Little research has been conducted on the housing modification needs of the elderly. The fault lies partly with lack of reliable data. Studies on housing modification needs of the elderly are often spotty, anecdotal, and unsystematic. Many rely on decennial census data which provide a limited and unsatisfactory portrait of special housing needs of elders in general. This paper seeks to fill this void by reporting a comprehensive study of elderly housing modification needs. The logistic regression analysis found that gender, health, and living alone were the three best predictors of need for special housing modifications. Men, people who live alone, and persons with poor health have greater needs for housing modifications.

Overview

   Lack of housing that is modified for the frail elderly forces many elderly persons to live with relatives or in institutional settings (Wan & Weissert, 1981; Weissert & Scanlon, 1983; Brody, Poulshock & Magciocchi, 1978; Doty, 1986; Smyer, 1980; Palmore, 1976; Greenberg & Ginn, 1979). In examining research on housing modifications, only about ten percent of elderly headed households with at least one member with a mobility problem has made a housing modification to the dwelling (Struyk & Katsura, 1987: 26-29; Katsura, Struyk, & Newman, 1989; Struyk, 1982; Pynoos et al., 1987). The most popular modification is extra handrails or grab bars (7 percent) with all other modifications (flashing lights, raised lettering or braille, sink, faucet and cabinet adjustments, etc.) made by one percent or less of elderly persons with disabilities. Struyk & Katsura (1987), however, examine the American Housing Survey (AHS) to determine modifications in place but do not explore questions which indicate whether certain modifications would help solve mobility and equipment usage problems.

   In examining institutionalization and independent living, Reschovsky and Newman (1990), analyzing a survey of Houston households, found that independent living requires adequate financial resources and informal support (special unpaid assistance from family members, friends, or neighbors). They also note that renting makes independent living more difficult. However, they do not address questions dealing with dwelling modifications, but only informal support. Reschovsky and Newman (1991), using the same data set, find that income is the major determinant of home upkeep.

   In another study of Houston, Gilderbloom & Rosentraub (1990) found that a large percentage of elderly persons require special architectural modifications in the home and an infrastructure that is accessible if they are to age in place.

   Only one major study has analyzed elderly modification needs, through an examination of the AHS Housing Modifications Supplementary Survey (Newman et al. 1990). Newman et al. (1990: S181) attempt a rigorous empirical analysis of the hypothesized positive relationship between modifications for elders with activity limitations and being able to remain at home. The authors conclude that housing modifications are correlated with increased institutionalization. However, those who have modified dwelling units tend to be the most frail persons who live alone (see also Soldo & Longino 1988: 116-119). Insofar as race, whites have a higher rate of institutionalization even when other factors like income, home ownership and living alone are part of the equation. When examining both modifications and support services, moreover, Newman et al. (1990: S183) report results derived from their path analytic model that "living in a specially modified dwelling and relying on formal support increase the probability of becoming institutionalized." In addition, lack of informal support in and of itself without regard to modifications has been correlated with higher rates of institutionalization.

    The Newman et al. (1990) analysis, is however, flawed by several methodological concerns . One major concern is the danger of combining two different data sets, the Annual Housing Survey and the National Long-Term Care Survey, for analysis. As Newman et al. (1990: S174) note, this combination of the two data sets is a "weakness" which might be better solved by looking at one unified data set. The key question of whether housing modifications lead to greater aging in place for those with activity limitations could be addressed by conducting a separate analysis of the AHS. Our study will do a separate analysis of the AHS with greater specification of each dependent variable and procedures that will minimize multicollinearity. Based on a number of demonstrations and studies, providing in-home services to the frail elderly may enable them to live at home better but not necessarily longer. They appear to enter institutions at least at the same rate as those not receiving such services. The same could be true for those benefiting from housing modifications.
 

Procedures

Research Questions

   A central question is whether a demand exists for housing modifications for elderly persons with activity limitations. The AHS supplementary survey was specifically designed to test whether persons unable to perform essential daily living activities could be helped by modified housing. Oddly, few researchers have tapped this data set to answer this crucial question. Moreover, we have no knowledge of the specific kinds of modifications (special telephone devices, braille, special kitchens, etc.) which are vital for independent living, nor are we aware of how needs for these break down by gender, household type, or age.

