OregonSaves Auto-IRA Program Works for Employers

Overview

In recent decades, employer-provided retirement plans have served as the main vehicle for retirement savings in the United States. Today, less than 15% of U.S. households save for retirement outside of the workplace.¹ However, about a third of Americans lack access to a retirement plan at their job, in part because many employers find starting a plan to be too expensive and burdensome.²

In response, half of U.S. states are exploring the use of automatic enrollment into individual retirement accounts (IRAs) for private sector workers who don’t have access to workplace retirement savings plans. These IRAs are known as auto-IRAs or Secure Choice programs, and seven states—California, Colorado, Connecticut, Illinois, Maryland, New Jersey, and Oregon—are currently implementing them.³ Workers without access to an employer-provided retirement plan are automatically enrolled and contribute a preset percentage of their wages or salaries (employees can opt out of the program or change the contribution percentage).

In a first-of-its-kind look into auto-IRA programs and to help inform policymakers considering the establishment of such a program, The Pew Charitable Trusts surveyed over 2,500 private sector businesses from 2019 to 2020 that participate in OregonSaves—the first auto-IRA in the U.S., launched in 2017. The survey was designed to help better understand how employers—which are required to facilitate the program, but otherwise have minimal involvement with it—experience various elements of OregonSaves for at least a few months after registering with the program, and whether the program imposes costs or burdens. The survey also enabled Pew to examine participants’ satisfaction with the program across various employer characteristics such as business size, which may help to identify opportunities for improvement in the employer experience.

In general, responses show strong levels of satisfaction with OregonSaves. (See the methodology for more information about the survey.)

• Overall, 73% of participating employers are satisfied or neutral (neither satisfied nor dissatisfied) with their experience in the program.
• Expressed satisfaction is 21/2 times higher among employers who have begun to process payroll contributions into funded accounts for their employees compared with those that have not yet started, suggesting that greater familiarity with the program may lead to higher levels of satisfaction.
• The stay-at-home order issued in March 2020 in Oregon does not seem to have had an effect on employer satisfaction levels with the program. This survey was carried out from 2019 to 2020 in three rounds, with each round including newly registered businesses. The last round, conducted in March and April 2020, coincided with statewide restrictions due to the COVID-19 pandemic.
• Businesses that register with the program well in advance of registration deadlines are more likely to express satisfaction with OregonSaves than those that register closer to their deadline. Because employers are not required to sign up for the program until their registration deadline, and can even register after the deadline, getting started “early” and expressing higher satisfaction suggests that a strong demand exists for auto-IRA programs such as OregonSaves among a number of businesses.
• Employers that have spent more time participating in OregonSaves are more likely to indicate higher satisfaction, which indicates that employers’ increased familiarity with OregonSaves may lead to greater satisfaction with the program.
• Dissatisfaction was highest with setting up payroll lists during registration and with administering payroll contributions, which are more time-intensive tasks. Still, about 70% of employers were satisfied or neutral regarding payroll-related tasks.
• Business characteristics—such as size—are associated with satisfaction levels. For example, smaller companies expressed higher satisfaction with OregonSaves than larger ones; smaller firms, often with fewer financial resources, are less likely than bigger companies to offer their own retirement plan, and so may be more receptive to an alternative such as OregonSaves. Employers with more part-time workers, or those in certain industries such as leisure and hospitality, showed higher levels of dissatisfaction, which suggests that payroll tasks associated with OregonSaves may be more challenging for high-turnover workforces.
• An employer’s financial situation may affect its satisfaction with OregonSaves. For example, firms that said they had increasing revenues over the past year expressed higher satisfaction with OregonSaves. In addition, employers that reported out-of-pocket costs because of OregonSaves are more dissatisfied than firms that did not report out-of-pocket costs.

Overall experience with OregonSaves

Figure 1 shows that more than 73% of businesses rated their experience with the program as very satisfied, somewhat satisfied, or neutral (expressed as neither satisfied nor dissatisfied), with almost 45% of respondents either very or somewhat satisfied.⁴ Dissatisfaction was at around 27%.

Businesses registered with OregonSaves were at varying stages of the program at the time of the survey. For instance, although all businesses in the survey had registered, 39% indicated that they had not yet administered payroll contributions on behalf of their employees into funded retirement accounts—most commonly because the business does not have any participating employees (79% of those that said they had not started). If a business does not have any participating employees, most commonly all employees have either opted out or have set their contribution percentage to 0%.

