The research was conducted in a 16-county area of east-central North Carolina in which a large number of migrant farmworkers are employed. The counties are Caswell, Craven, Cumberland, Duplin, Edgecombe, Greene, Halifax, Harnett, Johnston, Lenoir, Nash, Person, Sampson, Wake, Wayne, and Wilson. These counties are served by four organizations that participated in the research: North Carolina Farmworkers Project, Carolina Family Health Center, Kinston Community Health Center, and Piedmont Health Services.
This research focused on housing occupied by migrant farmworkers. All participants in this research resided in employer-provided housing, which is generally referred to as a “camp.” In North Carolina, virtually all migrant farmworkers reside in employer-provided housing. Lists of camps were obtained from the partner organizations. Over the course of data collection, the list of camps was expanded as new camps were encountered. All identified camps were contacted to participate in the study. Project field staff traveled to the camps and explained the nature of the study to residents. If camp residents agreed to participate in the study, a camp census was conducted to assess general camp characteristics and to determine eligibility. A total of 186 camps were enrolled in the study. Residents in an additional 36 camps declined to participate, and the grower refused to permit participation in another four camps. The resulting camp participation rate was 82.3 percent (186/226). Data collection was not completed in five of the186 participating camps due to intervention by the grower. For three of these camps, insufficient information was collected on adherence to housing regulations and they could not be included in this analysis. Therefore, the final sample for this analysis consisted of 183 camps. Each camp that participated in the study was given a volleyball as a token of appreciation.
Three residents of each camp were selected as participants. Inclusion criteria were being male, being currently employed as a farmworker, migrating for employment, and residing in the camp. Two residents were asked to complete an interview questionnaire, help with assessing their sleeping rooms, and provide biological samples. One resident was asked to help with a camp and housing assessment. The final sample included 371 men who completed interviews and 182 men who assisted in the camp assessments; 231 men refused to participate when asked. The participation rate was 70.5 percent (553/784); however, the true rate could be lower, as individuals who did not want to participate could have avoided the recruiters. Each participating camp resident was given a $30 cash incentive.
Two procedures were used to collect data used in this analysis: interviews completed with the two residents in each participating camp; and the camp assessment conducted with the assistance of a resident farmworker. Data collection forms were developed in English and translated into Spanish by a native Spanish speaker familiar with Mexican Spanish. The forms were reviewed by staff members of the community partners who were native Spanish speakers. Revised forms were field-tested, with the interview questionnaires being pretested with four male migrant farmworkers. All materials were revised based on the field test.
Interviews were completed by trained staff members who were fluent Spanish speakers. Interviews were completed in a location in which the participant was comfortable, generally their sleeping room. Interviews assessed demographic information, housing features, and perceptions of housing quality. Farmworkers who completed the interviews helped with an assessment of their sleeping rooms. Interviews took approximately 90 minutes to complete.
Housing assessments were completed by trained staff members who were fluent Spanish speakers with the assistance of a farmworker. The housing assessment form included 129 items. The inspector observed, asked questions, and used instruments such as a flashlight and an extending mirror to inspect inside cabinets and behind appliances for signs of pest infestation and exposed wires. The housing assessment form consisted of five sections: 1) general camp; 2) toilet facilities; 3) bathing and showering facilities; (4) kitchen/eating area; and 5) laundry facilities. Inspections took 60 to 90 minutes to complete. Housing assessments also included recording temperature and relative humidity in participants’ sleeping rooms and a common room using a hygrometer, with measurements taken three times consecutively.
Housing safety, security, hygiene, and privacy were the foci for this analysis. Housing safety included three dichotomous measures of structural problems: 1) the presence of water damage in either of the participants’ sleeping rooms; 2) the presence of holes or other damage in dwellings’ floors, walls, or ceilings; and 3) the presence of leaks when raining. A final measure was the sum of the three individual structural problems. Housing safety also included measurement of temperature in the participants’ sleeping rooms. The heat index was based on room temperature, and had four values: 1) no danger, heat index < 80; 2) caution, heat index ≥ 80 and < 90; 3) extreme caution, heat index ≥ 90 and < 105; and 4) danger, heat index ≥ 105.