   To what extent does gender influence which people 65 or older with limited essential daily living activities could be helped by modified housing? How do other factors (living arrangement and health) influence the need for specific modifications (e.g., handrails, braille, special telephones, special features in kitchens and bathrooms) that are vital for independent living?

Data set

American Housing Survey

   The AHS (the Annual Housing Survey from 1973 to 1981 as data were collected annually, and the American Housing Survey since 1981 as data were collected biennially) provides a highly detailed national database of elderly housing needs (Annual Housing Survey, 1978a, 1978c). Funded by the U. S. Department of Housing and Urban Development and administered by the Census Bureau, the AHS survey began in 1973 to examine a wide range of housing concerns (Apgar 1987: 69).

   The American Housing Survey provides a detailed national analysis of elderly housing and neighborhood needs. As a survey of housing conditions, the AHS is a major improvement over the decennial census. The original survey had two major parts: (1) the national survey which examined between 60,000 and 80,000 units on a yearly basis and (2) surveys of 60 selected metropolitan areas in the United States every three to four years. The survey has data on nearly 18,000 total households with a resident 65 years or older.

   The scope, precision, and sophistication of the AHS makes it one of the richest data sets about elderly living needs. For example, the AHS finds that three-fourths of all households report at least one major neighborhood or local problem: litter, police protection, poor schools, abandoned buildings. The survey indicates that over time the quality of neighborhood life has steadily worsened (Apgar, 1987; Goering, 1979; Bratt, Hartman & Meyerson, 1986). The degree and impact of these problems are checked by asking whether they will prompt respondents to move. Less than 16 percent of all renters and homeowners feel impelled to move because of these problems (Goering, 1979: 9). The AHS also allows cities and regions to be compared with one another along with different kinds of housing and neighborhood situations.

AHS Supplementary Survey of Housing Modifications

   An important source of data from the AHS is the housing modification survey conducted in 1978. The Housing Modification Supplement was designed to determine whether certain modifications have been made or would be useful to make the unit more accessible to the disabled (Annual Housing Survey, 1978b; U. S. Department of Housing and Urban Development, 1993). The questionnaire was precise and detailed. The survey attempted to identify the disability and asked a variety of questions measuring whether the home had or did not have specific "barrier free" features: ramps, rails, wide doorways, etc. Because a substantial number of elderly persons have a disabling conditions, these data might be useful to ascertain the housing needs of disabled elderly (Gilderbloom, Rosentraub, & Bullard, 1987).

   This nation-wide data set is one of the most comprehensive analysis of disabled housing needs ever undertaken. Yet few researchers have ever published results from this data set, in part because it is not user friendly. Some might note that the response patterns have changed in 15 years. Reschovsky & Newman (1990: 544), however, note that there have been few changes in the housing situation of the elderly in the past fifteen years.

   The survey used screening procedures to ensure a representative sample of disabled Americans. This provided a database of over 3,000 households of elderly/disabled. The interviewers asked close to 150 questions concerning infrastructure design and housing needs. Thirteen variables show which, if any, of the following special features were in each home:

• handrails or grab bars
• elevator or stair lift
• raised lettering or braille
• special sinks, faucets, or cabinets
• specially equipped telephones
• any other features
• ramps
• extra wide doors or hallways
• push bars on doors
• special wall sockets or light switches
• flashing lights
• no special features

Methods of Analysis

   These data sets provide the opportunity to conduct sophisticated statistical analysis. The data files also allow greater for a more exact determination of the need for specific location, affordability, and design needs of elderly persons as well as desired housing options.

   In order to examine whether housing modifications may assist the disabled elderly to age in place, the supplementary housing modifications data set was merged with the core AHS data set. The supplementary survey defines activity limitations on the basis of four questions:

• Does anyone in this household have difficulty going in or out of this house (apartment)?
• Does anyone in this household have difficulty going up and down stairs?
• Does anyone in this household have difficulty moving around inside this house (apartment)?
• Does anyone in this household have difficulty using the bathroom facilities, kitchen equipment, or other equipment in this house (apartment), such as telephone, light switches, and so forth?