Table 1

Reason Business Has Not Started Processing Payroll Contributions

Note: N=953. Percentages may not total 100 due to rounding.

Source: Pew’s Survey of Employers Registered With the OregonSaves Retirement Savings Program, 2019-20

© 2021 The Pew Charitable Trusts

Figure 2 shows differences in satisfaction levels with OregonSaves between businesses that have started processing payroll contributions for their employees and those that have not. Among those that have started processing contributions, 78% said that they were either satisfied or neutral with their experience. Meanwhile, 65% of employers that had not yet processed payroll contributions reported a satisfied or neutral experience. The proportion that expressed outright satisfaction was significantly higher for those employers that had started processing contributions (58%) than with those that had not (22%). Although it is up to employees to decide whether to remain in the program or opt out, businesses that have registered for OregonSaves but have not actually started processing payroll contributions to employees’ accounts are less likely to witness the benefits of the program.

The majority of businesses (61%) report that they’ve started to process contributions to OregonSaves. Statistical models (see the Appendix for details on model design and full results) show that starting to process payroll has a statistically significant association with higher satisfaction; on average, a business that is processing payroll contributions (and therefore seeing employee participation and benefits) is over five times as likely to be satisfied (as opposed to either neutral or dissatisfied) with their experience. As more businesses start processing payroll contributions, one might expect satisfaction levels will continue to increase.

Although the third round of the survey data collection coincided with the early emergence of the COVID-19 pandemic and Oregon’s corresponding stay-at-home order⁵, there is little evidence in the data of a significant impact on satisfaction with OregonSaves: The satisfaction levels between rounds two (pre-pandemic) and three of the survey were not statistically different. (See Figure 3.) Although the overall proportion of those satisfied or neutral in survey round one was not statistically different from rounds two or three, it is worth noting that there was a higher proportion of employers that were satisfied in the first round of data collection.

One explanation for this difference may be the length of time that employers had been enrolled in the program:

On average, businesses surveyed in the first round of data collection had been registered with the program for a longer amount of time. The mean length of time in the program for round one of the survey was 314 days compared with 150 days and 181 days for rounds two and three, respectively. And the statistical estimates imply that being enrolled in OregonSaves for a longer amount of time was significantly associated with higher levels of being satisfied: On average, each 30 days of enrollment was associated with a 5% increase in the likelihood of being satisfied rather than neutral.

In addition, surveyed employers that registered for the program well in advance of their deadline expressed higher levels of satisfaction with their experience. (See Figure 4.) Satisfaction among businesses that registered more than 90 days in advance of their deadline was almost 16 percentage points higher than satisfaction among those that didn’t register more than 30 days in advance of their deadline (or even after the deadline had passed). The modeling results indicate that this association is statistically significant.⁶ A high proportion of employers in round one of the survey registered more than 90 days early (45%). Comparatively, the proportion of employers that registered that early was only 13% and 22% among round two and three respondents, respectively.

The high satisfaction levels expressed by early registrants suggests a demand for OregonSaves, which finds support in anecdotal comments by employers. In the survey, responses to an open-ended question asking about impressions about the program show that OregonSaves addresses critical needs in the workplace. Among the answers were:

• “As a very small business it has been so appreciated as other options seemed out of reach for us.”
• “It is great having a free option for savings for our employees. We eventually want to offer our own program, but this is nice for the time being.”
• “I do appreciate the program overall. It helps younger staff start saving early. From a small business that can’t afford to have a retirement plan it is a nice option for our team.”

In addition to asking employers about their satisfaction with the program, the survey asked them whether they supported or opposed the program when thinking about its overall intent, objectives, and design. There was significant overlap between program support and satisfaction with the program.⁷ Does program satisfaction lead to program support? Or do those who already support the program express higher levels of satisfaction? Because the questions are asked contemporaneously,⁸ future research is needed.

Satisfaction with the registration process and facilitation of OregonSaves

In addition to rating their overall experience with OregonSaves, employers were asked to rate individual elements of the program, including those that are part of the registration process and those that are part of the ongoing payroll functions of the program. Satisfaction levels were high across all elements of the registration process (see Figure 5), with employers expressing the strongest level of satisfaction with customer service: 82% said that they were satisfied or neutral with the help they received from OregonSaves during the registration process.

