Office Of Refugee Resettlement: Resources

Synthesis of Findings From Three Sites

F. Housing

Despite low wages and incomes, about 30 percent of refugees in Houston and Miami and 38 percent in Sacramento reported owning their own home. Very few received public housing assistance (public housing or Section 8 housing) in Miami, 17 percent received assistance in Houston, and 19 percent received assistance in Sacramento (see Table VI.10).

Table VI.10: Housing

Measure	Houston	Miami	Sacramento
Housing (%)
Own with mortgage or loan	27.6	26.9	37.9
Own without mortgage or loan	*	3.9	*
Rent	69.8	65.1	60.5
Occupy without payment of cash rent	1.6	3.9	0.0
Public programs (%)
Public housing	11.7	2.1	8.5
Section 8 housing	5.1	*	10.2
Receipt of energy assistance	6.8	*	41.6
Number of bedrooms in home (%)
No bedrooms	2.2	6.3	0.0
1 bedroom	30.5	25.8	8.3
2–3 bedrooms	57.8	61.3	69.2
4 or more bedrooms	9.2	6.0	21.9
Crowded housing (%)
2 or more household members per room	11.2	7.3	11.7
Average monthly housing expenses ($)	663	949	1,068
Sample size	316	335	306

Source: Refugee Assistance Survey
* Indicates a category that contains fewer than five individuals

The most common type of housing was two- and three-bedroom units, with over half of all refugees living in this type of housing in all sites. Houston and Miami refugees were more likely to live in homes with fewer than two bedrooms. While refugees in Sacramento had more bedrooms, they were also more likely to have two or more household members per room. Monthly housing expenses were highest in Sacramento ($1,068), reflecting both the higher housing costs in California and the larger families, followed by Miami ($949) and Houston ($663).

Back to Table of Contents

VII. Relationship of Program Services to Outcomes

The statistical analysis in this section expands on the descriptive analysis discussed earlier in this report by presenting the findings from the multivariate regression analysis. While descriptive analysis illustrates how outcomes vary by participant characteristics and services received, it does not establish clear relationships between participant characteristics, services received, and outcomes. Regression analysis, on the other hand, examines the partial effect of each variable on an outcome while holding all other variables constant.48 The results of the analysis demonstrate which client characteristics or conditions are statistically associated with various client outcomes and the strengths of the relationships.

The regression analysis utilizes data from the administrative sources for both Houston and Miami. Given the high rate of missing information for some of the key characteristics in the Sacramento administrative data, Sacramento could not be included in the comparative regression analysis of administrative data. However, the discussion includes regression results from Sacramento survey data when they are relevant and statistically significant.

As with all studies using regression analysis, this analysis has some potential limitations and should be interpreted with caution. While regression analysis shows the relationship of independent variables to the dependent variables, this does not necessarily imply causality. The analysis does show, however, which factors are associated with service receipt and employment outcomes.

Back to Table of Contents

A. Service Receipt

The first regression model examines the relationship between the type of services refugees received and fiscal year cohort. It controls for differences in sociodemographic characteristics, country of origin, and service receipt in the first two years. As Table VII.1 shows, there are some differences between Houston and Miami in the factors that predict service receipt.

In Houston, males were 17 percent more likely than females to receive employment services, yet gender did not appear to be associated with the receipt of ESL services. The opposite is true in Miami, where men were slightly less likely than women to receive both employment and ESL services. Survey data in Sacramento has a pattern consistent with Houston; men were 12 percent more likely than women to receive employment services, but gender was not correlated with the receipt of ESL services.

In both Houston and Miami, being married reduced the likelihood of receiving employment services, yet it was not significantly correlated with the receipt of ESL services. In Houston, those who completed high school were 14 percent more likely than those who had not completed high school to receive employment services. At the same time, high school completion did not have a significant association with ESL participation in Houston. In Miami, completion of high school was correlated with a higher probability of receiving both types of services.

