A PANEL STUDY OF INCOME DYNAMICS: PROCEDURES AND TAPE CODES (DOCUMENTATION) 1988 INTERVIEWING YEAR VOLUME I WAVE XXI A SUPPLEMENT Conducted with Grants from the National Science Foundation, the Office of the Assistant Secretary for Planning and Evaluation of the Department of Health and Human Services, the National Institute for Child Health and Development, and the Food and Nutrition Service of the Department of Agriculture. Survey Research Center INSTITUTE FOR SOCIAL RESEARCH THE UNIVERSITY OF MICHIGAN ANN ARBOR, MICHIGAN 1991 Morgan, James N. PANEL STUDY OF INCOME DYNAMICS 1968-1988 (WAVES I- XXI) [computer file]. Ann Arbor, MI: Survey Research Center [pro- ducer], 1991. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 1991. ii Preface This volume documents the twenty-first wave of data collected by the Panel Study of Income Dynamics, interviews taken in 1988 on income for 1987. Volumes I and II of A Panel Study of Income Dynamics: 1968-1972 Interviewing Years (Waves I-V) contain tape codes, indexes, available data, questionnaires and procedures specific to our first five years of data col- lection (1968-1972). These volumes also describe the early history of the study and some of the basic procedures that are common to all twenty-one years of interviewing. Sixteen supplemental volumes, including this one, cover procedures, codes and questionnaires for Waves VI-XXI. Ten volumes of findings entitled Five Thousand American Families--Patterns of Economic Progress are available, covering ten years of PSID findings from 1969 through 1978. Years of Poverty, Years of Plenty by Greg J. Duncan and col- leagues, based on PSID data, is also available. This book is an accessible summary of findings regarding poverty and employment dynamics through the late 1970s. All documentation for the PSID is available from the Inter-University Consortium for Political and Social Research, P.O. Box 1248, Ann Arbor, MI, 48106. The User Guide The User Guide to the Panel Study of Income Dynamics, a supplement to the PSID Procedures and Tape Codes volumes, is also distributed by the Inter-University Consortium for Political and Social Research. The Guide clarifies features of the study design and provides information needed to use the panel study data effectively. The guide is in loose-leaf format and purchasers are sent updates to add to their copies. Staff Greg J. Duncan, Martha S. Hill, and James N. Morgan are the study's principal researchers. James M. Lepkowski directs the survey operations. Charles Brown is in charge of labor market content. Tecla C. Loup oversees data collection and processing and compiles the documentation with the as- sistance of Anita Ernst. Bonnie Bittman supervises family composition editing, Thomas Gonzales supervises income editing, and Anne Sears super- vises the coding procedure. Data processing is divided into several parts: Charles Stallman deals with raw data files and consistency checks, Ron Amos generates variables for the final single-year files, Margaret Hoad proces- ses the family history data, and Marita Servais and Barbara Browne build the merged files. Kathryn Terrazas manages the field production. Joan Brinser and Priscilla Hildebrandt are responsible for general "care and feeding" of and payments to respondents. Deborah S. Laren and Naomi K. Sealand assist Greg Duncan with data analysis. Jean Yeung assists Martha Hill with data analysis and sponsor communication, and also keeps the bibliography of publications which use Panel Study data. Mary Wreford is an administrative manager. Peggy Gunnesch and Sarah Olson provide secretarial support. Users who wish to communicate with the study staff regarding questions about PSID data content should contact Tecla C. Loup at (313) 936-0316. iv TABLE OF CONTENTS Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii Section I: Procedures for the 1988 Interviewing Year . . . . . 1 Part 1: The 1988 Questionnaire, Data Processing, Interview- ing Procedures, Occupation Codes, Data Quality, In- dependent Part Samples, Weights . . . . . . . . . . 1 Part 2: 1988 Questionnaire . . . . . . . . . . . . . . . . . 3 Part 3: Editing Procedures and Worksheets . . . . . . . . . 9 Part 4: Coding Procedures . . . . . . . . . . . . . . . . . 11 Part 5: Generated Variables, Additional Data and Hot Topics 11 Part 6: The 1985 Ego-Alter, Demographic Event History and 1968-1985 Relationship Files . . . . . . . . . . . . 22 Part 7: The Work History Supplement File . . . . . . . . . . 25 Part 8: Data Available . . . . . . . . . . . . . . . . . . . 25 Part 9: Creating Family-Level Data from the Cross-Year Family-Individual Tape . . . . . . . . . . . . . . . 26 Part 10: PSID User Guide . . . . . . . . . . . . . . . . . . 27 Part 11: PSID CD-ROM . . . . . . . . . . . . . . . . . . . . 27 Section II: Tape Codes for Wave XXI . . . . . . . . . . . . . . 