The highest level of dissatisfaction with a registration element was with the time it took employers to set up the payroll lists—which makes sense because payroll setup is the most time-intensive part of registration. Still, almost 70% of employers said they were either satisfied or neutral with respect to setting up payroll in the OregonSaves system.

Employer satisfaction was similarly strong with elements of OregonSaves associated with facilitating the program (see Figure 6): At least 80% of employers were either satisfied with or neutral about the program’s online employer portal, help received, and the email communications. And, as with registration, the most time-intensive element—administering payroll contributions—had the highest dissatisfaction. Even so, dissatisfaction remained low at around 30%.

Business characteristics associated with OregonSaves satisfaction

This brief also examines which business characteristics—such as the size of a business or its industry—are associated with differences in satisfaction with OregonSaves. Although the overall satisfaction with OregonSaves is high, understanding the variation among registered employers can highlight any potential burdens as well as opportunities for further improvement in the employer experience. In addition to examining differences in response levels in the survey, Pew analyzed whether the observed associations are statistically significant, holding the other factors constant (see the Appendix for more details on the statistical models employed in the analysis).

The survey asked about various characteristics, including:

• Age of the business.
• Number of full-time, part-time, contractor, and seasonal employees.
• Whether the business outsources payroll or handles payroll internally.
• Changes in earnings over the past year.
• Whether the business experienced out-of-pocket costs associated with OregonSaves.

Satisfaction with OregonSaves does not appear to be associated with the length of time that a business has been in operation: Organizations expressed similar levels of satisfaction regardless of the age of the firm. But some variation in employers’ satisfaction with the program was evident when looking at each of the other business characteristics.

Organizations of varying sizes expressed differences in their satisfaction (see Figure 7): Satisfaction levels were high across all employers, but the proportion of businesses that were either satisfied or neutral was higher as the size of the organization decreased, with 69% of businesses with 20 or more employees either satisfied or neutral compared with 77% of the smallest employers (those with four or fewer workers). This relatively small 8% difference could be related to businesses with more employees having more information to upload when setting up payroll and processing payroll contributions. Additionally, larger firms likely have more resources to offer a retirement plan compared with small firms. As such, larger employers that choose not to offer their own plan may be predisposed to not offering a plan in the future and thus may be expressing their dissatisfaction with having to facilitate the OregonSaves program.⁹

A higher proportion—57%—of the smallest employers (those with four or fewer workers) reported being satisfied than was true for the other groups. Because registration remains open for these employers, all businesses of this size included in the survey registered with OregonSaves well in advance of the deadline. So to check whether early registrants—who, as noted above, are more likely to be satisfied with the program—are responsible for higher satisfaction shown among smaller employers, Pew narrowed the analysis in a statistical model to include only those businesses that joined OregonSaves close to (or after) their registration deadline. Even then, statistical results implied that smaller employers were more likely than larger firms to be satisfied rather than dissatisfied.

The survey also measured employment another way, with employers asked about the relative distribution of their worker types (full-time employees, part-time employees, contractors, and seasonal employees). The statistical model (see Table A2 in the appendix for full results) allows for an examination of association between having more of each type of worker and satisfaction with OregonSaves, holding all else equal. This shows that the relationship between the number of employees and satisfaction with the program is dependent on the type of worker.

On average, a business with fewer full-time employees (holding the number of all other types of employees constant) is more likely to be satisfied than dissatisfied: All else equal, a business with zero to four full-time employees is almost 11 percentage points more likely to be satisfied with its experience than a business with 20 or more full-time employees—a finding consistent with smaller firms expressing higher satisfaction than larger firms.

Having comparably fewer part-time workers (again, holding the number of all other employees constant) is also associated with a higher likelihood of satisfaction compared with dissatisfaction: A business with zero to four part-time employees, is about 9.5 percentage points more likely to be satisfied when compared with a business that has five to nine part-time employees and 11 percentage points more likely to be satisfied than a business with 20 or more part-time employees.

A decrease in likelihood of satisfaction also shows up when a business has comparably higher numbers of seasonal workers. With more turnover among seasonal workers than among full-time workers by definition, employers with more seasonal workers would spend more time setting up payroll lists with OregonSaves.

On the other hand, the statistical model indicates that having more contractors is associated with higher likelihood of employer satisfaction than dissatisfaction—logical because employers are not responsible for payroll deductions for their contracted workers.

