Human Resources and Functional Illiteracy Public Administration

Human Resources and Functional Illiteracy

Public Administration Human Resources Research Study Design

Public Sector Human Resource Policy and Functional Illiteracy

Public Sector Human Resource Policy and Functional Illiteracy

Anderson and Ricks (1993) examined the role of illiteracy in determining worker status within the public sector, in part because it had not been done before. They were interested in understanding how worker stratification was determined, after scholars had raised the possibility of a worker caste system within government agencies. This issue is still relevant today, given the increasingly diverse workforce both ethnically and linguistically. This report will analyze the research methods used by Anderson and Ricks (1993) and summarize their findings

Methods

The sample studied by Anderson and Ricks (1993) were human resource professionals currently engaged in managing employees in the public sector. A 41-item questionnaire was used to determine the 'functional literacy' that these managers were noticing on the job, with functional literacy being defined as having the basic English reading, writing, and comprehension skills necessary to function in a specific job position. The survey respondents were also queried for information about problems caused in the workplace due to illiteracy and whether remedial actions were being taken to correct deficiencies. The survey instruments were mailed to a random sample of 1215 local government offices, representing 922 cities, 263 counties, and 30 special districts.

Results

The response rate to the surveys was 44%, much higher than expected (Anderson and Ricks, 1993). A total of 398 cities, 122 counties, and 13 special districts returned the completed surveys. Based on the demographic information included in the surveys, close to 48% of all employees represented by the survey respondents were members of a collective bargaining organization. In addition, a comparison between surveys received early and late revealed no difference in terms of the agency characteristics. The percent union employees represented by survey respondents and the lack of a demographic difference between early and late respondents, were interpreted by Anderson and Ricks as indicating no obvious bias in the survey sample.

Of the 533 agencies responding to the survey, 24% knowingly hire basic-skills deficient (BSD) workers (Anderson and Ricks, 1993). Another 37% do not hire BSD workers, but are aware that such employees were hired before their deficiencies were discovered. Approximately 16% do not hire BSD workers, but will contract with companies that hire such workers.

To determine whether the prevalence of BSD workers is influenced by geographic location, Anderson and Ricks (1993) used chi-square to test for significant differences. Their findings revealed that the Southeast and northern Midwest regions of the U.S. were more likely to employ BSD workers (p < 0.01). Also examined were jurisdiction (p > 0.10), local unemployment conditions (p > 0.40), and agency size (p > 0.05) and no significant differences were found, although agency size may be trending towards significance. The authors also compared unionized, partially unionized, and non-unionized agencies and found no significant difference (p > 0.05), unless they grouped unionized and partially-unionized together for the comparison (p < 0.01).

Overall, agencies that hired BSD workers were more likely to be non-union (p < 0.01), from areas with high unemployment (p < 0.001), and located in the Southeast and northern Midwest (p < 0.001). By comparison, agency size and jurisdiction did not determine prevalence of BSD workers.

A one-way analysis of variance (ANOVA) was used to look for significant interactions between agencies intentionally hiring BSD workers for low-level positions and suspected dependent variables (Anderson and Ricks, 1993). This agency group experienced more problems with BSD workers (p < 0.001), in part because BSD workers were unable to comprehend rules and policies. However, these agencies encountered few problems when promoting these same employees (p < 0.001), despite BSD workers having higher rates of absenteeism, accidents, and job loss across all types of agencies. It should be noted that the authors mentioned that increased accident rates tend to be associated with low-skill jobs. These agencies were also more likely to adapt the job to the BSD employee's skill level through de-skilling the job (p < 0.01), rather than provide basic skills training (p < 0.05).

Discussion

The study by Anderson and Ricks (1993) represented a quantitative survey study. They used a 41-item survey instrument to collect the data, but also left room for individualized responses concerning problems human resource professionals experience with the hiring and managing BSD workers. The primary hypothesis tested was whether functionally illiterate workers were being hired by government agencies and they found that only 24% intentionally hire BSD workers. Anderson and Ricks (1993) then asked whether there existed factors which could predict the practice of knowingly hiring BSD workers. They found that geographic location, local unemployment rates, employment of union members, and a willingness to provide basic skill training were significant predictors of whether a government agency intentionally hired BSD workers or not.

The survey utilized a Likert scale and the data collected was therefore primarily ordinal in nature (Anderson and Ricks, 1993). The grouping of survey respondents by BSD worker hiring policies created categories. The use of the chi-square to test for significant interactions between categories and the ordinal data was therefore appropriate. The one-way ANOVA can be used for ordinal data by treating agency categories as independent variables. However, the data must meet certain assumptions before ANOVA can be used. Bias must be minimized, independent variables truly independent, the data randomly sampled, and the variance equivalent across all variables. With respect to the bias requirement, the authors examined whether unionization was over- or under-represented in the data and whether there were significant differences between early and late respondents and found no evidence of bias. In addition, the authors limited the ANOVA analysis to one independent variable. Given the lack of evidence for bias, the data can probably be assumed to be randomly sampled. No information was provided concerning the magnitude of variance, so this cannot be examined. The use of the chi-square and the one-way ANOVA therefore appears to have been appropriate given what is known about the data.