Job Automation and Concerns During the COVID-19 Pandemic Among Middle and Low-Income US Adults

Abstract

Objective 

This study examined job automation and concerns about job automation during the COVID-19 pandemic.

Methods 

Data from a national sample of 6607 middle- and low-income US adults in May to June 2020 were analyzed.

Results 

Among recently employed participants, 19.1% reported partial or complete job automation during the pandemic. Male sex, racial/ethnic minorities, students, higher education, full-time employment, and any history of psychiatric disorder were associated with job automation. In the total sample, 57.5% had concerns about job automation, which were associated with younger age, male, racial/ethnic minorities, students, veterans, more minors in the household, Northeast residence, any history of psychiatric diagnosis, COVID-19 infection, and any recent job changes.

Conclusions 

Job automation is impacting the lives of middle- and low-income US adults, and particular groups are vulnerable to ongoing changes in the nature of work.

In Brief

Male, racial/ethnic minority, middle and low-income U.S. adults disproportionately reported experiencing job automation during the COVID-19 pandemic and may particularly benefit from retraining and vocational assistance programs. Broad-scale interventions and policies are needed to alleviate concerns about future job automation.

Advanced technologies (eg, robotic technology) and computing infrastructure (eg, artificial intelligence) continue to transform the nature of work across the globe through job automation by means of computer-controlled equipment.¹ While job automation improves productivity, economic growth, and community development, there is a growing concern about how job automation may disrupt the supply and demand of labor.^2,3 Confirming this concern, 65% of Americans believe that robots and computers will take over many jobs currently performed by humans in the next 50 years.² These changes may disproportionately impact certain groups in society as 76% of Americans believe income inequality will increase with job automation.² Concerns about job automation have also been found to vary by socioeconomic standing and political ideology.⁴ One study of vocational rehabilitation programs found most job placements have high probabilities of being automated in the future, such as in housekeeping/janitorial service (66% probability), administrative and clerical tasks (96%), food service (87%−96%), and warehouse positions.³ Other studies have simulated future changes in work with the development of automation technologies and argue that the creation of new jobs, retraining, and government interventions can help plan for job automation.⁵ Regardless, there is expected to be consequences of job automation on both developed and developing countries.²

The coronavirus disease 2019 (COVID-19) pandemic has dramatically changed the landscape of work. The pandemic has not only been deleterious and led to increased morbidity and mortality, but it has also impacted the global economy and the nature of work. Because governments across the world implemented mitigation strategies to reduce virus contamination, millions of individuals either permanently or temporarily lost their jobs, and many employers turned to new strategies to shore up labor shortages and handle higher labor costs.^6,7 Research done by the European Centre for the Development of Vocational Training revealed that jobs that are considered low risk of automation provided a shield against job loss during the COVID-19 pandemic.⁸ However, it is unclear whether concerns about job automation have increased during the COVID-19 pandemic and how they may manifest in the future of work.

Although job automation has boomed recently, there has not been an adequate examination of this issue in the context of the COVID-19 pandemic. It is unclear how susceptible the general public is to job automation, what concerns there are about automation, and which groups in society are most susceptible to automation and automation concerns. Therefore, in the current study, we used a national demographically representative sample of middle- and low-income US adults to examine (1) the prevalence of job automation experienced by the population during the COVID-19 pandemic, (2) sociodemographic and clinical characteristics associated with job automation experiences, and (3) the prevalence and correlates of job automation concerns during the COVID-19 pandemic.

METHODS

Sample

Data were from a national sample of 6607 middle- and low-income US adults recruited May to June 2020 to examine health and social well-being during the COVID-19 pandemic. Eligibility for the study were individuals who were 22 years or older, lived in the United States, and reported annual personal income of $75,000 or less. Participants were recruited through Amazon Mechanical Turk (MTurk), which is an online labor market with more than 500,000 participants across 200 countries. To ensure data quality, only participants who had completed 50 or more human intelligence tasks and had a human intelligence task approval rating of 50% or higher were eligible to participate. Data derived from MTurk have been found to be of similar quality or higher than that from traditional subject pools, such as community samples, college students, and professional panels.⁹

Of 9760 individuals who initially agreed to participate in the study, 6762 (69.3%) met eligibility criteria and 155 individuals after validity checks (ie, they failed 3 items from the validity scales from the Minnesota Multiphasic Personality Inventory 2). The final study sample consisted of 6607 participants (67.7% of initial recruitment) from all 50 US states and the District of Columbia. All participants provided informed consent and were compensated for participation, and study procedures were approved by the institutional review board at the University of Texas Health Science Center at Houston.