   While the unit of analysis in both the core file and the supplemental file is the housing unit, most relevant data are on the household, and data on activity limitations are on individual members of the household. A significant problem in merging the two databases is that individual members of the household (persons number one through six) do not necessarily correspond. Thus, for instance, person number one in a household in the supplementary database is not necessarily person number one in the core database. To correct for this, various compute statements were written to match the individual persons in both data sets. As such, the demographic independent variables represent the characteristics of the disabled person, not the respondent to the survey.

   Once this was done, there were 1,895 cases of the elderly disabled. Because by far the greatest number of valid responses applied to person number one, this person is used as the unit of analysis. The case depletion from 2,793 to 1,895 is significant and a potential cause of concern. However, an analysis of the demographic characteristics of the cases before and after the person match indicated no significant differences.

   Missing values pose a significant issue. The dependent variables are dichotomous in nature, measuring whether a modification would or would not help the disabled elderly person in activities. A significant number of persons (from about eighty to two hundred) did not answer the questions on modifications. These responses are counted as missing values. An even higher proportion of responses are not applicable. This is because in many cases the modification to assist in performing an activity does not apply to a person because they do not have difficulty in performing that activity, but have trouble in performing some other activity which defines them as disabled. In addition, the cases are weighted in accordance with the original weighting scheme of the clustered sample of the supplementary database to adjust the standard errors. The dependent variables represent the unmet need for housing modifications among the disabled elderly. As such, only persons who do not currently have the modification are included in the analysis. Other research (Struyk and Katsura 1987; Katsura, Struyk, and Newman 1989; Struyk 1982; Pynoos et al. 1987) has examined the role of existing modifications to assist the elderly disabled to age in place. As such, these persons are excluded from the analysis.

1. First, no discussion is made about possible multicollinearity problems which might result in the sign of coefficients for housing modification becoming reversed or insignificant (Lewis-Beck, 1980: 60; Nie et. al., 1975; Norusis, 1982, 1983; Gilderbloom & Appelbaum, 1988; Gilderbloom et. al, 1992). The first sign of possible multicollinearity is when a key variables shows the "wrong" sign. (Lewis-Beck, 1980: 60). This "wrong" sign might be the result of two highly correlated independent variables. For example, instead of running separate equations that separate the highly correlated variables race and income, Newman et al. run the variables simultaneously in the equation. Two other variables that might be highly correlated with each might be income and housing modifications. No table is provided showing the simple correlations between the independent variables (Lewis-Beck, 1980; Nie et. al., 1975) This approach, however, is considered somewhat primitive compared to the more sophisticated tolerance tests which regress each independent variable against other independent variables (Norusis, 1982, 1983).

   A second major weakness with the statistical analysis was the failure to conduct a split regression analysis which would have effectively dealt with potential multicollinearity problem of simultaneously controlling for race and income. A split regression analysis examine make a separate analysis of elderly whites and elderly blacks in the sample and compare housing modification coefficient. This would seem critical given the finding that controlling for income whites are more likely than blacks to need institutionalization. Without the split regression analysis this finding has little meaning.

   A third major weakness is the operationalization of housing modification variable. No explanation is given defining housing modification beyond "grab bars, ramps and raised toilets" (Newman et al., 1990: S177). It is unclear whether this definition also included other AHS items which measure special housing features such as elevators or stair lifts, raised lettering or braille, sinks, faucets or cabinets, specially equipped telephone, extra wide doors or hallway, push bars on doors, wall sockets or light switches, or flashing lights. Combining all of these housing modifications and creating a scale would have made a more powerful predictor variable.

   A fourth weakness was the measurement of the dependent variable which Newman et al. (1990: S174) claim is "not without limitations." Newman et al. (1990: S174) note that measures were used to estimate institutional outcome that were not current but two years old. According to Newman et al. (1990: S174): "This approach is open to criticism not only because many characteristics potentially associated with risk may change over a two-year period (e.g. health status, social support) but also because it may be the change itself rather than prior status that precipitates institutionalization.
 