29 Part 1: Twenty-first-Year Family-Level Tape Code . . . . . . 29 Part 3: Index of the 1988 Employment Sections . . . . . . . 523 v vi 1 SECTION I PROCEDURES FOR THE 1988 INTERVIEWING YEAR Part 1: The 1988 Questionnaire, Data Processing, Interviewing Procedures, Occupation Codes, Data Quality, Independent Part Samples, Weights The 1988 Questionnaire The 1988 questionnaire included a major supplement about parental and other help given and received in terms of both time and money. We con- tinued to ask about marriages, divorces, adopted and natural children. Up- dating questions were again asked to account for new children and marital changes for those whose data were collected from 1985 through 1987; new Heads and Wives/"Wives"1 were asked about all of their children and first and last marriages. Employment event dating questions for 1988 were redesigned to ask about spells with employers instead of the position- oriented approach used from 1984 through 1987, and only information about prior-year employers was asked in detail. Food stamp eligibility, mortgage and utility payment questions were omitted for 1988. Data Processing We continue to use a direct data entry coding system that is fully compatible with the OSIRIS Statistical Software System. PSID tapes are released in OSIRIS format. OSIRIS interfaces with other systems (e.g., SPSS, SAS, BMDP), allowing easy access to other statistical and data management software. Interviewing Procedures Nearly all of the 1988 interviews were taken by SRC interviewers in the field by telephone. Interviews were taken with 7114 heads of families out of 7350 assigned, for an overall response rate of 96.8%. Subtracting from the base 61 respondents who had died since the last interview, had moved into institutions that precluded an interview, were too ill to be interviewed, or had rejoined sample ex-spouses raises the response rate to 97.6%. The interview total includes interviews with 252 splitoffs (out of a total of 289) with a response rate of 87.2 percent. For the reinterview panel only, again with the deceased and others removed from the base, the response rate was 98.0%. Slightly more than fifty percent of the persua- sion letters written to reluctant respondents resulted in interviews. The average length of the interview was 37.0 minutes (Table 1). The increase in interviewing time was due to the parent and help supplement. Respondents were each paid $12.50 for their interviews and an additional $5 per family for returning an address correction postcard in January 1988. ___________________________________________________________________________ 1The PSID uses the term "wife" (in quotes) in referring to long-term female cohabitors. 2 Occupation Codes We continue to use the 1970 Census three-digit occupation and industry codes for the current main jobs of employed Heads and Wives/"Wives." They are also used for the most recent jobs held by Heads and Wives/"Wives" who are not currently working, and in coding the employment histories and extra or second job questions. For comparability with past data, one-digit oc- cupation codes are used to code Head's first job and Head's father's oc- cupation, since these data items were collected only for new Heads in 1988. Data Quality Ninety-two percent of the 1988 interviews were taken by telephone (Table 3). The remaining eight percent of respondents have no telephones, prefer personal interviews due to party lines or hearing difficulties, or live out of range of our interviewers and complete their own question- naires. The rate at which Heads responded for themselves declined to 76.9%; Wives/"Wives" accounted for almost all of the proxy respondents. There is very little year-to-year variation in the number of data im- putations (Table 5); the quality of the data, according to this indicator, continues to be good. Table 2b shows response rates based on original sample individuals, annually and cumulatively. Since it would be impossible to know how many individuals were eligible but nonresponding in 1968, we used the 1968 sample as the base for further calculations. Table 2b also includes columns that remove the deceased from the base. Individuals born into the sample are not included in this table. Independent Part Samples The use of part samples is suggested for separating the selection of a preferred model from the assessment of its stability and power. Simple random subsamples are not independent of the rest of the sample because of the clustered nature of area probability samples. Four independent quarter-samples have been selected for users and are designated in the variable V16178. How much of the sample should be reserved for statistical testing depends on how unsure one is about the best model and how important the estimation and testing of one optimal model is felt to be. For il- lustrations of the results of this separation of the searching from the as- sessing procedures, see the volumes of findings, Five Thousand American Families--Patterns of Economic Progress, Volume I, pp. 6-8 and pp. 342-344; Volume II, Chapter 9; and Volume IV, Chapter 2 (Survey Research Center, Ann Arbor, Michigan). There are also several variables for use in defining paired sampling error computing units within half-sample strata for repeated replication to compute sampling errors (V16179-V16182). These variables have also been added at the individual level (V32050-V32053). See Chapter 17 of Vol. IX of Five Thousand American Families, and Section I, Part 5 in this volume. 