Measures

Background information was collected by asking participants their age, sex, race/ethnicity, income level education, marital status, student status, employment status, geographic region, any active service in the military, and the number of minor children in their household.

COVID-19 status was assessed by asking participants whether they had been tested for COVID-19 and what the outcome was (ie, positive, negative, not tested). They were also asked whether anyone close to them (eg, friends, family) had tested positive for COVID-19.

Physical health status was assessed by asking participants whether they had ever been diagnosed with any of 22 different physical health conditions (eg, cancer, heart disease, arthritis) and the total number was summed.¹⁰

Psychiatric history was assessed by asking participants whether they had ever been diagnosed with any of 8 mental health or substance use disorders (schizophrenia/schizoaffective disorder, posttraumatic stress disorder, alcohol use disorder, bipolar disorder, anxiety disorder, major depressive disorder, drug use disorder, traumatic brain injury).

Social connectedness was assessed with the Medical Outcomes Study (MOS) Social Support Survey–Short Form.¹¹

Job changes experienced during the COVID-19 pandemic were assessed with a question that asked: “During COVID-19 social distancing measures, has your employer changed your job in any way?” Participants responded by checking any of the following that applied: “Automated your entire job (ie, used robots/machines/machines),” “Automated aspects of your job,” “Moved to virtual communication (eg, videoconference calls),” “Reduced your paid hours,” “Other (please specify),” and “N/A, no employer.” Any job change was coded as any of the previously mentioned response except for “N/A” or “No change.”

Concern about job automation was assessed with a question that asked: “How worried are you about job automation in your occupation (ie, robots, machines, computers replacing your job)?” Participants responded on a 5-point scale from 1 (Not at all) to 5 (Extremely).

Data Analysis

Poststratification weights were applied to all of the following analyses. First, descriptive analyses were conducted with the total sample to describe reported job changes during the COVID-19 pandemic. Second, among recently employed participants (ie, reported either being employed or had recently lost their job at study enrollment), we divided participants into those who reported experiencing any job automation during the COVID-19 pandemic and those who did not. These 2 groups were compared on sociodemographic and clinical characteristics with t tests and chi-squared tests. Third, we conducted a logistic regression analysis to identify participant sociodemographic and clinical characteristics associated with job automation, including only significant variables found in the bivariate analyses. Fourth, we conducted descriptive analyses with the total sample to describe concerns about job automation during the COVID-19 pandemic. Fifth, participants in the total sample were then divided into those who reported they were “moderately” or “extremely” worried about job automation (herein labeled the automation concerned [AC] group) and those reported they were “not at all,” “slightly,” or “somewhat moderately” worried about job automation (herein labeled the automation unconcerned [AU] group). Six, the AC and AU groups were compared with bivariate tests followed by a multiple regression analysis to identify sociodemographic and clinical characteristics associated with greater concerns about job automation.

RESULTS

Table 1 shows reported job changes in the total sample (N = 6607), which showed nearly one-third reported their job moved to virtual communication and nearly one-fifth experienced reduced hours or reduce salary during the COVID-19 pandemic. In addition, in the total sample, 1143 (weighted 14.3%) reported that their job were partially or entirely automated during the pandemic (excluded 161 participants who indicated both partial and entire automation). Overall, 4501 participants (weighted 58.0%) reported that their employer changed their job in at least one of these ways during the pandemic.

TABLE 1 - Reported Job Changes Due to COVID-19 Social Distancing Measure Among Middle and Low-Income Adults (N = 6,607)

During COVID-19 Social Distancing Measures, Has your Employer Changed Your Job in any Way? (Check all that Apply)?	n	Weighted % of Cases
N/A, no employer, self-employed, or not working	1,939	39.2%
Moved to virtual communication (eg, videoconference calls)	2,427	31.4%
Reduced hours/reduced salary	1,435	18.6%
Automated aspects of your job	860	10.7%
Automated your entire job (ie, used robots/machines/machines)	444	5.7%
No changes	222	3.6%
Implemented social distancing measures (eg, masks, physical barriers)	142	2.1%
Laid off/furloughed	145	2.0%
Business closed	45	0.8%
Other/more responsibility and tasks	27	0.4%
Raise/increase in work hours	21	0.3%
Total	7,707

Experiences With Job Automation During COVID-19 Pandemic

We then focused analyses on only participants who were recently employed (n = 5531). Among participants who were recently employed, 1122 (weighted 19.1%) reported that their jobs were partially or entirely automated during the COVID-19 pandemic.

