Evaluating the impact of free private well testing outreach on participants’ private well stewardship in New Jersey

Over 1 million people in New Jersey (NJ) are estimated to receive drinking water from private wells. The most commonly detected contaminants in NJ private well water are naturally occurring arsenic and gross alpha (8.3 and 10.9%, respectively). Between 2015 and 2018, three free and voluntary private well testing events tested a total of 571 at-risk wells and 226 (40%) were identified as having one or more contaminants exceeding drinking water standards. Participants were invited to complete a survey to evaluate household characteristics, participant experience, and private well stewardship behavior patterns. Of 529 delivered surveys, 211 (40%) participants completed surveys. Among respondents, 63% reported plans to test their private wells in the future. Among failed wells, 45% of households reported performing mitigative action in response to the event, either through the installation of water treatment system or switching to bottled water. The survey evaluation identified previous knowledge of well contamination risks and discussing test results with a third party as important factors for promoting self-reported stewardship behavior. The evaluation provides guidance for outreach organizers to develop effective testing events and further considers the private well owners’ experience of the outreach events to identify information for ‘best practices’ and improvements of future programs.


INTRODUCTION
An estimated 1 million people in New Jersey (NJ) receive their drinking water from private wells (Flanagan et al. 2016). While public water supplies are protected under both the federal and state Safe Drinking Water Acts, private well owners are solely responsible for monitoring and maintaining the quality of their drinking water (Shaw et al. 2005;Zheng & Flanagan 2017). Chemicals from natural or anthropogenic origin occur in the environment and contaminate private well water supplies (DeSimone et al. 2009). More than 20% of tested wells in the US have one or more contaminants exceeding standards (Colley et al. 2019). Exposure to contaminants over the recommended drinking water standards (USEPA 2020) could result in acute and chronic health conditions including gastrointestinal illness, cancer, cardiovascular disease, and developmental or birth defects (Lewandowski et al. 2008;Cappello et al. 2013;Paul et al. 2015;Zheng & Flanagan 2017). The most commonly detected contaminants to exceed primary Maximum Contaminant Levels (MCLs) in NJ private well water are arsenic in Northern NJ and gross alpha in Southern NJ (8.3 and 10.9%, respectively) (NJDEP 2021), which naturally occur in the geologic formation. events were provided with a recruitment or informational flyer regarding event details including contact information for the event organizers. Tailored results letters were structured to be easy to understand and provide clear recommendations and outlined health risks associated with exposure to contaminated drinking water.
A total of 571 wells were tested. For each outreach event, water analyses were performed by a certified laboratory, and all participants received individualized letters through email or standard mail which included information regarding recommendations on future testing, water treatment, health effects, and additional resources. In early 2019, participants in the events were mailed the same self-administered surveyapproximately 1-3 years following participation in one of the three outreach events. To encourage survey return, participants were mailed a tote bag which said 'Protect Your Family -Test Your Well water' upfront in appreciation of their time and a prepaid return envelope.

SURVEY
The survey contained three broad sections: household characteristics, event experience, and stewardship actions (see Supplementary Material). Household characteristic questions included open-ended and multiple-choice questions including whether a pregnant individual resided in the household at the time of participation in the event, number of adults in the home (coded as '1 to 2 adults' and 'greater than 2 adults'), number of children in home ('0 children' or '1 or more children'), number of children outside the household who use their tap water ('0 children' or '1 or more children'), the age of the youngest child living in the home (in years), residence status of the home (homeowner or renter), years lived in home ('1 to 10 years' and 'greater than 10 years'), frequency of using well water for drinking (coded as frequent 'mostly or always' or less frequent 'rarely, sometimes or never'), and discussion of test results with someone outside the home ('yes' or 'no'). The combined total number of children utilizing the tap water ('0 children' or '1 or more children') was calculated from survey responses and respondents were coded to the event they participated in (Event A, B, and C). If any of the tested contaminants described above exceeded their respective NJ MCL, the well test result was deemed as fail, and test results were coded as a binary variable (pass, fail).
Questions regarding event experience included participant feedback on communication during the event, satisfaction with their participation in the event, and prior knowledge before event participation. There were six questions pertaining to communication, three regarding satisfaction and three regarding knowledge prior to the event. The event experience questions were asked in a Likert scale range including 1 (disagree), 2 (neutral), and 3 (agree) and an additional 'not applicable' category was included for responses to communication questions when a respondent did not have a relevant communication experience ('not applicable' was coded as missing). The questions in each group (communication, satisfaction, and prior knowledge) were summed for a total score and the average score of each category was obtained. Additionally, the event experience variables were categorized as either 'agree' (equal to maximum total score) or 'disagree' (less than maximum total score).
The primary self-reported stewardship outcomes included multiple-choice responses to the following: having previously tested their well water before their participation in the event ('previous testing'), reporting plans to test their well water in the future ('future testing'), and performing mitigative action due to event recommendations defined as installing water treatment system or using bottled water as a source of drinking water ('performed mitigative action').

