Back to Appendix B: Kentucky Retirement Survey: Background Information
and Sample CharacteristicsFrom Planning for the Future
p. 85-88, published 2002
The estimated percentages shown on page 13 are based on the effect of Social Security dependence on standard of living while holding constant a number of other socioeconomic and demographic factors. (See Table C.1 for a crosstabulation between standard of living and “dependence” on Social Security.) We used a cumulative logit model for ordinal responses to estimate the relationship between the dependent variable and several independent variables. The dependent variable is the respondent’s self-assessment of their current standard of living. The actual question is: Compared to the end of your working career, would you say your current standard of living is: (1) much better now; (2) better now; (3) about the same; (4) worse now; or (5) much worse now. We collapsed the dependent variable into three categories so that “worse” or “much worse” equals one, “about the same” equals two, and “better” or “much better” equals three. The independent variables in the model are listed below, along with parameter estimates in Table C.2 and a correlation matrix in Table C.3:
METROCO — This dichotomous variable is equal to 1 if the individual lives in an urban county and 0 for a rural county. We used Beale Codes to categorize the respondent’s county. If the county is designated as “0” through “3” then METROCO equals 1. Otherwise, if the Beale Code is equal to 4 through 9 then METROCO equals 0.
AFRDMED — The question is: Can you afford to pay for all your medical needs? This variable equals 0 for “no” and 1 for “yes.”
GENDER — This dichotomous variable is equal to 1 for males and 0 for females.
Q2INC, Q3INC, Q4INC, MISINC — These dichotomous variables reflect if the individual’s total household income from all sources before taxes is in the second, third, fourth quartile, or missing. The first quartile is left out of the model and is therefore the comparison group. The variable equals 1 if the individual’s income falls in the quartile (or is missing in the case of MISINC) and 0 if it does not. The variable Q1INC equals 1 if income is between “none” and $14,999. The variable Q2INC equals 1 if income is between $15,000 and $29,999. The variable Q3INC equals 1 if income is between $30,000 and $49,999. The variable Q4NC equals 1 if income is $50,000 or higher.
SOMEPSE — The survey respondents were asked to please circle the last grade in school you completed. SOMEPSE equals 1 if they circled “1 or 2 years college, no degree,” “graduated junior or community college,” or “vocational/technical degree.”
BAorMORE — The survey respondents were asked to please circle the last grade in school you completed. If they circled “bachelor’s degree,” “some graduate school work,” or “graduate degree (ex: MA, MS, PhD, JD)” then BAorMORE equals 1.
OVER65 — If the respondent is over 65 years old, then OVER65 equals 1 (otherwise equals 0).
SSMAJ — The respondents were asked which source of income is or will be their most important source of income in retirement. If they indicated “Social Security,” then SSMAJ equals 1. Otherwise, if they listed another source, like pensions, employment, or savings, then SSMAJ equals 0.
Table C1: Crosstabulation Between Standard of Living and Social Security Dependence
Table C2: Estimating Standard of Living, Parameter Estimates
Table C3: Correlation Matrix: Pearson Correlation Coefficients/Significance/Number of Observations
To view a list of all chapters in this book, click here. To read the chapters in sequential order, please follow the arrows below.
Back to Appendix B: Kentucky Retirement Survey: Background Information
and Sample Characteristics
