Method
This study uses federal data from the Current Population Survey (CPS) to examine library use. The following are the questions the study attempted to answer:
- Do library usage rates vary between immigrants from different regions of the world, as represented by households of immigrants from these regions?
- In the United States, are households of native-born residents more likely to report public library use than those of foreign-born residents?
- What are some sociodemographic factors that affect rates of household public library use?
The CPS is a household sample survey that is conducted several months out of every year by the U.S. Bureau of the Census to collect labor force data. Other information is occasionally gathered by including supplemental questions. Data for the present study were collected with the CPS October 2002: School Enrollment/Library Use supplement. The sample universe for the CPS is the U.S. civilian noninstitutionalized population living in households. The sampling method is a multistage probability sample of households from all fifty states and the District of Columbia, and data are gathered using a series of structured, closed-ended questions. Households are selected, then data are gathered about all persons in the household. One person in the household becomes the “reference person” and answers questions for all members of the household if possible. The reference person must be fifteen years of age or older. Since households are chosen and not individuals or families, data are gathered about all household members no matter their relationships to each other. The October 2002 dataset contained 159,887 respondents that lived in 57,148 households.29
Some of the questions regularly asked by the CPS concern each individual in the household, such as the country in which a person was born, country in which each parent was born, education level, and other demographic information. Other questions are asked about the household, such as “Is there a telephone in the household?” All of the library use questions in this supplement were asked of the household, and not of individuals in the household. It is not possible to determine which individuals in a household used the public library in the past month, only whether anyone in the household used the library. For this study a household variable was created that specified the immigrant status of the household overall. Households were coded as consisting of U.S.–born persons, immigrants by world region of origin, or mixed. Households in which children of immigrants were under age 25 were coded to the immigrant region of the parents and not as mixed households, even though children of immigrants born in the U.S. are considered native-born Americans. Immigrant households in which U.S.-born children or other U.S.–born adults age 25 and older resided were classified as mixed households and not analyzed in this study. Households from mixed world regions, but not containing native-born Americans age 25 or over were coded to the world region of the reference person.
Native-born U.S. citizens who were born abroad were coded in the U.S.–born household category. These individuals might, for example, include those born to military families stationed abroad, travelers, and others whose parents were abroad at the time of their birth. Respondents coded as being from “elsewhere” were eliminated from the analysis. Respondents from Australia, New Zealand, Fiji, and the Pacific Islands were also not analyzed because there were too few to make meaningful comparisons.
Statistical Methods Used
Two techniques of data comparison were used in the analysis. The main discussion compares percentages of different groups’ use of the public library. Secondly, chi-square was used to determine whether the amount of difference between two groups was likely to have occurred from chance alone, or whether the difference was large enough to suggest an actual difference between groups. When the difference between groups is large enough to be unlikely to have happened by chance, that is referred to as statistically significant. The chi-squares in this analysis were conducted on the frequencies (actual numbers) and were calculated for 2 x 2 contingency tables at a .05 alpha level. Chi-square is sensitive to sample size, so this statistic could not be calculated for the native-born U.S. citizens because fifty thousand households were too many for the test to have meaning. Since the immigrant groups were represented by far fewer households, it was appropriate to conduct chi-square on differences between these groups. In the data analysis discussion, when it states that the difference between certain groups was significant, it is referring to the chi-square calculation between those groups showing statistical significance. Chi-square was used in the data analysis section on library use by immigrant groups, but it was not used in the section on additional demographic variables because the multilevel analyses resulted in smaller sample sizes per table cell. In addition, percentages were not calculated when sample size for a cell was twenty-five or fewer households.
Limitations of the Study
There are limitations to using available data. Researchers using available data are restricted to the questions asked by the original researchers. Questions often do not represent exactly what the current researcher wants to know. How the data were coded and entered is also under the control of the original researchers and may not be as useful for secondary analysis as it might have been if coded in a different manner. A specific limitation of the data in this study is that the library questions were asked at a household level. A more in-depth picture of immigrant library use could have been established if the questions had been answered by individuals in the household. There were no qualitative data available for this study. As a result, only the general picture of library use is available, but not the type of individual reasoning that might have been available with additional qualitative data.
Data Analysis
Three questions were asked about general library use. “In the past month, that is since (month) (day), has any member of your household used a public library or bookmobile for any reason?” Respondents who answered “no” to the first question were then probed for additional affirmative responses with the following question, “How about to borrow materials, take a class, to use the computers or for activities for children?” Responses to these two questions were combined to get an overall figure for public library use in the past month. Respondents who answered “no” to these two questions were then asked, “Has anyone in your household used a public library or bookmobile in the past year, that is since October, 2001?” Answers to the library use questions were cross-tabulated with the immigrant household variable to produce the following figures.
As can be seen in figure 1, respondents from some geographic regions were more likely to have used the public library in the past month than respondents from other regions. Differences between groups had a range of 21.5 percentage points from a low of 23.2% for European households to a high of 44.7% for South Asian households. Excluding these two outliers, all other households fell within 10 percentage points of each other, from Central American or Mexican households at 27.2% to Southeast Asian households at 37.3%.
Not all of the differences between groups were large enough to be meaningful. Using the lowest (European) and highest (South Asian) library use as the comparison groups for calculation, the point where the chi-square test showed a significant difference between groups divided them into low, middle, and high categories of library use in the past month as follows:
- Low use: Households classified as European, Central American or Mexican, Canadian or other North American, Caribbean, East European, and Middle Eastern
- Middle use: Households classified as African, U.S. native-born, East Asian
- High use: Households classified as South Asian, Southeast Asian, and South American