As shown in Table 2, participants who reported experiencing job automation during the COVID-19 pandemic were significantly younger, with lower levels of income, higher levels of education, and were more likely to be male, racial/ethnic minority, attending school, married/living with partner, have more minor children, to live in the West and South regions of the country, and to be a veteran compared with participants who reported no job automation. Participants who reported job automation also reported significantly more physical health conditions, reported lower social support, and were more likely to report any lifetime history of psychiatric disorder, a positive test for COVID-19, and a positive COVID-19 test among a loved one.

TABLE 2 - Comparison of Characteristics between US Adults Who Had their Jobs Automated and Those Who Did Not

	Total (N = 5,531)		Automation (n = 1,122)		No Automation (n = 4,409)		ToD
	Raw n/Mean	Weighted %/SD	Raw n/Mean	Weighted %/SD	Raw n/Mean	Weighted %/SD	χ2 or t Test
Background characteristics
Age	44.01	15.13	39.76	12.60	45.02	15.50	13.379***
Male	2,630	46.50%	710	63.00%	1,920	42.60%	189.536***
Race							420.408***
White	4,147	76.00%	670	60.60%	3,477	79.70%
Black	805	14.20%	325	31.40%	480	10.20%
Asian	393	4.00%	72	3.80%	321	4.00%
Other	186	5.80%	55	4.20%	131	6.10%
Ethnicity, Hispanic	877	21.40%	351	43.70%	526	16.10%	512.814***
Income level	37,481.01	21,931.29	34,174.10	25,205.37	38,263.54	21,008.64	5.622***
Education							236.437***
High school or below	384	7.50%	40	4.10%	344	8.30%
Some college	1,013	21.00%	147	12.90%	866	22.90%
Associates/bachelor’s degree	2,876	49.20%	545	46.50%	2,331	49.80%
Graduate degree	1,258	22.30%	390	36.40%	868	19.00%
Part-time/full-time student	1,444	22.90%	567	50.00%	877	16.50%	721.246***
Marital status							319.192***
Single	1,767	26.10%	240	16.50%	1,527	28.30%
Divorced/Separated/Widowed	463	14.30%	35	3.60%	428	16.90%
Married/Living with partner	3,301	59.60%	847	79.90%	2,454	54.80%
No. minors in household	1.73	0.96	2.23	0.90	1.61	0.93	−22.854***
Currently employed full-time	3,741	63.30%	946	84.10%	2,795	58.40%	321.726***
Region							101.397***
Northeast	1,066	18.30%	207	17.60%	859	18.40%
Midwest	1,061	19.80%	164	14.10%	897	21.10%
South	2,117	38.70%	400	35.40%	1,717	39.50%
West	1,287	23.30%	351	32.90%	936	21.00%
Military veteran	731	16.40%	403	37.80%	328	11.30%	579.339***
Clinical characteristics
No. physical health conditions	1.78	2.44	2.79	4.01	1.55	1.76	−11.126***
Any lifetime history of psychiatric diagnosis	2,892	48.20%	912	83.40%	1,980	39.90%	855.775***
Tested positive for COVID-19	342	5.70%	220	18.90%	122	2.50%	565.017***
Had loved one test positive for COVID-19	1,443	24.30%	507	45.30%	936	19.30%	413.183***
MOS Social Support Survey score	21.29	5.94	22.01	4.34	21.12	6.25	−6.308***

MOS, Medical Outcomes Study; ToD, test of difference.

***P < .001.

These bivariate findings were followed with a logistic regression (Table 3), which showed being male, from a racial/ethnic minority group, having higher levels of education, being a student, working full-time, and having any history of psychiatric disorder were all independently significantly associated with job automation (all odd ratios >1.50).

TABLE 3 - Logistic Regression (Dependent: Job Automation or No Job Automation)