 

Analysis of Statistical Results

   Through the process described above, we produced a subset of 1,895 households in which there was an elderly person (65 years or older) who was disabled in the sense that their activities were limited by at least one of four difficulties: (1) getting in/out of the unit, (2) getting around inside, (3) getting up and down stairs, and (4) using bathroom, kitchen, or other equipment in the unit. We analyzed what the disabled elderly person said about whether introduction of the following housing modifications would help these individuals: handrails or grab bars, ramps, elevator, extrawide door or hallways, door handles instead of knobs, raised lettering or braille, push bars on doors, special sinks, faucets, or cabinets, specially equipped telephone, and flashing lights.

   For each of the four kinds of difficulty, the interviewers asked about the value of only the relevant modifications because, for example, neither braille nor special telephones will help an elderly person who has difficulty climbing stairs.
Tables 1, 2, and 3 show the results of the bivariate relationship between the need for housing modifications and gender, whether the person lives alone or not, and health, respectively. Table 1 shows that, overall, men more than women feel that a particular modification would help them to age in place. For example, 27 percent of men and only 16 percent of women think that extra handrails or grab bars would assist them in getting in and out of their housing unit. Table 2 shows that persons who live alone feel that a modification would assist them in aging in place more than persons who do not live alone. Table 3 indicates that the need for modifications increases as health becomes worse.

Logistic Regression

    As shown previously, a maximum of 21 percent of disabled elderly persons thought a modification would assist them (handrails or grab bars for going in and out of the unit). Yet, these simple cross tabulations do not simultaneously control for all relevant influences. In order to determine the unmet need for housing modifications, all relevant influences are simultaneously controlled in a logistic regression model. Based upon variables found to be important in previous literature (Gilderbloom & Rosentraub, 1990; Struyk & Katsura, 1987; Reschovsky & Newman 1990, 1991; Soldo & Longino, 1988; Struyk, 1982; Pynoos et al., 1987), Table 4 presents the results of logistic regression for each of the 34 questions taking into consideration demographic, geographic, and unit characteristic control variables that may have an influence on the need for certain housing modifications that would help with the disabled elderly to age in place.
These control variables are:

 
Age (1 = 65B74, 2 = 75B84, 3 = 85 and older)
Health (an index of disabling medical conditions, 0 = good health, 6 = poor health)
Gender (0 = female, 1 = male)
Alone (0 = not live alone, 1 = live alone)
Race (0 = nonwhite, 1 = white)
Income (in 1978 current dollars)
Metropolitan (0 = not live in metropolitan area, 1 = live in metropolitan area)
Suburban (0 = central city residence, 1 = suburban residence)
South (0 = non-South, 1 = South)
Housing adequacy (1 = adequate, 2 = moderately inadequate, 3= severely inadequate)
Age of unit (high = old)
Size of unit (number of rooms)
Detach (0 = attached unit, 1 = detached unit)
Number of units (number of units in structure)
Length of residency (high = longer length of residence)
Own (0 = rent, 1 = own)

    The variables that most frequently predict whether a modification would help a person to age in place are health (significant for 19 questions), whether the person lives alone or not (significant for 18 questions), and gender (significant for 16 questions).

    Whether the elderly person lives alone (single elderly) or not (elderly couple) had the most powerful impact on whether modifications would help them to age in place among all the control variables. For all questions, elderly respondents who lived alone consistently said that a modification would help. This relationship is particularly strong for using equipment in the unit (Bs of from 1.23 to 2.88). These answers seem to suggest not a straightforward answer to "would it help?," but whether such a modification is "necessary," "makes sense," or would be "worth the money." If someone else in the household can do things that are difficult for one elderly resident, then while a modification "would help" in a noncost-conscious context, the expressed answer may often have been "no."

    Gender had the second most powerful impact of all the control variables. For all questions the direction of the influence is the same: men are more likely to say that certain features would help with a particular mobility or use problem.