Housing security measures addressed feeling secure, locks for the dwelling, storage, and communication. The first measure of housing security was based on participants’ responses to an interview item which asked how secure the doors and windows of their dwelling were for protecting their belongings and themselves (“¿Qué tan seguras son las puertas y ventanas de su vivienda para proteger sus pertenencias (ejemplo: dinero, documentos) y usted?”). If either of the two interview participants answered, “not secure,” the camp was coded as “belonging and self are not secure.” Measures of dwelling locks included whether participants had a key to the exterior doors of their dwelling, whether the dwelling doors had locks, and whether the dwelling windows had locks. A final measure was the sum of the three individual lock measures. Measures of storage included whether the participants felt they had sufficient space for storing personal items, sufficient space for storing food in the refrigerator, and sufficient space for storing food and cooking equipment in the kitchen. A final measure was the sum of the three individual storage measures. Measures of communication included whether participants had access to a phone for emergencies, had access to a phone for personal calls, or had limited phone access; and whether a mobile signal was always available.
Hygiene measures addressed cleanliness, odor, and drinking water. A lack of cleanliness was assessed using eight dichotomous measures: whether the camp was unclean upon participants’ arrival, and, at the time of the inspection, the bathroom was not clean, the shower had mold, the kitchen was not clean, the buildings had roach infestation, the buildings had mouse infestation, the outside trash receptacles were without tight-fitting lids, and a working washing machine was not available. These were summed to a total score, with lower scores indicating cleaner housing. An additional measure was the sum of the eight individual hygiene measures. Odor included two dichotomous measures: whether participants, while in the camp, could smell animals in a nearby concentrated animal feeding operation (CAFO); and whether their sleeping room had a musty or mildew smell. The survey included a dichotomous measure of whether the participants drank the tap water in their camp; participants who did not drink the tap water were also asked whether each item on a list of possible concerns was a reason that they did not drink the tap water.
A single measure of privacy concerns was constructed with the values: no privacy concerns, one privacy concern, two privacy concerns, and three or four privacy concerns. Camps were rated as having one privacy concern for each of the following: if the toilets were not separated by a solid barrier, if toilets did not have doors or curtains, if showers were not separated by a solid barrier, and if showers did not have doors or curtains.
Eight camp characteristics were included as predictors in this analysis. H-2A status was a dichotomous measure indicating whether any farmworkers with H-2A visas were living in the camp. Housing type had the values: any barracks in the camp versus no barracks in the camp. Camps with barracks could also have non-barrack housing, such as houses and trailers. Non-barracks camps had only houses or trailers. Number of camp residents was classified into three categories: 1 to 10, 11 to 20, and 21 or more. Number of housing units in the camp had the values 1, 2, and 3 or more. Presence of female residents was a dichotomous measure, as was the presence of a posted NCDOL certificate of inspection. All camps inspected by the NCDOL should post the certificate of inspection; as all of the camps included in this study housed migrant farmworkers, all should have been inspected. Hidden camp is a dichotomous measure: a camp was classified as hidden when it was obstructed from view from any public road by natural or constructed structures (e.g., dense trees, barns), or more than 0.15 miles from any public road. The hidden status of each camp was based on direct observation by the project data collection teams to indicate whether the camp was obstructed from view, and information from Google Earth satellite imagery to indicate whether the camp was more than 0.15 miles from any public road.. Geographic information system (GIS) software (ESRI ArcInfo (Version 9.3 ESRI Redlands, CA; Google Earth Pro, Mountain View, CA; and Atlas. ti 6.0; Cologne, Germany) were used in providing the analysis for this measure. The final measure was data collection period, which had the values of early season (June through mid-July), mid season (mid-July through August), and late season (September and October).
Descriptive statistics were used to describe the various camp characteristics. Bivariate associations of discrete housing quality attributes with camp characteristics were assessed using chi-square tests or Fisher’s exact tests when necessary. Continuous housing quality attributes whose distributions were approximately normal were examined using analysis of variance (ANOVA). All statistical analyses were performed using SAS 9.2 (Cary, NC). A p-value of less than 0.05 was considered statistically significant.