	B	SE	Wald	df	Sig	Odds Ratio
Age	−0.016	0.003	24.016	1	<0.001	0.985
Sex: Male (reference: not a male)	0.412	0.075	29.899	1	<0.001	1.510
Race (reference: White)
Black	0.660	0.093	49.876	1	<0.001	1.934
Asian	0.454	0.183	6.121	1	0.013	1.575
Other	0.356	0.166	4.581	1	0.032	1.427
Hispanic (reference: non-Hispanic)	0.528	0.083	40.997	1	<0.001	1.696
Income more than $35,000 (reference: $35,000 or less)	−0.111	0.079	1.978	1	0.160	0.895
Education (reference: high school or below)
Some college	0.193	0.186	1.081	1	0.299	1.213
Associates/bachelor’s degree	0.329	0.172	3.636	1	0.057	1.389
Graduate degree	0.622	0.180	11.873	1	0.001	1.862
Part-time/full-time students (reference: not a student)	0.749	0.080	86.669	1	<0.001	2.114
Marital status (reference: single)
Divorced, separated, or widowed	−0.577	0.186	9.598	1	0.002	0.561
Married or living with partner	0.359	0.104	12.045	1	0.001	1.432
No. minors in the household	0.241	0.039	38.268	1	<0.001	1.272
Full-time employee (reference: other than full-time)	0.794	0.095	70.618	1	<0.001	2.213
State of residence (reference: northeast)
Midwest	−0.303	0.125	5.909	1	0.015	0.739
South	−0.112	0.105	1.142	1	0.285	0.894
West	0.146	0.110	1.778	1	0.182	1.157
Veteran (reference: not a veteran)	0.358	0.100	12.921	1	<0.001	1.431
No. physical health conditions	0.108	0.016	45.914	1	<0.001	1.114
Any lifetime history of psychiatric diagnosis (reference: no lifetime history of psychiatric diagnosis)	1.131	0.088	164.636	1	<0.001	3.100
Have family/friend tested positive for COVID-19 (reference: no one tested positive)	0.378	0.083	20.566	1	<0.001	1.459
Tested positive for COVID-19	0.191	0.138	1.919	1	0.166	1.210
MOS Social Support Survey score	0.004	0.007	0.261	1	0.609	1.004
Total R 2	0.251

Note: Odds ratios >1.50 were bolded.

MOS, Medical Outcomes Study.

Concerns About Job Automation During COVID-19 Pandemic

In the total sample (N = 6607), participants rated their worries about job automation as a mean of 1.78 (SD, 1.08) on a 5-point scale or a weighted 57.5% reported being at least “slightly” worried about job automation. When participants were divided into 2 groups, 774 (weighted 9.6%) were grouped in the AC group (ie, reported being “moderately” or “extremely” worried about job automation) and 5824 (weighted 90.4%) were grouped in the AU group (ie, reported “not at all,” “slightly,” or “somewhat moderately” worried about job automation).

As shown in Table 4, the AC group was significantly younger, had higher levels of education but lower income, and was more likely to be male, from a racial/ethnic minority group, attending school, married/living with partner, have more minor children, employed full-time, living in the South region of the country, and is a veteran than the AU group. The AC group also reported significantly more physical health conditions, was more likely to have a history of psychiatric diagnoses, and was more likely to have tested positive for COVID-19 as well as have a loved one who tested positive for COVID-19 than the AU group. Regarding job changes, the AC group was significantly more likely to report experiencing any job automation, move to virtual communication, and reduced work hours/salary during the COVID-19 pandemic compared with the AU group.

TABLE 4 - Comparison of Characteristics between US Adults With Different Levels of Worry About Job Automation