    The health of the respondent proved to be the third most powerful of whether the housing modification would assist in daily household activities. The AHS Housing Modifications Supplementary Survey contains 39 variables indicating whether the respondent has various disabling medical conditions. These conditions range from arthritis to Parkinson's disease and includes such disabling conditions as asthma, tuberculosis, bronchitis, lung problems, effects of a stroke, cerebral palsy, and diseases of the respiratory system. Each condition was coded as zero (not have condition) or one (have condition). The variables were then summed to yield an index ranging from one to six, with the latter indicating worse health.

    As can be seen from the table, the direction of the impact of health is consistent. As a respondent's health worsens, they are more likely to believe that the modification would assist them in the four areas of household activity. The strength of this relationship, moreover, is moderately strong for numerous questions. These include push bars on doors for getting around inside (B of .76), extra hand rails or grab bars for getting in and out of the unit (B of .52), and special sinks, faucets, or cabinets for using equipment (B of .51).

    Age had a statistically significant but modest impact on housing modifications needs for only one the 34 questions. Clearly, once other factors, especially living arrangement, gender, and health, are controlled for, the age of the elderly person has little impact on the need for dwelling modifications.

    The remaining variables had an infrequent or slight impact on the need for dwelling modifications. Race significantly affected three modification needs. Whites compared to nonwhites said that extra hand rails or grab bars to go in and out of the unit (B of -.70), and to go up and down stairs (B of -.56) would not help. Whites were, however, more inclined to say that an elevator or stair lift would help them to go up and down stairs (B of .93).

    Total family income was likewise used as a control variable. This interval variable measures 1978 current income up to $50,000 (about $120,000 in 1995 dollars). Surprisingly, income significantly affected none of the housing modification questions, despite the fact that a higher income makes modifications more affordable.

    Geographic control variables had an infrequent impact on the need for housing modifications. Persons living in metropolitan areas relative to non metropolitan areas thought that extra wide doors or hallways to get around in the unit would not help, and that no feature would help. Persons living in the South thought that extra hand rails or grab bars would not help in getting in and out of the unit (B of -.58) and in using equipment (B of -.61).

    Characteristics of the unit had very little influence on dwelling modification needs. Housing quality, measured as an adequate, moderately inadequate, or severely inadequate unit, affected no modification need. The age of the unit influenced only elevators or stair lifts in getting in and out of the unit, and extra handrails or grab bars in going up and down stairs. Whether the person lived in a detached or attached unit affect no modification need. The number of units in the structure affected only three, and the length of residency and tenure one of the numerous modification needs.

    All control variables were checked to determine if any multicollinearity problems were present. The highest correlation was between detached units and number of units, which had a correlation of -.66 (significant at the .01 level). A tolerance test was performed as a further test for multicollinearity. These results showed the lowest tolerance to be .45 (for number of units). Thus no evidence of multicollinearity could be found (Lewis-Beck, 1980: 60; Nie et al., 1975; Norusis, 1982, 1983; Gilderbloom & Appelbaum, 1988; Gilderbloom et al., 1992).

    One of the highest pseudo r squares (.47) is for elderly people who have difficulty going in and out of the unit and say that extra handrails or grab bars would help. They tend to be males, persons who live alone, and persons in poor health. This pattern repeats itself for many of the dwelling modifications.

Discussion

    This research has sought a greater understanding of the housing modification needs of the disabled elderly. The results indicate that men more than women, persons who live alone, and persons in poor health have greater housing modification needs. Previous research has shown that the most disabled or those in the worst health would benefit most from dwelling modifications (Newman et al., 1990; Struyk & Katsura, 1987; Reschovsky & Newman, 1990, 1991; Newman, 1985; Katsura, Struyk & Newman, 1989; Soldo & Longino, 1988; Struyk, 1982; Pynoos et al., 1987). Yet, an examination of the actual use of dwelling modifications found that use was "most prevalent" among persons who were slightly disabled (Manton, Corder & Stallard, 1993: 175). Indeed, at higher levels of disability dwelling modifications "appear insufficient to support individuals in the community" (Manton, Corder & Stallard 1993: 176).

