of the items examined here the public is assessing what security measures should be allowed in regards to those who handle or want to handle secrets. In such cases it tips the balance in favor of security and away from individual privacy and personal freedoms. In other circumstances the public would probably give more emphasis to these concerns (Smith, 1993). The majority that agrees that the government classifies too many documents as secret represents such a tipping away from security in favor of public access to information.
Security is not one, simple dimension in people's minds. Even among these 11 items, three related, but distinct, factors appeared: a high and low approval of background checks and other protective security measures.
Moreover, the complexity of security attitudes is demonstrated by the distinctive pattern of associates that the security scales have. In particular, different demographics relate to the background check and protective security scales and while these scales share a number of important attitudinal associates (being pro-military, politically and religiously conservative, patriotic, anti-crime) , the details often are different. Thus, while one knows some of the general factors that associate with more support for security measures in general, it is more difficult to stipulate the precise attitudes that will lead to support for specific security measures. The complicated association of education with the 11 measures illustrates this point.
1. The negative signs in Table 2 reflect items coded in the opposite direction. When coded in a consistent manner, all correlations are positive and all but 7 of the 55 inter-item correlations are statistically significant.
2. While the asking question formed two factors, they differ essentially between items with high approval (drugs, crime, mental health, and drinking) and those with low to medium approval (sexual orientation, foreigners, and finances). Since these factors inter-correlated at .46-.47 and did not differ greatly in their association with other variables, they have been combined together into a single asking scale.
Strongly agree 34.8% Agree 40.9 Neither agree nor disagree 11.7 Disagree 7.1 Strongly disagree 3.0 Don't Know 2.5
2. Given the world situation, the government protects too many documents by classifying them as SECRET and TOP SECRET. Do you ...
Strongly agree 13.9% Agree 42.0 Neither agree nor disagree 17.1 Disagree 17.5 Strongly disagree 4.4 Don't Know 5.0
3. When faced with a conflict between loyalty to one's employer and loyalty to a co-worker who is violating rules protecting SECRET and TOP SECRET information, a person should...
Report the co-worker to a 40.8% company official Ask the co-worker to stop, 5.9 but do nothing further Ask the co-workers to stop, but report 41.4 him/her if the behavior continues Mind one's own business and not 8.3 get involved Don't Know 3.5
4. Before giving an individual a SECRET or TOP SECRET clearance, the government should have the riqht to ask him or her detailed, personal questions in the following areas:
Definitely Probably Probably Definitely Don't Should Should Should Not Should Not Know Financial and credit history 54.9% 26.9 10.9 5.8 1.4 Criminal arrests and convictions 89.8% 8.4 0.6 0.4 0.8 Illegal drug use 84.7% 11.6 1.9 1.0 0.8 Mental health history 76.9% 17.9 3.3 0.7 1.1 Alcohol use 68.5% 24.1 4.6 2.1 0.7 Sexual orientation 28.7% 18.7 24.8 25.5 2.3
5. When faced with the conflict between an individual's right to privacy and the government's need to collect personal background information before giving a SECRET or TOP SECRET clearance, the government should ...
Favor protecting an individual's right to privacy by not gathering information 14.9% Favor the government's need to protect security by gathering information 80.1 Don't Know 5.0
[Tables 2 through 6 on statistical correlations and distributions are not included here, but may be requested from the FAS Project on Government Secrecy.]