	Total (N = 6,598)	AC (n = 774)			AU (n = 5,824)		ToD
	Raw n/Mean	Weighted %/SD	Raw n/Mean	Weighted %/SD	Raw n/Mean	Weighted %/SD	χ 2 or t Test
Background characteristics
Age	48.61	17.70	40.75	12.54	49.44	17.97	19.462***
Male	2,967	42.6%	439	56.0%	2,528	41.2%	78.514***
Race							324.742***
White	4,999	78.0%	447	60.2%	4,552	79.8%
Black	890	12.1%	226	29.6%	664	10.3%
Asian	484	3.7%	74	5.5%	410	3.5%
Other	225	6.2%	27	4.6%	198	6.4%
Ethnicity, Hispanic	947	17.2%	205	38.1%	742	14.9%	325.969***
Income level	34,733.60	21,869.11	32,253.19	24,082.87	34,997.12	21,605.34	3.387**
Education							80.131***
High school or below	550	8.5%	51	7.5%	499	8.6%
Some college	1,302	22.1%	115	14.5%	1,187	22.9%
Associates/bachelor’s degree	3,351	48.5%	377	47.1%	2,974	48.7%
Graduate degree	1,395	20.9%	231	30.9%	1,164	19.8%
Part-time/full-time student	1,595	18.3%	290	37.3%	1,305	16.3%	256.898***
Marital status							126.976***
Single	2,161	24.9%	204	21.8%	1,957	25.2%
Divorced/separated/widowed	623	20.7%	45	8.4%	578	22.0%
Married/living with partner	3,814	54.4%	525	69.8%	3,289	52.8%
No. minors in household	1.62	0.93	2.10	0.99	1.56	0.91	−15.962***
Currently employed full-time	3,736	46.2%	540	68.5%	3,196	43.9%	211.522***
Region							91.009***
Northeast	1,235	16.8%	162	18.7%	1,073	16.6%
Midwest	1,279	20.6%	124	15.4%	1,155	21.2%
South	2,547	38.9%	239	31.0%	2,308	39.8%
West	1,537	23.6%	249	34.9%	1,288	22.4%
Military veteran	814	15.2%	187	27.6%	627	13.9%	126.452***
Clinical characteristics
No. physical health conditions	2.04	2.39	2.59	3.68	1.98	2.20	−5.118***
Any lifetime history of psychiatric diagnosis	3,352	43.1%	545	72.1%	2.807	40.1%	361.523***
Tested positive for COVID-19	353	4.4%	119	15.7%	234	3.1%	330.468***
Had loved one test positive for COVID-19	1,609	20.9%	307	40.1%	1,302	18.9%	236.863***
MOS Social Support Survey score	21.21	6.26	21.14	5.82	21.21	6.31	0.348
Job changes during COVID-19 pandemic
Automated your entire job (ie, used robots/machines/machines)	442	5.6%	144	19.5%	298	4.2%	383.861***
Automated aspects of your job	860	10.8%	230	31.1%	630	8.6%	455.753***
Automated some or all aspects of your job	1,141	14.3%	313	41.4%	828	11.4%	637.649***
Moved to virtual communication (eg, videoconference calls)	2,425	31.4%	311	40.6%	2,114	30.5%	41.523***
Reduced your paid hours/reduced salary	1,431	18.6%	252	33.2%	1,179	17.0%	150.381***
Business closed	45	0.8%	2	0.3%	43	0.9%	3.279
Implemented social distancing measures (eg, masks)	142	2.1%	5	0.6%	137	2.2%	10.940***
Other/more responsibility and tasks	27	0.4%	0	0.0%	27	0.4%	3.731
Raise/increase in work hours	21	0.4%	0	0.0%	21	0.4%	3.730
Laid off/furloughed	144	2.0%	18	1.8%	126	2.0%	0.309
Any job change (ie, any of the above)	4,492	58.0%	663	84.7%	3,829	55.2%	309.805***

Any lifetime history of psychiatric diagnosis included schizophrenia/schizoaffective disorder, posttraumatic stress disorder, alcohol use disorder, bipolar disorder, anxiety disorder, major depressive disorder, drug use disorder, and traumatic brain injury.

*P < 0.05, **P < 0.01, ***P < 0.001.

AC, automation concerned; AU, automation unconcerned; MOS, Medical Outcomes Study; ToD, test of difference.

Table 5 shows results of a multiple regression analysis that included significant variables found in the previously mentioned bivariate analysis. Younger age, male sex, racial/ethnic minority status, student status, number of minors in the household, residence in the Northeast, veteran status, any lifetime history of psychiatric diagnoses, testing positive for COVID-19, and experiencing any job changes were each independently associated with greater concerns about job automation (all P < 0.001). Together, these variables explained approximately 27% of the variance in concerns about job automation during the COVID-19 pandemic.

TABLE 5 - Multiple Regression Analysis of Characteristics Associated With Job Automation