    The current research lends nuances and clarifications to these two streams of findings. Previous research found that elderly persons thought that various modifications would help them to age in place (e.g. Newman et al., 1990). Other research on the actual use of dwelling modifications has found that such equipment is used by persons of lower disability levels, as measured by ADLs and IADLs (Manton, Corder & Stallard, 1993). The current research supports previous research in that the expected usefulness of dwelling modifications, in the subjective opinion of the elderly person, increases as health worsens. It does not, however, directly address other research (Manton, Corder & Stallard, 1993) which operationalizes disabilities by ADL and IADL measures to examine current use of housing modifications.

    The current findings suggest that as the health of an elderly person worsens, certain dwelling modifications are thought by the elderly to help in aging in place. Among the most helpful of the modifications are handrails or grab bars to get in and out of the unit, handrails or grab bars to use equipment in the unit, push bars on doors to get around inside the unit, and special sinks, faucets, or cabinets to use equipment in the unit. The current research clearly supports the previous research (Newman et al., 1990; Struyk & Katsura, 1987; Reschovsky & Newman, 1990, 1991; Newman, 1985; Katsura, Struyk & Newman, 1989; Soldo & Longino, 1988; Struyk, 1982; Pynoos et al., 1987), yet also lends indirect support to research on the actual use of equipment (Manton, Corder & Stallard, 1993) insofar as most modifications are regarded as insufficient to assist aging in place. Moreover, as the bivariate analysis in Table 3 indicates, the expected assistance of most modifications in aging in place drops to near zero percent for those in the worst health.

    Previous research has found that women more than men find dwelling modifications would help them to age in place (Manton, Corder & Stallard, 1993; Newman et al., 1990; Newman, 1985; Struyk & Katsura, 1987; Reschovsky & Newman, 1990, 1991; Katsura, Struyk & Newman, 1989; Soldo & Longino, 1988; Struyk, 1982; Pynoos et al., 1987). The current research, however, shows that men would benefit more from housing modifications. The relationship between gender and home modification needs may be explained in the context of well known differences in the type and incidence of various kinds of conditions between men and women (Newman et al. 1990), as well as the fact that women live longer, and in that women are more frail (Manton, Corder & Stallard, 1993: 175).

    Perhaps a more surprising observation that researchers have not pointed out has been the reluctance of individuals who have movement difficulties within the house to endorse modifications as helpful. Why is it that 40 percent of the respondents who have difficulties in daily activities view housing modifications as not helpful? This question deserves further attention by researchers. A preliminary hypothesis might be in order. First, the expected high cost of home modifications may deter the desire for modifications, even when income is controlled. Although an elderly person may think a modification may help them age in place, the sheer cost may be a deterrent to the desire for such a modification. Second, the elderly might see housing modifications as a change that spoils the aesthetic value of the home. In addition, elderly persons may believe that housing modifications would lower the value of their home. This may be a particularly powerful obstacle because of the desire to pass on the value of the home to their heirs. Fourth, resistance to housing modifications may reflect an individual's determination to overcome the disability by getting better instead of accepting the condition. Fifth, as shown by Manton, Corder & Stallard (1993), there may be little that can be done to help those highly disabled to age in place. Finally, the disabled elderly also face the problem of few organizations involved in making home modifications. There is, moreover, the problem of the elderly not trusting the few existing housing modification contractors (Pynoos, 1992).
 

End notes

2. Altogether, three of the answers to the 34 questions showed statistically significant patterns of difference by race. This is somewhat surprising. These results consider race alone, without taking into account any control variables which factor out other influences and often show a variable to be either more important or less important than when considered alone. Given that race failed to predict housing modification needs, what factors can help explain the need? Our analysis continued with logistic regression. This should not lead to the conclusion, however, that no differences exist among elderly whites and blacks for housing and neighborhood needs. Differences between elderly blacks and whites do seem to exist on other measures (social services, quality of housing, neighborhood needs, price of housing) but not on housing modifications (Gilderbloom & Markham 1994).