Satisfaction with OregonSaves also varied across nine categories of industries (see Figure 8).¹⁰ Employers in the Education and Health Services category expressed the highest level of satisfaction with the program, with almost 80% either satisfied or neutral. The lowest levels of satisfaction came from two industries: About 67% of those within the largest industry in the state—Leisure and Hospitality, which accounted for 23% of respondents to the survey and as an industry employs many part-time and seasonal workers—were either satisfied or neutral. And little over 64% of employers in the Natural Resources and Mining industries were satisfied or neutral.

Whether a business outsources its payroll to a processing company or instead handles payroll internally is also associated with OregonSaves program satisfaction. (See Figure 8.) Some 47% of businesses that outsourced payroll were satisfied compared with 41.5% that handled payroll internally.¹¹ The statistical model also indicates that this difference is significant: Outsourcing payroll is associated with an increase in the likelihood of being satisfied rather than dissatisfied even when accounting for other factors. Since setting up payroll and processing payroll contributions were the registration and facilitation elements with the highest levels of dissatisfaction, it makes sense that a business that outsources this function would be more likely to be satisfied.

Figure 10 shows that about 75% of businesses that reported an increase in earnings over the past year were either satisfied or neutral with OregonSaves. The number was similar for businesses for which earnings had stayed about the same, although the share of neutral rather than satisfied was higher. About 66% of businesses that saw a decrease in earnings were either satisfied or neutral. The statistical model also indicates a significant relationship between changes in earnings and satisfaction levels. Prior research from Pew found that cost is a big barrier to adding benefits such as retirement savings;¹² increasing earnings may remove or reduce that barrier.

Finally, whether a business self-reported having out-of-pocket (OOP) costs in its experience with OregonSaves was an important factor associated with the overall level of satisfaction (see Figure 11). While there are no direct costs to the employer associated with program participation, 22% of businesses reported having OOP costs, most commonly office supplies or the time it takes to register and facilitate the program. More than twice the proportion of businesses that reported OOP costs were dissatisfied (49%) than businesses that reported no OOP costs (20%); the statistical model indicates that this association is significant, holding all other business characteristics constant. More research is needed to understand why some businesses self-report OOP costs and others do not.

Conclusion

State-facilitated auto-IRA programs such as OregonSaves offer an innovative way for states to address workers’ lack of access to employer-sponsored retirement plans. Pew’s survey of nearly 2,400 employers participating in the OregonSaves program is the first such analysis of auto-IRA programs; the hope is that the findings might help inform the design and implementation of similar no-cost retirement programs in the more than two dozen states considering or implementing them. Overall, employers are generally satisfied with OregonSaves, and future work by Pew will look specifically at any costs or burdens generated by this and other state-facilitated auto-IRA programs.

The survey reveals a business demand in Oregon for a no-cost retirement savings program for workers. Businesses in the state did not need to register prior to their assigned deadline, yet almost 27% of employers registered at least 90 days in advance of the deadline. And after beginning participation in the program, the group most in need of providing retirement benefits—smaller firms—was more likely to express satisfaction with the program compared with larger firms.

The survey results also show that support for OregonSaves tends to increase over time. Employers who have been participating in the program for longer and those that have begun processing payroll show more satisfaction with the program than those who have registered for the program but have not yet begun processing payroll for the program. Although it’s difficult to know whether these metrics mean the program is improving over time or whether employers are getting more comfortable as the program matures, OregonSaves received high marks for customer service in helping employers get going and resolving issues going forward.

In terms of policy, the fact that OregonSaves enjoys support from many employers may alleviate concerns that such programs overly burden small employers. But helping employers with payroll contribution processing is a critical issue and would remove a source of dissatisfaction for a significant minority of employers while boosting program assets and financial sustainability. Improved payroll processing might also lead to increased satisfaction by reducing perceived out-of-pocket costs because issues such as time spent on payroll administration were commonly cited in the survey as examples of costs associated with the program. Firms with high employee turnover or a high proportion of part-time or seasonal workers may need additional or targeted help in implementing an auto-IRA program.