	B	SE	β	t	P
Age	−0.005	0.001	−0.076	−6.036	<0.001
Sex: male (reference: not a male)	0.187	0.02	0.085	9.218	<0.001
Race (reference: White)
Black	0.364	0.03	0.109	12.052	<0.001
Asian	0.328	0.051	0.057	6.448	<0.001
Other	−0.010	0.04	−0.002	−0.252	0.801
Hispanic (reference: non-Hispanic)	0.233	0.027	0.081	8.611	<0.001
Income	<0.000	0.000	−0.037	−3.768	<0.001
Education (reference: high school or below)
Some college	−0.066	0.038	−0.025	−1.74	0.082
Associates/bachelor’s degree	−0.055	0.035	−0.025	−1.562	0.118
Graduate degree	−0.062	0.039	−0.023	−1.576	0.115
Part-time/Full-time students (reference: not a student)	0.176	0.027	0.063	6.430	<0.001
Marital status (reference: single)
Divorced, separated, or widowed	0.095	0.034	0.036	2.838	0.005
Married or living with partner	0.057	0.025	0.026	2.237	0.025
No. minors in the Household	0.078	0.011	0.067	6.858	<0.001
Full-time employee (reference: other than full-time)	0.034	0.024	0.016	1.405	0.160
State of residence (reference: northeast)
Midwest	−0.18	0.031	−0.067	−5.819	<0.001
South	−0.176	0.028	−0.079	−6.375	<0.001
West	0.014	0.030	0.006	0.482	0.630
Veteran (reference: not a veteran)	0.197	0.030	0.065	6.581	<0.001
No. physical health conditions	0.013	0.004	0.028	2.890	0.004
Any lifetime history of psychiatric diagnosis	0.259	0.021	0.118	12.096	<0.001
Have family/friend tested positive for COVID-19 (reference: no one tested positive)	0.044	0.025	0.017	1.758	0.079
Tested positive for COVID-19	0.365	0.052	0.069	7.011	<0.001
Job change due to COVID-19 social distancing measures (reference: no change)	0.197	0.030	0.065	6.581	<0.001
Total R 2	0.272

DISCUSSION

This national study examined experiences and concerns with job automation among middle- and low-income US adults during the COVID-19 pandemic. More than half of the participants reported that their employer changed their job in some way during the pandemic, which is not surprising given the unique COVID-19 social distancing measures implemented across industries impacting employment and wages.^12–14 Our results revealed that 19.1% of participants reported that their jobs were entirely or partially automated during the COVID-19 pandemic consistent with increased demand for job automation in various sectors during the pandemic.^15,16 The Pew Research Center recently surveyed Americans regarding job automation and found that approximately 48% of participants reported that job automation in the workplace has mostly hurt American workers.¹⁷ Some experts have described the COVID-19 pandemic as contributing to the acceleration toward a “fourth industrial revolution.”¹⁸

Thus, it seems justified that we found more than half of participants reported at least some concerns about job automation, although only approximately one-tenth were characterized as moderately/extremely concerned about job automation. We did find that individuals with certain backgrounds were at greater risk of reporting job automation than others. Specifically, males, racial/ethnic minorities, students, and those with any history of psychiatric disorders reported experiencing job automation at higher rates during the COVID-19 pandemic. This may be due to the jobs that individuals with these characteristics are more likely to hold, an adverse impact on individuals with these characteristics, or some other vulnerabilities associated with these characteristics that put them at risk of job automation. Because there is already great concern about racial/ethnic health disparities, it may be important to examine this further because disparities in job automation may exacerbate health disparities. Similarly, individuals with psychiatric problems may need additional employment support given the challenges they already face obtaining and retaining employment. Some vocational rehabilitation programs also place individuals in jobs that have high probabilities of being automated.³

We found characteristics associated with job automation were also associated with concerns about job automation, including individuals from racial/ethnic minority backgrounds and those with psychiatric histories reporting greater job automation concerns. Students, veterans, those who tested positive for COVID-19, and those who experienced job changes during the COVID-19 pandemic were also more likely to report job automation concerns. Many of these individuals were directly or indirectly impacted by changes in the nature of work during the COVID-19 pandemic, which may have shaped their concerns about job automation. Vulnerable populations such as those who frequently interacted with others (eg, students) or those with health conditions and disabilities were advised to take extra precautions during the COVID-19 pandemic, which may have caused real concerns about job automation.^18,19 The findings are also consistent with studies that have found many individuals with disabilities are concerned about losing their job due to the COVID-19.^20,21 Taken together, these findings suggest that job automation has impacted middle- and low-income US adults during the COVID-19 pandemic and thus concerns about job automation may be warranted, especially among those from certain backgrounds who may be particularly vulnerable. Initiatives to plan for job automation should be informed by these vulnerabilities.

There were several study limitations worth mentioning. First, the data were cross-sectional and thus no inferences can be made about the causality between variables. We mostly focused on job automation as a potential threat, but others have pointed to potential benefits, which may be important to not dismiss.^22,23 Second, information about job changes was based on self-report and not validated by employers or companies. Third, we did not have data on specific employment industries or job types, which likely influence experiences and concerns with job automation. These limitations notwithstanding, the strengths of the study include the use of a national sample, focus on middle- and low-income adults, and unique examination of a range of sociodemographic and clinical characteristics associated with job automation concerns. The findings capture the extent to which different segments of the US population has been affected by job automation during the COVID-19 pandemic and are concerned about job automation, which will be important in planning recovery efforts after pandemic.