3 Table 1 AVERAGE LENGTH OF INTERVIEW __________________________________ ---------------------------------- Number of Average Length Year Interviews in Minutes ---------------------------------- 1968 4802 63.1 1968 4460 61.8 1970 4645 60.5 1971 4840 59.1 1972 5060 66.2 1973 5285 20.1 1974 5517 23.1 1975 5725 26.9 1976 5862 48.2* 1977 6007 25.0 1978 6154 26.9 1979 6373 28.1 1980 6533 29.0 1981 6620 26.5 1982 6742 20.8 1983 6852 23.8 1984 6918 34.7 1985 7032 49.9* 1986 7018 34.9 1987 7061 29.5 1988 7114 37.0 ---------------------------------- *Includes both Head's and Wife's interviews. Weights The sample was entirely reweighted in 1984 for each year of data. This is a population weight for reducing bias in estimates, not a variance weight for efficiency. See Section I, Part 5, pp. 66-71 of the wave XVII documentation or the PSID User Guide, Chapter E for a discussion of reweighting theory and techniques. Part 2: 1988 Questionnaire The 1988 questionnaire follows, with cross-year family-level variable numbers added at the appropriate questions for both edited and directly coded items. The 1988 questionnaire with variable numbers from the merged family tape was included in the original published documentation. It is not in- cluded in this machine readable version. You may obtain a paper version of the "1988 Questionnaire with Vari- able Numbers" by sending a request by e-mail to "psid.staff@umich.edu" or 4 Table 2a FAMILY ANNUAL AND CUMULATIVE PANEL RESPONSE RATES** __________________________ -------------------------- Percent Year ------------------- Annual Cumulative -------------------------- 1968 76 76 1969 89 68 1970 97 66 1971 97 64 1972 97 62 1973 97 61 1974 97 59 1975 97 57 1976 96 55 1977 97 53 1978 97 51 1979 97 49 1980 97 48 1981 97 47 1982 97 46 1983 97 45 1984 97 44 1985 96 42 1986 96 40 1987 96 38 1988 97 37 -------------------------- **The deceased, those too ill to be interviewed, and recombined families have not been removed from the base. by US mail to Jean Yeung, 3263 ISR, University of Michigan, PO Box 1248, Ann Arbor, MI 48106-1248. 5 Table 2b INDIVIDUAL ANNUAL AND CUMULATIVE PANEL RESPONSE RATES ___________________________________________________________ ----------------------------------------------------------- Annual, Cumulative, Annual, Cumulative, Deceased Deceased Deceased Deceased Sample Included Included Removed Removed Year Size in Base in Base from Base from Base ----------------------------------------------------------- 1968 18224 100.0 100.0 100.0 100.0 1969 16046 88.0 88.0 88.5 88.5 1970 15476 96.4 84.9 96.9 85.7 1971 15108 97.6 82.9 98.2 84.1 1972 14713 97.4 80.7 98.1 82.4 1973 14295 97.2 78.4 97.9 80.6 1974 13908 97.3 76.3 97.9 78.8 1975 13548 97.4 74.3 98.1 77.2 1976 13096 96.7 71.9 97.2 75.0 1977 12706 97.0 69.7 97.7 73.1 1978 12417 97.7 68.1 98.2 71.7 1979 12056 97.1 66.2 97.7 70.0 1980 11683 96.9 64.1 97.6 68.1 1981 11382 97.4 62.5 98.1 66.7 1982 11125 97.7 61.0 98.5 65.5 1983 10828 97.3 59.4 98.1 64.1 1984 10515 97.1 57.7 98.0 62.6 1985 10183 96.8 55.9 97.7 60.1 1986 9826 96.5 53.9 97.4 59.1 1987 9504 96.7 52.2 97.9 57.6 1988 9225 97.1 50.6 98.0 56.1 ----------------------------------------------------------- 6 Table 3 PROPORTION OF INTERVIEWS BY TELEPHONE ____________________________________________ -------------------------------------------- Number of Unweighted Year Sample Size Telephone Percent of Interviews Sample -------------------------------------------- 1968 4,802 -- -- 1969 4,460 -- -- 1970 4,645 67 1.4 1971 4,840 108 2.2 1972 5,060 134 2.6 1973 5,285 4,047 76.6 1974 5,517 4,554 82.5 1975 5,725 4,836 84.5 1976 5,862 5,360 91.4 1977 6,007 5,040 83.9 1978 6,154 5,283 85.8 1979 6,373 5,635 88.4 1980 6,533 5,829 89.2 1981 6,620 6,081 91.9 1982 6,742 6,257 92.8 1983 6,852 6,401 93.4 1984 6,918 6,369 92.1 1985 7,032 6,423 90.6 1986 7,018 6,454 92.0 1987 7,061 6,479 91.8 1988 7,114 6,520 91.5 -------------------------------------------- 7 Table 4 PROPORTION OF FAMILY HEADS INTERVIEWED __________________________________ ---------------------------------- Proportion of Year Sample Size Interviews by Head ---------------------------------- 1968 4,802 92.6 1969 4,460 93.1 