DATA ANALYSIS
Descriptive statistics and analyses were conducted using IBM SPSS statistical software (version 26.0). Pearson's Chi-square test, with a significance level of 0.05, was used to compare differences in proportions between Pass/Fail results of respondents and by each of the three event locations ('A', 'B', 'C') and Kruskal-Wallis test was used to compare differences in medians. Total scores, means, and standard deviation were calculated for event experiences (communication, satisfaction, and knowledge) and were analyzed by each event. Differences between event experience and event location were determined using one-way ANOVA, with significance , 0.05. Binary logistic regression was performed to estimate the association between dichotomized household participant characteristics and dichotomized event experience ('agree' v. 'disagree') on stewardship behavior outcomes ('future testing' and 'performed mitigative action').

Participant characteristics
Among the 571 mailed surveys, 34 (6%) of surveys were unable to be delivered (return to sender, unable to forward, or vacant), and 8 (1%) participants moved since their participation in the outreach event for a remaining total of 529 eligible households. A total of 211 (40%) completed surveys were returned and included in the analysis (five surveys were unable to be paired with private well test results and their event location and excluded from analysis). While survey responses cannot be compared between respondents and non-respondents, there was no significant difference between percentage of failed wells between survey respondents and non-respondents (data not shown). Between the three events, Event A participants had the lowest response rate at 28% (n ¼ 105 out of 376), while Events B and C had comparable response rates of 58% (n ¼ 58 out of 100) and 50% (n ¼ 47 out of 94), respectively (Table 2). There was not a significant difference between the percentage of survey respondents with passing water results versus failed (p-value ¼ 0.18) Event C had the highest rate of failure (81%), followed by Event B (56%), and Event A (30%) ( Table 2).
The following results are summarized in Table 2. Survey respondents were mostly homeowners (91% (n ¼ 192)) with 16 years as the median years in residence (range 0-70). More than half of the households had children living in the home (56% (n ¼ 118)) with a median age of the youngest child being 11 years old (range 0-19) and 27% (n ¼ 56) of respondents reported that additional children, besides their own, use their well water. Most households (67%) reported using well water for drinking purposes 'mostly' or 'always.' There were no significant differences between these characteristics among respondents with failing water results compared with those with passing results. The only significant difference was the frequency results were discussed with someone outside the home, with 74% of households with failing results reported discussing results with someone outside of the home compared with 60% of households with passing results (p , 0.05) ( Table 2). Of the respondents who reported discussing their results with another person, most reported discussing their test results with either a friend/relative (25%) or a neighbor (23%), while another 33% (n ¼ 66 out of 202) of survey participants did not discuss results with anyone (results not shown).

Participant event experience
A majority of the participants reported that communication of results (83%), risks (72%), and recommendations (70%) were clear. Participants also reported high satisfaction regarding the event, with 84% reporting that the events were easy to participate in, 85% were glad they participated, and 84% reported that they would participate again. Prior to participating in the event, 12% of participants reported having no prior knowledge of possible private well contamination, 7% reported not knowing they should test their private wells, and 17% reported not knowing their wells were vulnerable to contaminants (data not shown).
There was not a significant difference in respondents' overall experience with communication across the three events (p ¼ 0.13) (Table 3). However, respondents who participated in Event C reported having lower trust that event organizers would keep them up to date with any changes compared with higher levels of reported trust among Event B respondents (p-value , 0.05). Satisfaction with participation in the event was generally positive but there was a significant difference between the events, as Event C had the lowest levels of satisfaction (p , 0.001). Although not indicative of event experience, participants were asked to rate their prior level of knowledge pertaining to private well water risk. Event C participants also reported significantly lower levels of prior knowledge compared to the other two events (p , 0.001) ( Table 3).