3 . Because logistic regression is highly insensitive to interval control variables which have a high degree of variation (Aldrich & Nelson, 1984; Demaris, 1992; Hosmer & Lemeshow, 1989; Joreskog, 1977) income was collapsed into the following categories: low-$999=1, $1000-$2999=2, $3000-$3999=3, $4000-$4999=4, $5000-$5999=5, $6000-$6999=6, $7000-$7999=7, $8000-$9999=8, $10000-$12499=9, $12500-$14999=10, $15000-$17499=11, $17500-$19999=12, $20000-$22499=13, $22500-$24999=14, $25000-$29999=15, $30000-$34999=16, $35000-high=17.

4. The "40 percent" also provides a estimate natonally of what percentage of persons need to have attendant care from nursing home to home health care services. This study found that 22 percent of the elderly have activity limitation which prohibits them from performing certain essential activities. Forty percent of those persons with activity limitations would not be helped with a housing modification and require additional attendant care. Consequently, we estimate that 8.8 percent of the elderly currently living outside of nursing homes require greater attendant care. Hence, the number of elderly over 65 who might need attendant care services might be as high as 13.75 if we add the 8.8 percent to the current percentage of the elderly population already residing in nursing homes nationally of 4.95 percent. Once you factor in home health care service the percentage of the elderly population needing nursing homes should be around 8 to 10 percent.

5. The Housing and Neighborhood Development Strategies (HANDS) program is a prime example of a public-private partnership that demonstrates how to create affordable, attractive and accessible housing. With the financing of developers and nonprofit organizations, ten different attractive, accessible and affordable housing units have been created. For example, three bedroom, one bath cottages sell for $51,000 (mortgage payment of around $350 a month in 1995). HANDS has shown that creating accessible, barrier-free housing adds an additional three percent to the cost of the house. So for example, a $50,000 cottage-style house would cost an additional $1,500 to design using universal design techniques. For more information on this program, please write to John I. Gilderbloom, HANDS, Urban Studies Institute, University of Louisville, Louisville, Kentucky 40292.

References

Aldrich, J., & Nelson, F. (1984). Linear probability, logit, and probit models. Sage university paper series on quantitative applications in the social sciences, number 07-045. Beverly Hills, CA: Sage Publications.

Annual Housing Survey. (1978a). National core file [machine-readable data file]. Conducted by the Bureau of the Census for the Department of Housing and Urban Development. Washington, D.C.: U.S. Bureau of the Census [producer and distributor].

Annual Housing Survey. (1978b). Disability supplement [machine-readable data file]. Conducted by the Bureau of the Census for the Department of Housing and Urban Development. Washington, D.C.: U.S. Bureau of the Census [producer and distributor].

Annual Housing Survey. (1978c). National core file technical documentation. Prepared by Data Access and Use Staff, Data user Services Division, Bureau of the Census. Washington, D.C.: U. S. Bureau of the Census.

Apgar, W. C. (1987). The leaky boat: A housing problem remains. In P. D. Salins (Ed.), Housing america's poor (pp.67-89). Chapel Hill, NC: University of North Carolina Press.

Bratt, R. G., Hartman, C. & Meyerson, A. (Eds.). (1986). Critical perspectives on housing. Philadelphia: Temple University Press.

Brody, S., Poulshock, R., & Magciocchi, C. (1978). The family caring unit: A major consideration in the long-term support system. The Gerontologist, 18, 556-561.

Demaris, A. (1992). Logit modeling. Sage university paper series on quantitative applications in the social sciences, number 07-086. Beverly Hills, CA: Sage Publications.

Doty, P. (1986). Family care of the elderly: The role of public policy. The Milbank Quarterly, 64, 556-561.

Gilderbloom, J., & Appelbaum, R. (1988). Rethinking rental housing. Philadelphia: Temple University Press.

Gilderbloom, J., Appelbaum, R., Dolny, M., & Dreier, P. (1992). Sham rent control research: A reply to Tucker. Journal of the American Planning Association, 58(2), 220-224.