1970 4,645 93.2 1971 4,840 93.3 1972 5,060 93.5 1973 5,285 91.1 1974 5,517 90.0 1975 5,725 88.3 1976 5,862 92.6 1977 6,007 90.0 1978 6,154 90.2 1979 6,373 88.5 1980 6,533 85.8 1981 6,620 84.3 1982 6,742 83.8 1983 6,852 82.2 1984 6,918 81.0 1985 7,032 87.1 1986 7,018 81.5 1987 7,061 79.0 1988 7,114 76.9 ---------------------------------- 8 Table 5* SUM OF ACCURACY CODES FOR THREE TAXABLE INCOME ITEMS FOR HEAD AND WIFE _________________________________________________________________________ ------------------------------------------------------------------------- Year of Data 0 1 2 3 4 or More Total ------------------------------------------------------------------------- 1968 94.0 2.5 2.6 0.2 0.8 100.0 1969 95.6 1.6 1.9 0.1 0.8 100.0 1970 96.9 1.3 1.3 0.1 0.5 100.0 1971 97.7 0.9 0.9 0.1 0.4 100.0 1972 97.8 0.8 1.1 0.0 0.3 100.0 1973 97.9 1.1 0.7 0.1 0.2 100.0 1974 98.2 0.9 0.7 0.0 0.2 100.0 1975 98.3 0.8 0.8 0.0 0.2 100.0 1976 97.0 1.2 1.6 0.1 0.2 100.0 1977 97.4 1.1 1.2 0.0 0.3 100.0 1978 97.4 0.7 1.3 0.1 0.5 100.0 1979 96.1 0.8 2.3 0.1 0.7 100.0 1980 95.8 0.8 2.4 0.2 0.8 100.0 1981 95.6 1.2 2.5 0.2 0.4 100.0 1982 95.3 1.1 2.7 0.1 0.8 100.0 1983 94.5 1.6 2.9 0.2 0.8 100.0 1984 94.3 2.0 2.7 0.2 0.8 100.0 1985 94.2 2.9 2.3 0.2 0.4 100.0 1986 94.7 1.4 3.0 0.1 0.7 100.0 1987 94.6 1.5 3.0 0.1 0.8 100.0 1988 95.1 1.0 2.8 0.1 1.0 100.0 ------------------------------------------------------------------------- *Table 5 is based on four variables: Accuracy of Head's Labor Income (1988: V14914 + V14919) Accuracy of Wife's Labor Income (1988: V14921) Accuracy of Asset Income of Head and Wife (1988: V14934) Overall accuracy is indicated by the number of assignments made by the editors in order to impute data missing from an interview. The more as- signments, the less accurate the data. The accuracy code values and their meanings are: 0. Adequate response: No assignments made. 1. Minor assignment: Response was inadequate, but estimates could be made within a probable error of under $300 or 10 percent of the as- signment by using previous years' data or other data in the inter- view. 2. Major assignment: Response was inadequate, and estimates had a probable error of at least $300 and at least 10 percent of the value of the assignment, using any information available in previous interviews or in the current one. Usually these values were imputed from an assignment table. This table shows the sum of the accuracy codes for the three different income measures. The maximum value possible here would be eight for mar- ried couples, six for single heads. Part 3: Editing Procedures and Worksheets The PSID editing process serves three main purposes: (1) accounting for all year-to-year changes in family membership, (2) rectifying dis- crepancies within the interview before coding, and (3) calculating and recording numeric data on the worksheets and interviews for coding. It is a complicated task requiring a high degree of accuracy; each interview is checked by another editor. Family composition editing and occupation coding comprise the first step. Next, an extensive edit of income and work is done. Discrepancies that require additional respondent contact to correct are noted, and each interview with such problems is sent back to the field for additional data collection. Interviews passing through this process are considered "clean" for coding. The full array of past interviews is available to editors, though only the prior year's interview is usually consulted to solve problems. Prior data are used when the current interview is vague, contradictory, or incom- plete despite attempts to clarify the family's situation. Project staff closely oversee the editing process and make substantive decisions regard- ing the handling of specific problems. Most of the techniques used to edit previous waves were again used to edit Wave XXI. However, the addition of employment history questions from 1984 through 1988 provided a new dimension in the editing of work weeks. Specifically, the work hours and employment histories were cross-checked for inconsistencies, and interviews were returned to the field for the resolution of discrepancies. Information on annual work time is probably slightly more accurate than in the past. Questions were added beginning in 1984 to distinguish time unemployed and looking for work from time out of the labor force, so that measures of unemployment hours are cleaner. Detailed discussion of techniques for editing other variables will be found in the Waves I-V (1968-1972) documen- tation, Volume I, pp. 270-339. Specific changes since that time have been included annually in Section I, Part 3 of the succeeding documentation volumes. Family Composition Editing All people in a panel family at the time of the previous year's inter- view must be accounted for in the current year. They may remain