C

Identifying Salary Ranges for Jobs Relevant to the Data Life Cycle

The main text identifies certain job descriptions with associated salary ranges, from L (low) to VH (very high). This appendix identifies possible job titles and associated salary ranges observed in workplace and occupational surveys conducted by the Bureau of Labor Statistics (BLS, 2019a).

DATA USED

Occupational Employment Statistics

Collection methods, estimation methodology, and coverage are described in BLS (2019b). The committee downloaded the data from https://www.bls.gov/oes/special.requests/oesm18nat.zip on October 30, 2019. From the downloaded data, national_M2018_dl.xlsx was used.

Occupational Information Network

The Occupational Information Network (O*NET) database is a U.S. Department of Labor–sponsored database developed by the National Center for O*Net Development.¹ The database provides standardized descriptions of hundreds of occupations within the U.S. economy. The database comprises worker attributes and job characteristics. Information is collected using a two-stage design in which the following occurs:

• A statistically random sample of businesses expected to employ workers in the targeted occupations is identified.
• A random sample of workers in those occupations within those businesses is selected. Data are collected by surveying job incumbents using a randomly assigned standardized questionnaire on occupation characteristics, out of three questionnaires. Additional questions cover tasks and demographic information.
• Abilities and skills information is developed by occupational analysts using the updated information from incumbent workers (National Center for O*NET Development, 2019a).

___________________

¹ See https://www.onetonline.org/, accessed August 12, 2020.

A data dictionary (National Center for O*NET Development, 2019b) provides additional information.

Version 23_2 of the data (National Center for O*NET Development, 2019c)² was used for committee determination of salaries. Both Occupation Data.xlsx and Alternate Titles.xlsx were used.

METHODS

Mapping Job Titles to Standard Occupational Classifications

O*NET is structured around Standard Occupational Classification (SOC; BLS, 2019c). The committee’s main text has a normative list of job descriptions based on data management practiced at university libraries. These may not match reported standard occupation titles. The O*NET data provide a long but not exhaustive list of alternate mentions of job titles for specific occupations (Alternate Titles.xlsx). Using both the standard occupation title as well as the alternative mention, the normative job title is matched via probabilistic matching, using the Jaro-Winkler distance (Winkler, 1990) as implemented in the R package fuzzyjoin (Robinson, 2019). All reasonable matches (d < 0.05) were kept to obtain a list of similar occupations and their SOC codes.

Mapping SOC into Salary Ranges

Occupational Employment Statistics computes for each SOC code a salary range, comprising annual salary and hourly wages, and characterized by the 25th and 75th percentile, as well as the median. The annual salary distributions were attached to each of the identified occupations (Table C.1), and then these statistics were collapsed to a triplet of information for each normative job description (Table C.2). To do so, the minimum of all observed 25th percentiles, the median of all observed medians, and the maximum of all observed 75th percentiles were chosen. No weights were applied. An alternative implementation might use the employment shares to create weighted statistics. Reliability statistics were not computed, as the resulting table is meant to be indicative, not precise.

RESULTS

Table C.1 lists the annual salaries, as of 2018, by job title (median, and the 25th and 75th percentile), for all occupations identified as having similar names as the normative description in Chapter 2. Blank salaries (“NA”) indicate that no occupation code could be found on O*NET based on the normative description. Table C.2 lists the ranges, as defined above, for each of the normative descriptions (Chapter 2), based on the underlying occupations identified. Table C.3 lists the statistics associated with each of the salary categories, from low to very high. While the categories are defined based on the experience of members of the committee, ex ante, they match up well with observed median salaries in 2018.

FULL CODE AND DATA

The code and data underlying this appendix, including an exhaustive list of the committee’s edits (inclusions and exclusions) to the list of occupations, are available at https://github.com/labordynamicsinstitute/job-description-and-wages.

TABLE C.1 Annual Salaries (2018) by Job Title for Occupations

NOTE: IT, information technology; SOC, Standard Occupational Classification.

TABLE C.3 Statistics Across Each of the Salary Categories (used in Chapter 2 of this report)

REFERENCES