The U.S. Department of Labor’s Bureau of Labor Statistics releases annual survey data about union membership. This data is based on two questions included in the Community Population Survey overseen by the U.S. Census Bureau.
The questions are: 1) “On this job, are you a member of a labor union or of an employee association similar to a union?” and 2) “On this job, are you covered by a union or employee association contract?” If survey responders answer “yes” to the first question, they are counted as union members. If they answer “no,” they are prompted with the second question. Answering this question in the affirmative classifies the respondent as being represented by a union. Respondents must answer “no” to both questions to be counted as a nonunion employee.
Extrapolating from CPS results, the U.S. Bureau of Labor Statistics estimates that there were 7,062,000 public sector union members nationwide in 2022 — that is, the total number of local, state and federal government employees who would answer “yes” to the first question. This is down from 7,216,000 in 2017, the last data available before the Janus decision. That’s a loss of 154,000 members, a 2.1% decline. The overall percentage of public sector workers saying they are union members dropped from 34.4% to 33.1% over this five-year period.
The number of public sector employees who said they are covered by a union contract but were not members of a union also increased during this time. These are employees who presumably work in a unionized job but are not members of a union. The BLS estimated there were 773,000 such public employees in 2022, up from 735,000 in 2017, a 5.2% increase.
The CPS covers all 50 states, but only union members in states without right-to-work laws were affected by the Janus decision. Public employees in right-to-work states were already able to opt out of joining the union at their workplace. Breaking out the 22 non-right-to-work states, the BLS estimated that there were 5,234,327 public sector union members in non-right-to-work states in 2017. In 2022, that number fell to 5,173,817. That’s a drop of 60,510 union members, a change of 1.2%. 
The number of employees in non-right-to-work states who were covered by a union contract but not members of a union, however, increased. In 2017, the BLS estimated there were 327,303 such public workers. In 2022, there were 366,946, an increase of 12.1%. 
But this federal survey data has severe limitations.
A major shortcoming is that some respondents misunderstand the questions. For instance, in North Carolina, South Carolina and Virginia, public sector collective bargaining has been or still is illegal. Despite this, about one in 10 public employees in those states claim to be covered by a collective bargaining agreement, according to the CPS. Further, in Georgia, Indiana, Tennessee and Texas, collective bargaining is not allowed for certain public employees. Yet, according to the CPS data, these states have similar public sector unionization rates with states that permit collective bargaining for all government employees.
Another issue relates to the survey sample sizes. As noted by other researchers, the CPS results are imprecise and subject to sudden swings, especially at the state level. The sample size upon which the data is based comes from only one-quarter of the 60,000 people surveyed, or 15,000 responses from across the country. These responses include private sector union members, so filtering the results to include only public employees reduces the sample size further. State-level results for public sector union members are based on even smaller sample sizes.
Despite the issues, this survey is probably the most-cited data by media outlets and reporters. This is likely because it includes official statistics from the federal government and is relatively easy to understand. But it is the least reliable way to measure trends in public sector union membership.