in the family or may have moved out, died, or entered institutions. Sample mem- bers 18 or older who move out and form their own households are followed and interviewed as new panel families (i.e., splitoffs). In addition to the usual tasks for family composition editing, the type of institution for those families in the armed forces, educational or health facilities, and other such institutions (V14823) has been coded since 1985. The more detailed relationship to Head and birthdates have been coded since 1983. The marital and childbirth histories collected since 1985 have placed more demands on the task of family composition editing through the addition of unique individual identifier for each spouse or child mentioned. For 1988, 9 10 individual identifiers were assigned and coded for parents, helpers, and those helped. Wave XXI Changes The extensive edit of income and work remained similar to 1984-1987 procedures. We returned from the 1987 procedure for calendar year 1986 to collection of an employment history that included all of calendar year 1987. We did not ask much history from January of the current year up un- til the time of the interview, under the assumption that those data are in- cluded in the 1989 wave. Assignment Tables Again in Wave XXI we did not use inflation factors for our assignment tables, as we had in 1985 and earlier years. The tables were simply created using the unweighted data from the last year (Wave XX). Proration Variables for Income Variables detailing adjustments to total income for family members who joined or left the family, begun for 1986, continued to be coded in 1988. See Section I, Part 3, page 68 in the 1986 documentation for details. The 1988 edit worksheets with variable numbers from the merged family tape was included in the original published documentation. It is not in- cluded in this machine readable version. You may obtain a paper version of the "1988 Edit Worksheets with Vari- able Numbers" by sending a request by e-mail to "psid.staff@umich.edu" or by US mail to Jean Yeung, 3263 ISR, University of Michigan, PO Box 1248, Ann Arbor, MI 48106-1248. 11 Part 4: Coding Procedures Production coding the questionnaire is the final step in putting the data onto computer tape. This occurs after the questionnaire has been edited as described in Part 3 above. The coding process converts numeric and non-numeric answers into machine readable data. Coders entered the edited variables and coded the questionnaire using the system designed by the Computer Support Group at ISR's Survey Research Center for direct data entry. The system, Interactive System for Input of Survey Data (ISISD), is compatible with OSIRIS System and has been used by the PSID for nine years. It incorporates wild code and data consistency checking into the coding process. These checks insure that coders cannot enter invalid code values, and thus the necessity of later data cleaning by the Panel Study staff is greatly reduced. Approximately 10 percent of the interviews (720) were coded twice, once by the coder and a second time by a Panel Study staff member or a check coder. Check coding consists of an item-by-item check of all data values for a case which have been independently coded by a second person. It enables the study staff to determine early in the processing whether any individual coder is having difficulty and whether some codes are causing problems. Coders are trained by a member of the study staff before they are al- lowed to production-code interviews. Training begins with a short intro- duction on the history and purpose of the study and a question-and-answer session. The coders code two practice interviews which have been coded previously by the study staff member. The coders and the study staff mem- ber review the practice interviews in detail, discussing any coding dif- ferences with particular emphasis on problems that could arise during production coding and responses that may present coding difficulty. A difference is a disagreement between the coder and the check coder. Differences become errors when so judged by the check coder. For example, a coder may use an erroneous code value, enter a wrong digit on the ter- minal keyboard, or miss a specific direction in the code book. Some dis- agreements are not errors. They involve coding of open-ended questions; decisions on the final codes chosen rest with the study staff member. Coding Differences for Wave XXI Coding reliability rates were again excellent for 1988. The overall difference rate was 1.44 per case. The error rate was .38 per interview. The questions for why the Head moved (1988: V15150) and why the Head might move in the next few years (1988: V15153) are our most consistently problematic for coding reliability, although the rates are improving. Table 6 shows the reliability rates on these two data