Not surprisingly, speaking English at entry into the United States was correlated with lower likelihoods of participation in ESL services in both Houston and Miami. Speaking English at entry also reduced the likelihood of receiving employment services in Miami, but it did not have a statistically significant association in Houston. Being an asylee did not have any significant correlation with the receipt of employment services in either Houston or Miami. Yet, asylees in Houston were 26 percent less likely than non-asylees to participate in ESL services. In Miami, asylees were 8 percent more likely to participate in ESL services than non-asylees.

Because each study site consists of refugees from different countries of origin, it is not possible to compare country of origin across the sites. However, regression results show that in both Houston and Miami, country of origin was, in most cases, correlated with the likelihood of service receipt. Cubans were more likely to receive employment services in Miami, relative to other groups, but there is little association with receiving ESL. African refugees in Houston were more likely to receive employment services in Houston and less likely to receive ESL. This might be because some of the refugees from Africa speak English. Survey data from Sacramento also suggest that country of origin is related to the likelihood of service receipt for ESL services (refugees from the former USSR were more likely to receive ESL). There is not a significant correlation, however, between country of origin and employment service receipt.

Finally, the type of services that the refugees received in their first two years after entry was often correlated with other service receipt. In both Houston and Miami, receipt of education services during the first two years was positively correlated with the receipt of employment services. In Miami, receipt of education services was also positively correlated with the receipt of ESL services. Receipt of employment services in both study sites was negatively correlated with ESL service receipt.

Table VII.1: Regression Results: Employment Service and ESL Service

	Employment Service a				ESL Service b
	Houston		Miami		Houston		Miami
Fiscal Year Cohort c
2001			0.025	***			0.018	***
2002			-0.051	***			0.148	***
2003	0.029		-0.176	***	0.212	***	0.307	***
2004	0.054		-0.149	***	0.307	***	0.287	***
Sociodemographic characteristics
Age at entry	-0.001		0.005	***	0.005		-0.008	***
Age at entry squared	0.000		-0.000	***	-0.000		0.000	***
Male	0.172	***	-0.006	*	-0.005		-0.029	***
Married	-0.123	***	-0.008	**	0.019		0.003
Completed high school	0.136	***	0.034	***	-0.020		0.053	***
Speaks English at entry	0.015		-0.304	***	-0.123	***	-0.036	***
Asylee	-0.009		0.089		-0.264	***	0.081	***
Country of origin d
Miami
Haiti			-0.370	***			-0.009
Colombia			-0.228	***			0.018
Other, non-Cuban			-0.144	***			0.065	***
Houston
Sudan	0.092	*			0.188	***
Cuba	-0.098	**			0.140	***
Vietnam	-0.036				0.154	***
Other, non-African	-0.093	***			0.133	***
Service receipt in first two years
Education	0.082	*	0.036	***	0.066		0.333	***
Employment services					-0.122	***	-0.072	***
ESL	-0.120	***	-0.091	***
Case management/orientation	-0.106	***			-0.033
Driver’s education	-0.029				-0.008
Constant	0.567	***	0.789	***	0.216		0.201	***
Observations	1,674		52,266		1,674		52,266
R-squared	0.144		0.413		0.122		0.287

Sources: RSS and TAG program data provided by states
* significant at 10%; ** significant at 5%; *** significant at 1%

a For Houston, employment service since entry; for Miami, in first two years
b For Houston, ESL since entry; for Miami, in first two years
c For Houston, excluded category is 2002; for Miami, excluded category is 2000
d For Houston, excluded category is Other Africa; for Miami, excluded category is Cuba

Back to Table of Contents

48 Technically, the equation estimated the form Y i = β 0 + β 1X 1i + β 2X 2i + …. + ε i., where Y i is the value of the outcome for person i, the variables X 1i through X ni are the explanatory variables for person i in the model that are hypothesized to affect the outcome, ε i is a random term that indicates that the model cannot perfectly predict Y, and the β terms are the parameters to be estimated.