|
Adkins, D. & Bala, E. (2004). Public library outreach as a function of staffing and metropolitan location [Electronic version].
Library & Information Science Research, 26(3), 338-350.
Abstract
This study was based on literature indicating that library staff and service population status (metropolitan or nonmetropolitan,
based on U.S. census definition) may affect the provision of outreach services. Data from a 1999 survey of Arizona public
libraries was analyzed and presented in three models showing the effects of: librarians alone, librarians and clerks(combined
to represent ‘staff’), and finally library staff and metropolitan status. Researchers found a positive correlation
between the number of librarians on staff and provision of outreach services. They found also that, though metropolitan areas
also showed increased outreach activities, this was not proven to be due exclusively to metro status but may have been due
simply to the likelihood of a greater number of staff in the metropolitan libraries.
Methodology
The survey used was sent to 168 Arizona public libraries. Tribal and prison libraries were excluded, since they have different
missions and serve different populations than public libraries. Bookmobile staff was excluded from the survey also, since
bookmobiles themselves are a type of outreach service. . In addition to survey results, researchers analyzed the number of
librarians employed, clerks employed, a combination of the two, and whether the library service area was metropolitan. Three
variables had a possible range of values, and one variable was categorical and dichotomous (metro/not metro).
Because the survey questions of outreach services required a ‘yes’ or ‘no’, OLS regression was deemed
inappropriate for this study; researchers used logistic regression, the Wald statistic and likelihood ratio tests, Nagelkerke’s
R2 and chi-square distribution in their analysis. A test for collinearity was also done, but researchers concluded that variables
were not associated closely enough to affect the logical regression analyses.
LIS/IT use of methodology
Any study involving human behavior is subject to the vagaries of human nature. Because of the predictive quality of logical
regression, this type of methodology could be used in studies of a more exploratory or qualitative nature, such as my research
project for this class.
It would be appropriate methodology for a study intended to gather data by survey, questionnaire or interviews, and analyzed
to present findings supporting or disproving hypotheses, encouraging further research.
Flaws
The data used in this study was gleaned from a survey intended to measure library services (including outreach) to Hispanic
patrons. The authors mention in the article that only questions pertaining to outreach to the general population were used,
but the questions originally designed for a different purpose may have elicited different responses than questions designed
specifically for this study may have done. I don’t know whether this is common or generally acceptable practice; it
seems that this would diminish the validity of the study. Researchers could conduct a separate survey, or provide more information
about the questions used as basis for this study.
Outreach services are not clearly defined – there seems to be no generally accepted consensus as to what constitutes
‘outreach’. This may well have led to a distortion of the data – e.g., one library may have counted their
marketing efforts as outreach, where another might well have the same program yet not consider it outreach. As stated in the
article itself, “a systematic vocabulary for outreach needs to be developed, and outreach efforts need to be included
in major library data collection projects”.
One other flaw that may ultimately have a major effect on the conclusions drawn by the researchers is the existence of other
variables, which they acknowledge may well affect the logistic regression models. Reduced outreach services in rural areas
may be due to distance or economic factors as much as to lack of staff. Demographics – not just the size, but the educational,
cultural, and economic makeup of the population – may skew the data as well. Demographic information is readily available
and could be incorporated. Also, more narrowly focused studies could identify and assess other variables.
|