items for 1985-1988. Part 5: Generated Variables, Additional Data and Hot Topics Various indices, bracket variables, and complex measures of economic status have been constructed each year using variables derived directly from coded interview data. Inter-year changes in the interview schedule 12 Table 6 RATES OF CODING DIFFERENCE _______________________________________ --------------------------------------- 1985 1986 1987 1988 --------------------------------------- Why moved 7.3% 7.8% 5.5% 5.7% Why might move 11.5 9.7 8.0 6.8 --------------------------------------- have made addition and deletion of indices necessary. In general, if an index could not be built to be exactly comparable to a previous index, it was not constructed. Income Several measures of economic status have been generated for all twenty years, including money income variables and measures of income adequacy. Family Money Income, one of the simplest indices, is the total of all mem- bers' earnings, transfers, and asset income from the prior calendar year (1989: V16144). Bracket (Interval) Variables Until Wave X (1977), several numerical variables, such as family money income, had been given also as bracket (interval) codes. Such interval codes had been constructed for most of the measures where a distribution was useful and appropriate. This includes practically all of the income variables and their components. For Waves X-XXI we have provided two pieces of information in the family-level tape code which allow users to bracket as their own uses dictate: (1) weighted percent of nonzero cases, and (2) weighted mean value of nonzero cases. This information is provided for any variable bracketed in 1976, as well as for almost all other field amounts. Regional Data Measures In addition to personality and behavior, location and environmental factors are potentially important determinants of an individual's economic status. Consequently, the interview data have been supplemented with in- formation on some employment and income characteristics of the county in which the panel family resides (V16183-V16186). Questionnaires are sent each year to state employment offices asking about current labor market conditions in these counties. A question was added to the 1988 question- naire for minimum wage in the state of residence (V16186). Sampling Error Computation Unit (SECU) Variables The 1988 data tape includes variables that may be used for computation of variances under the stratified multistage design (V16181-V16182, V32051- V32052), as well as variables for balanced half-sample replication (V16179- 13 V16180, V32049-V32050). Please see the 1983 Documentation Volume, pp. 89- 90, for details. Families and Households: Householder The PSID concept of family has been described in the User Guide. Briefly, we began the study with a definition of a family that was similar to that used by the Census Bureau--the group of individuals sharing a household who are related by blood, marriage or adoption. Like the Census Bureau, we have never treated lodgers, conventional roommates, or live-in employees as members of our families, but we do regard Census "unrelated individuals" as single-person families. We have also followed the Census concept of Head of Household, or householder. Our following rules dictate that we not only continue to interview the Head in succeeding waves, but also attempt to interview family members who leave to form their own households. A corollary is that returning family members who have been successfully followed are not reintegrated into the family (with the notable exception of recombined married couples.) The result is that we have diverged from the Census definition of family in that several related individuals may share the household but are treated by us as separate families, each with its own family unit Head. The picture is complicated further by our rule that a valid PSID interview must contain an original sample member (or offspring) as the family Head or Wife/"Wife." In order to minimize the difficulties that our sample design require- ments cause for comparisons with Census Population Survey data, we code in- formation about the household member who would qualify as household Head according to Census rules, regardless of whether that individual is a mem- ber of our family unit or even a member of the panel at all. Beginning with the 1985 wave, we have included five variables describing the householder (1988: V14817-V14821.) These variables list his or her individual-level identifiers (1968 ID Number and Person Number), age, sex, and relationship to the Head of our family unit. For more information on multiple families within one household, see Linking Data: Families Sharing Households later in Part 5. New Heads and New Wives/"Wives" Two variables have been generated beginning in 1983 that indicate