Stewardship behavior
Most respondents reported testing their private wells prior to participation in the event (79%, n ¼ 161). With the option to select multiple responses, the most common reasons given for previously testing wells included real estate transaction (52%, n ¼ 106), participation in a prior community event (19%, n ¼ 38), and concern for water quality (26%, n ¼ 53); 21% (n ¼ 43) of participants reported not having previously tested their well (Supplementary Material, Table 1). A total of 130 out of 205 (63%) survey respondents reported that they plan to test their private wells in the future; reasons for testing included: to confirm recent test results (16%), to confirm the effectiveness of a water treatment system (9%), and general plans to test in the future (48%). There is no significant difference in reported plans of future testing between household with passing or failing test results (Supplementary Material, Table 1).
Among survey respondents with failed test results who responded either 'Yes, now drink bottled water' or 'Yes, installed a treatment system' (n ¼ 93), 42 (45%) reported taking action in response to the event, with 18% (n ¼ 17) reported using bottled water as a drinking source and 29% (n ¼ 27) reported installing a water treatment system. Of the participants reporting that they will perform no action due to having safe water (n ¼ 70), 26% had failed test results. Of the people who reported having treatment installed prior to private well testing event (n ¼ 53), 28% had failed test results. Among the events, Event A participants reported a higher percentage of the previous testing due to participation in a prior event (Supplementary Material, Table 1). Overall, 54% (n ¼ 114 out of 211) of survey participants reported some type of water treatment system already installed or which they installed following event participation, while 21% (n ¼ 44 out of 211) of households utilized bottled water as a drinking water source prior to and after event. In total, 59% of households reported utilizing some form of mitigation before or after event participation (Supplementary Material, Table 1). Survey respondents who reported discussing their private well test results with someone have greater odds of planning future well testing by a factor of 2.33 (95% CI: 1.26-4.30) compared to individuals who did not report discussing results (Table 4). Individuals who had prior knowledge of well water risks were 1.99 times (95% CI: 1.12-3.55) more likely to plan future well testing compared to those who had no prior knowledge. There were no significant differences between the household characteristics and whether a household reported plans to test their private well again in the future.
Among the wells with failed test results, survey respondents who reported discussing their results had higher odds of reporting taking mitigative action, when compared to those who did not discuss by a factor of 2.91 (95% CI: 1.02-8.32). Perception of the event experience was not found to be significantly associated with taking recommended mitigative action (Table 4). There were no significant differences between household characteristics or event experience and whether a household reported taking mitigative action due in response to recommendations (when results were restricted to failed households).