Gilderbloom, J., & Rosentraub, M. (1990). Creating the accessible city: Proposals for providing housing and transportation for low income, elderly and disabled people. American Journal of Economics and Sociology, 49(3), 271-282.

Gilderbloom, J., Rosentraub, M., & Bullard, R. (1987). Designing, locating and financing housing and transportation services for low income, elderly and disabled persons. Houston: University of Houston, Center for Public Policy.

Goering, J. M. 1979. Housing in America: The characteristics and uses of the annual housing survey. Washington, D.C.: U.S. Department of Housing and Urban Development, Office of Policy Development and Research.

Greenberg, J., & Ginn, A. (1979). A multivariate analysis of the predictors of long-term care placement. Home Health Care Services Quarterly, 1, 75-99.

Hosmer, D., & Lemeshow, S. (1989). Applied logistic regression. New York: Wiley.

Joreskog, K. G. (1977). Structural analysis of covariance and correlation matrices. Psychometrika, 15, 349-367.

Katsura, H. M., Struyk, R. J., & Newman, S. J. (1989). Housing for the elderly in 2010: Projections and policy options. Washington, D.C.: Urban Institute Press.

Lewis-Beck, M. (1980). Applied regression: An introduction. Beverly Hills, CA: Sage Publications.

Manton, K. G., Corder, L, & Stallard, E. (1993). Changes in the use of personal assistance and special equipment from 1982 to 1989: Results from the 1982 and 1989 NLTCS. The Gerontologist 33(2), 168-176.

Newman, S. J. (1985). Housing and long-term care: The suitability of the elderly's housing to the provision of in-home services. The Gerontologist, 25, 35-40.

Newman, S. J., & Struyk, R. with Wright, P., & Rice, M. (1990). Overwhelming odds: Caregiving and the risk of institutionalization. Journal of Gerontology: Social Sciences, 45(5), S173BS183.

Nie, N. H., Hull, C. H., Jenkins, J. G., Steinbrenner, & Kearin, B. (1975). SPSS statistical package for the social sciences. Second Edition. New York: McGraw-Hill.

Norusis, M. J. (1982). SPSS introductory guide. New York: McGraw Hill.

Norusis, M. J. (1983). SPSSX: Introductory statistics guide. New York: McGraw Hill.

Palmore, E. (1976). Total chance of institutionalization among the aged. The Gerontologist, 16, 504-509.

Pynoos, J. (1992). Strategies for home modification and repair. Generations, 16(2), 21-25.

Pynoos, J., Cohen, E., Davis, L., & Bernhardt, S. (1987). Home modifications: Improvements that extend independence. In V. Regnier, & J. Pynoos, (Eds), Housing the aged: Design directives and policy considerations (pp. 277-304). New York: Elsevier.

Reschovsky, J., & Newman, S. (1990). Adaptations for independent living by older frail households. The Gerontologist, 30, 543-552.

Reschovsky, J., & Newman, S. (1991). Home upkeep and housing quality of older homeowners. Journal of Gerontology: Social Sciences, 46, S288-S297.

Smyer, M. (1980). The differential use of services by impaired elderly. Journal of Gerontology, 35, 249- 255.

Soldo, B., & Longino, C., Jr. (1988). Social and physical environments for the vulnerable aged. In National Academy Press, (Ed), America's aging: The social and built environment in an older society (pp. 103-133).  Washington, D.C.: National Academy Press.

Struyk, R. J. (1982). The demand for specially adopted housing by elderly-headed households. Washington, D.C.:The Urban Institute Press.

Struyk, R., & Katsura, H. (1987). Aging at home: How the elderly adjust their housing without moving. Journal of Housing for the Elderly, 4 (Whole No. 2).

U.S. Department of Housing and Urban Development. (1993). Codebook for the american housing survey data base: 1973-1993. Washington, D.C.: U.S. Department of Housing and Urban Development.

Wan, T., & Weissert, W. (1981). Social support networks, patient status and institutionalization. Research on Aging,3, 240-256.

Weissert, W., & Scanlon, W. (1983). Determinants of institutionalization of the aged. Washington, D.C.: The Urban Institute Press.