the year in which the current Head most recently became Head (1988: V16190) and the year in which the current Wife/"Wife" most recently became Wife/"Wife" (1987: V16191). (It is possible that an individual becomes Head or Wife/ "Wife" more than once in the course of the panel due to marital breakups, reconciliations and remarriages.) These variables contain as code values the last two digits of the year in which the background data for Heads or Wives/"Wives" was most recently asked. In 1985, most background informa- tion was reasked. All Wives/"Wives" answered these data items afresh for the 1985 interview. Therefore, V16191 equals 85 for most cases. New Wives/"Wives" since then were asked the entire sequence and thus have values of 86, 87 or 88 for this variable. New Heads were, as usual, asked the entire sequence (V16055-V16127). For 1985 Heads who were also Heads in 1984, however, only 1985 variables 14 V11924-V11981 were asked. The values for 1985 in variables V11907-V11923 were simply transferred from previous years' data, as most of these items should not change from year to year. Variable 16190 indicates the year in which these items were most recently collected. Refer also to p. 72 of the Wave XX (1987) documentation for information regarding specific background variables. Education of Head and Wife/"Wife" at the Individual Level The income and work sequence asked about all of last year's family unit members other than the current Head and Wife/"Wife" includes some questions about completed years of schooling. These questions are reasked and coded each year for such individuals (1988: V300602-V30605.) Current Heads and Wives/"Wives" have values of zero for these variables. We have not included equivalent data for them in these tape locations because their years of schooling and much more education detail is available at the fami- ly level (Head: V16089-V16124, Wife/"Wife": V16024-V16050) and because their education and other background items are not reasked each year. For those users who want it, we have provided information in the individual- level variable descriptions for each year's completed years of education about how to generate comparable data for current Heads and Wives/"Wives" from the family-level data. Family Composition and Sample Member Data The variable Couple Status of Head (V16189) allows users to easily identify cases with female Heads and husbands at the family level. Ad- ditionally, through the coding of relationship to Head in more detail, long-term female cohabitors ("wives") are distinguishable from legally mar- ried couples at the family level. The variable Head and Spouse Sample Status (V16192) allows the user to know how the family-level weight (V16208) has been calculated for each case. The family-level weight is defined as the average of the Head's and the Wife's/"Wife's" individual weights; nonsample individuals have weights of zero. Weights for 1968-1988 In 1968, the original set of family weights was created to reflect differential sampling fractions, overlap of the Census and ISR samples, and initial response rates by area. At the beginning of the study, each in- dividual's chance of being in the sample was the same as that of his or her family's and so individual and family weights were identical in 1968.1 ___________________________________________________________________________ 1In later years, individual-level and family-level weights diverged as families regrouped through marriages of sample and nonsample in- dividuals. The family-level weights are generated from the individual- level weights; the family weight equals the mean of the head's and spouse/ cohabitor's individual weights. Thus, in a family with a single head (no spouse/cohabitor present), the family weight is identical with that head's individual weight. For couples, there are a few possible situations: a) Head and spouse/cohabitor are both sample members: their individual weights are identical, and therefore the family weight is also identi- cal. 15 Since that time, individual-level weights were periodically adjusted for subsequent nonresponse in 1972 and 1978, and, at the family level, the weights were additionally adjusted each year for marriages to and divorces from nonsample spouses. Thus, individual-level weights were carried for- ward intact in years when no new nonresponse adjustments were made, al- though, as stated above, the family-level weights were modified when neces- sary. No adjustment was made for mortality, however, and as a result our weights, particularly for older people, were becoming increasingly too large. In 1984, using the complete file of all individuals ever connected with the Panel Study of Income Dynamics, we produced mortality-corrected individual-level and family-level weights for each year of data (1968- 1984).2 The revisions performed