DISCUSSION
These voluntary private well testing events helped to increase the number of wells tested in NJ and raise awareness of potential risks of private well contamination. Overall, households reported a positive experience from participating in the water testing events, with a majority reporting that the communication throughout the event was timely and clear and that they would participate again in a similar event. This is encouraging as clear dissemination of results and recommendations can lead to a greater understanding of the issue and allow well owners to take action accordingly. Insight into the participants' opinions of the testing events can help strengthen future outreach programs. A majority of participants discussed their results with individuals outside the home, with households with failing results having higher rates of discussion. Individuals who discussed the event's test results with someone reported significantly higher rates of planning future testing than individuals who did not discuss results. In addition, households who discussed test results were 2.9 times more likely to report taking mitigative action, similar to findings published by Flanagan et al. 2018. Among the events, Event B had the largest percentage of households who discussed their test results. Project coordinators strongly encouraged participants in Event B to share their results with friends and neighbors, and it is encouraging to learn that this behavior is associated with increased stewardship during these events. Several studies have previously reported an association between stewardship behavior and discussion, which could be due to social network influencing behavior (Renaud et al. 2011;Colley et al. 2019).
Although it is difficult to confirm if actual stewardship action occurred, it is encouraging to note that verbal acknowledgement of intentions can increase the likelihood of said behavior (Morris et al. 2016). Studies also indicate that social networks are an important factor for information dissemination, cues to action, and setting social norms for behavior (Colley et al. 2019). Discussion of the testing events could further encourage individuals who did not directly participate in the events to test their own wells, as studies have found that individuals are more likely to be concerned for health and test wells if they live close to wells, which have been found contaminated (Morris et al. 2016;Flanagan et al. 2020). Future private well outreach initiatives may consider offering free testing 'coupons' for participants to share with their neighbors to further take advantage of the impact of social networks on subsequent mitigative action. Respondents who reported previous knowledge of risks of well water contamination were almost twice as likely to report plans to test their private well in the future. Therefore, it is possible that community-level private well educational efforts may improve private well stewardship following private well testing outreach.
Of the households who participated, most respondents reported having more than one adult living in the home, having at least one child in the home, and having additional children living outside the home utilizing the well water. Therefore, the well testing events have the potential to reduce hazardous drinking water exposure to the whole family and other water users; especially as most participants reported using their private well water for drinking. Event A had the greatest number of well samples tested and the greatest number children in the home. This event used a school-based method to recruit participants allowing targeting of homes with children (Rockafellow-Baldoni et al. 2020). By having children deliver the invitation to participate in the well testing may have motivated well owners to participate. Furthermore, children are particularly vulnerable to the developmental effects of contaminants, including arsenic and gross alpha, making them an important target group for hazard reduction.
Respondents from Event B scored communication during the event more positively as compared to respondents from Event C. Although the survey tool cannot be used to assess why scores were higher, some differences in Event B implementation include relying largely on online sign-ups with email required, although mail-in registration and communication was also available. The online sign-up allowed for program organizers to more easily and frequently communicate with participants on a regular basis including registration confirmation, reminders and follow-ups, and invitation to a public meeting to provide an overview of results (presented at an aggregate level to maintain confidentiality) and answer questions regarding health effects and treatment. Unlike Event C, for which communication was scored the lowest, both Event A and B were cosponsored by the local environmental health commissions which may have provided a community contact in which some participants may have more trust.
Most survey respondents (79%) reported having previously tested their private well. This rate is comparable between the event locations and with previous studies, where overall testing rates in high-risk areas were above 82% in New Jersey and 78-89% in Maine (Flanagan et al. 2016;Malecki et al. 2017). Of those who reported previously testing, almost half (52%) did so due to a real estate transaction, indicative of the effectiveness of NJ PWTA increasing testing rates. Of the individuals who reported previously testing their private wells, 39% failed the recent test results. This could suggest either previous testing did not include testing for contaminants targeted in the outreach events or lack of subsequent installation and/or maintenance of an effective treatment system. It is also possible that contaminant levels have changed over time, which highlights the importance of regular private well testing (Flanagan et al. 2016;Malecki et al. 2017).
Over half of participants reported plans to perform future testing (63%), regardless of the test result from the private well testing events. This means that individuals may continue to monitor private wells and thus reduce the risk of exposure, although the intention is not a guarantee that they will follow through on their plan. Individuals who reported having previous knowledge of private well risks also had a greater likelihood of reporting plans to perform future testing. Several studies have demonstrated that knowledge is an influencing factor, with lack of knowledge being negatively associated with well testing (Colley et al. 2019).
When asked if any action was performed due to recommendations of the event, 42 of the 93 (45%) failed households indicated either installing a water treatment system in the home (29%) or switching to drinking bottled water (17%). Additionally, 42 of the 119 (35%) failed households reported previously performing mitigative action prior to the testing event, with households reporting already having a water treatment system installed (28%) or utilizing bottled water as a drinking source (14%). In another survey of New Jersey residents, 54% of households with an identified problem installed a treatment system, while 10% reported resorting to drinking water bottles (Flanagan et al. 2016). While in Maine, 43% installed a treatment system and 27% switched to bottled water (Flanagan et al. 2016). Self-reported mitigation does not guarantee that households will install and maintain proper treatment systems (Flanagan et al. 2018).
Among those that reported having treatment installed prior to participation in the event (n ¼ 53), 49% still had failed results, and among respondents who reported taking no action because they have safe water (n ¼ 70), 26% had failing water results. This percentage of taking no action to reduce exposure is similar to rates seen in other studies which range from 27 to 40% (Flanagan et al. 2018). These results show that there are still individuals at risk of hazardous drinking water, due to either the inability to understand their results letter which explains that their water test failed, the mistaken belief that a failed water test is not a health risk, that they have an ineffective or unmaintained water treatment system, or they do not have the resources to obtain a treatment system. Studies indicate that individuals often inaccurately recall test results, and oftentimes downgrade problematic information (Malecki et al. 2017). Despite providing free water testing, communication regarding the need and maintenance of treatment is critical. Future outreach initiatives should seek to increase the comprehensibility of results and recommendations. Although the majority of participants felt the communication of results and recommendations were clear, there were households who disagreed with statements on clear communication (with 6% of households reporting unclear risks, and 7% reporting unclear recommendations). In addition, without results of the raw water, participants may not appreciate the underlying danger; and without results post-treatment, participants may have a false sense that they have effective treatment. Testing both raw and treated water may allow outreach organizers to communicate risk and provide recommendations more effectively.
This work is one of the first to evaluate subsequent private well stewardship behavior of participants participating in free and voluntary private well testing outreach in New Jersey. Several studies have published results from surveys of New Jersey private well owners' stewardship behaviors and attitudes, but these studies investigated the effectiveness of the PWTA (Flanagan et al. 2016(Flanagan et al. , 2018. Similarly, some studies used mailed surveys to collect data on stewardship information after dissemination of test results, but these studies differ as they recruited participants from different populations, only focused on the behavior of households with failed results (with focus on arsenic), or acquired well test results from geological survey testing (Severtson et al. 2006;Flanagan et al. 2015aFlanagan et al. , 2015bMalecki et al. 2017). Future direction can include further analysis of underlying demographic factors of the participants which could influence stewardship outcomes. Furthermore, future studies are needed to confirm if participants of private well testing events actually follow through on their self-reported plans to identify the effectiveness of programs.