Methodology

Qualtrics collected this data in Pew’s Survey of Employers Registered With the OregonSaves Retirement Savings Program. The target population was business representatives who were familiar with their organization’s experience with OregonSaves. Pew obtained a list of businesses maintained by the OregonSaves program administrator, Ascensus. Study participants were contacted using the email address on file with the program administrator. The survey was fielded online in three rounds during 2019 and 2020. OregonSaves registered participants over a series of deadlines, and each survey round was designed to capture newly registered businesses. The first round, conducted July 29 to Aug. 16, 2019, included all businesses that had registered on or before April 1, 2019. The second round, conducted Sept. 21 to Oct. 11, 2019, included all businesses that registered on or before May 31, 2019. The third and final round, conducted March 23 to April 10, 2020, included businesses that registered on or before Dec. 1, 2019.

Because all unique contacts received an invitation, this survey was a census and has a response rate of 21.8%. All analyses presented in this report are unweighted because the survey responses are representative across nine industry categories and employee size of businesses that had registered with OregonSaves on or before Dec. 1, 2019.

Model design

Although the analysis in this issue brief concentrates on bivariate cross-tabulations to explore the relationship between employer characteristics and satisfaction levels with OregonSaves, multivariate regression analysis is also used to supplement the analysis. Multivariate regression allows for assessing the relationship between each characteristic and expressed satisfaction while holding other characteristics constant. The analysis indicates in the text whenever the relationship between the employer characteristic and satisfaction is statistically significant in the regression model, holding all else equal.

The variable being explained in the regression analysis (dependent variable) is satisfaction with the overall experience with OregonSaves (“Please rate your business’s overall experience with OregonSaves”). Responses included very satisfied, somewhat satisfied, neither satisfied nor dissatisfied, somewhat dissatisfied, very dissatisfied, or N/A. “N/A” responses were excluded from the analysis. Since we are concerned with differences in overall employer satisfaction or dissatisfaction (or neither), we combined very satisfied and somewhat satisfied into one measure of satisfied (and we did the same with very dissatisfied and somewhat dissatisfied). In the analysis, we referred to neither satisfied nor dissatisfied as neutral.

Although these categories can be viewed as ordered, we used multinomial logistic regression to examine satisfaction rather than ordinal logistic regression. First, analysis showed that the proportional odds assumption underlying ordinal logistic regression is violated, indicating that the difference between each level is not the same. Second, there are two possible interpretations of neither satisfied nor dissatisfied. One might view this category as on a scale in between satisfaction and dissatisfaction (an ordered interpretation). However, one might also view this category as expressing an indifference between the two. Because there might be multiple theoretical interpretations of this category, multinomial regression is the preferred approach.

Each column in the multinomial regression results presents relative risk ratios. In the left-hand column, the results show the increase or decrease in likelihood of expressing satisfaction as opposed to dissatisfaction with OregonSaves. For example, the relative risk ratio reported for those who outsource payroll (with handling payroll internally as the reference category) is 1.4. This result indicates that those who outsource payroll are 1.4 times as likely to be satisfied with OregonSaves as opposed to dissatisfied compared with those who handle payroll internally. In other words, outsourcing payroll is associated with a 40% increase in the likelihood of being satisfied as opposed to dissatisfied. The middle column results show the increase or decrease in the likelihood of expressing satisfaction as opposed to being neutral, while the right-hand column results show the increase or decrease in likelihood of being neutral as opposed to dissatisfied.

The models all use various business characteristics as independent variables, including years in operation; whether payroll was handled internally or outsourced; whether earnings in the past year increased, decreased, or stayed the same; whether the business has started processing OregonSaves payroll contributions; whether the business reported experiencing out-of-pocket costs related to OregonSaves; the employer’s industry category; a measure of early registration; and how long the business had been enrolled in the program at the time of taking the survey. The model in Table A1 includes a measure of overall employer size from program information. The model in Table A2 is similar to Table A1, except it is limited to only those employers who registered within 30 days of their registration deadline or after the deadline and excludes the early registration measure. The model in Table A3 includes measures of how many full-time, part-time, contractor, and seasonal employees the employer reported having instead of using the overall size of the employer. Except for the discussion of the distribution of full-time, part-time, contractor, and seasonal employees, the analysis relies mainly on the results from Table A1.

Figures presented in the report are cross-tabulations from the survey responses. Any predicted probabilities referenced in the text used Stata’s margin commands, with covariates held at their mean.

Appendix

Table A.1

Note: 2,564 businesses responded to the survey. The model excludes 171 respondents (6.7%) that are missing the dependent variable, and, of those that are not, 338 respondents (13.2%) that are missing one or more covariates. In total 509 respondents (20%) of the sample are excluded. Statistical significance is indicated by * p < 0.05 ** p < 0.01 *** p < 0.001.

Source: Pew’s Survey of Employers Registered With the OregonSaves Retirement Savings Program, 2019-20

Table A.2

Note: 2,564 businesses responded to the survey. The model excludes 926 respondents that are employers that registered more than 30 days early, including all employers with four or fewer employees, leaving a subsample of 1,638. The model excludes 122 respondents (7.4%) that are missing the dependent variable, and, of those that are not, 213 respondents (13%) that are missing one or more covariates. In total 335 respondents (20.5%) of the subsample are excluded. Statistical significance is indicated by * p < 0.05 ** p < 0.01 *** p < 0.001.

Source: Pew’s Survey of Employers Registered With the OregonSaves Retirement Savings Program, 2019-20

Table A.3

Note: 2,564 businesses responded to the survey. The model excludes 171 respondents (6.7%) that are missing the dependent variable, and, of those that are not, 426 respondents (16.6%) that are missing one or more covariates. In total, 597 respondents (23%) of the sample are excluded. Statistical significance is indicated by * p < 0.05 ** p < 0.01 *** p < 0.001.

Source: Pew’s Survey of Employers Registered With the OregonSaves Retirement Savings Program, 2019-20

Endnotes

1. Investment Company Institute, “The Role of IRAs in U.S. Households’ Saving for Retirement, 2019” (2019), https://www.ici.org/pdf/per25-10.pdf.
2. The Pew Charitable Trusts, “Employer Barriers to and Motivations for Offering Retirement Benefits” (2017), https://www.pewtrusts.org/-/media/assets/2017/09/employer_barriers_to_and_motivations.pdf.
3. For a review of legislative activity at the state level, see Center for Retirement Initiatives, “State Initiatives 2020: New Programs Begin Implementation While Others Consider Action,” accessed Oct. 28, 2020, https://cri.georgetown.edu/states/.
4. In this analysis, we combine very satisfied and somewhat satisfied into one measure of satisfied. We also combine very dissatisfied and somewhat dissatisfied into one measure of dissatisfied.
5. Oregon Governor Kate Brown declared a state of emergency on March 8 and the state’s initial stay-at-home order on March 23, 2020. See J. Cowley, “Oregon State of Emergency vs. Stay-Home Order: What’s the Difference?,” KGW, May 4, 2020, https://www.kgw.com/article/news/health/coronavirus/oregon-stateof-emergency-versus-stay-home-order-whats-the-difference/283-95a91ace-9580-40ee-a8b6-727c6bec3216.
6. Because businesses that who registered early have also spent more time enrolled in the program, we separately estimated overall satisfaction for only those businesses that registered no more than 30 days ahead of the required deadline (including those that registered after the deadline). By excluding those that registered early, we can more clearly examine whether time in the program increases satisfaction for those that were more likely to have been motivated by the registration requirement. Among this group of employers, a smaller subsample, the statistical results show that time in the program is marginally significant (p=0.057) in predicting likelihood of being satisfied as opposed to dissatisfied.
7. When asked to think about the overall intent, objectives, and design of OregonSaves, 47.6% supported, 22.8% neither supported nor opposed, and 29.6% opposed. A chi-square test of independence showed a significant association between policy support and satisfaction with the program, X2 (4, N = 2,319) = 993.55, p < 0.001.
8. A measure of program support is not included in the multivariate analysis. Although there is clearly a relationship with program satisfaction, the relationship is highly endogenous. As such, we focus on the business characteristics that are likely exogenous to program satisfaction in the multivariate analysis.
9. The Pew Charitable Trusts, “Employer Barriers.”
10. 0The statistical model showed limited evidence of independent associations between industry category and satisfaction levels. Leisure and hospitality, the largest industry category and the category with the second-lowest level of satisfaction, was used as the reference category in the model. Holding all else equal, only education and health services and other and unclassified were statistically more likely to be satisfied as opposed to dissatisfied. The other categories were not statistically significant when controlling for other employer characteristics.
11. A test of proportions indicated that the difference in satisfaction between internal and outsourced payroll was statistically significant, Pr(|Z| > |z|) = 0.0064. 12 The Pew Charitable Trusts, “Employer Barriers.”