LIMITATIONS
This study had a few limitations. First, an original questionnaire was developed to evaluate the effectiveness of private well testing events and results may not be comparable to previous studies. There is also the possibility of bias in the study, due to the nature of self-reporting. With differences in the time frame between the event and the survey, there is the possibility of recall bias which could differ among event respondents due to the date of survey and differing dates of participation. In addition, non-response bias could occur, as the response rate was 41% and may not be a representative sample of voluntary event participants. The non-response also limited the sample size, which reduces the power of analysis. The generality of the stewardship questions limited the amount of information gathered. For example, although over half of participants reported plans to test their private well in the future, no information was gathered on the type of testing in which they plan to perform. It is possible that the impact of recommendations to encourage participants to take mitigative action is underestimated since the questionnaire does not capture those that may still take action. Lastly, this survey captured data from participants from three different private well testing events. The difference between the events could influence results and interpretation, including the contaminant of concern tested for and treatment options. For instance, effective treatment for radium is a water softener, which is commonly installed for the treatment of water hardness and iron, a common problem in some areas, and is more affordable than the treatment for arsenic which is also more difficult to install and maintain (Spayd et al. 2015).

CONCLUSION
This study not only demonstrates that free and voluntary private well testing events increased the number of wells tested for contaminants of local risk, leading to the reduction of hazardous drinking water in some homes, but also identifies factors that influence self-reported stewardship behavior and can guide outreach organizers to develop effective testing events. This evaluation also provides a framework and baseline to evaluate the success of future free and voluntary private well outreach events and also considers the private well owners' experience of the outreach events, which could provide vital information toward the improvement of future programs including allowing online registration and email communication, co-sponsoring with local trusted partners, providing raw and treated water tests, and encouraging participants to discuss their participation and results with neighbors and friends. Successful outreach programs not only identify barriers to the behavior but incorporate the needs of the community in which they serve (Morris et al. 2016). Securing a sustainable funding stream through federal, state, and local budgets will be integral for the promotion and continuation of free private well testing programs.

DATA AVAILABILITY STATEMENT
Data cannot be made publicly available; readers should contact the corresponding author for details.