in 1984 involved adjusting the family-level and individual-level weights in 1969, 1974, 1979, and 1984 to take account of mortality. The adjustments used individual-level weights and were based on expected mortality,3 as actual mortality is not currently known with certainty. Since annual mortality for most segments of the population has been quite modest, five-year adjustments were deemed sufficient. The five- year periodicity for updating the weight generally seemed adequate for non- response adjustments as well. Annual response rates were quite high (96 or 97 percent) in all years except 1968 and 1969. See Section I, Part 1, Tables 2a and 2b for actual year-by-year response rates. The lower response rates in 1969 prompted us to estimate a separate weight adjustment for 1969, and the high response rates of subsequent years allowed us to follow a pattern of five-year readjustments for the years following 1969. Thus, for 1988, the simple updating occurred. New mortality adjustments will again be made in 1989. ___________________________________________________________________________ b) Head and spouse/cohabitor are both sample members, but their weights are not identical: the family weight is the average of the two individual weights. c) Head and spouse/cohabitor are not both sample: one of the two has moved into a sample household and the family weight is therefore equal to half that of the sample individual. We have mistakenly, on rare occasions during the earlier years of the study, interviewed families where neither head nor spouse/cohabitor is sample. In such a case, the remainder of the family listing is scanned for sample members, and the family weight becomes the average of all such per- sons' weights. If no sample members are in the family, then the family weight is zero. 2The 1969 family-level weight was heretofore nonexistent. A variable was added for this (V1014), as there were a few unused tape locations be- tween the 1969 and 1970 family-level data. The tape locations for this new variable are 1037-1038. 3Expected mortality rates were derived from the Life Tables published in Vital Statistics of the United States (USGPO: 1969, 1974, 1979 and 1984; Table 5-1). 16 For more information on the complete reweighting in 1984, see the Wave XVII (1984) Documentation Volume, pp. 66-71. Linking Data: Splitoffs From the 1981 wave onward, data have been provided to assist the user in linking splitoff records with those of their main families. The family- level data for each main family in 1988 (V14807=0) contain values for V16193 representing the actual number of successfully interviewed 1988 splitoff families from each main family. Thus, splitoff nonresponse cases are not included. On each splitoff data record (V14807=1), the family por- tion of the record contains the current year's interview number (V14802) of the associated main family at V16194. The individual-level record of each member of a splitoff family also contains this interview number (V30622), as well as month and year the splitoff family was formed (V30620 and V30621). Month and year the splitoff family was formed are derived from actual move-out dates of splitoff individuals as reported on the main family coversheets. Thus, in the relatively rare event that two or more in- dividuals move at different times from the main family to form one splitoff family, each individual receives his or her actual date of move as code values for V30620 and V30621. Any other splitoff individuals who did not move out of a main family but simply appeared for the first time in the splitoff family, such as nonsample spouses, friends, miscellaneous rela- tives and newborn children, receive the move-out date given for the splitoff mover-out. When more than one splitoff mover-out date exists (a rare occurrence), these new persons receive the earlier date. For those individuals who move from institutions to form their own splitoff families, code values of 98 are inserted into the tape locations for both month and year. The other miscellaneous splitoff family members appearing for the first time in the study receive missing data code values of 99 for these month and year variables. Linking Data: Families Sharing Households It is not uncommon for two or more family units to share living quarters. Panel families involved in such a situation may live with per- sons or families who are not sample members and who are not included by the study as family members because the arrangement is supposedly temporary. The situation resembles that of roommates, where expenses are split between the individuals involved. However, the members of one panel family may also move in with the members of another panel family. Financial disasters such as divorce and unemployment contribute heavily to such patterns of be- havior. Most frequently, a former splitoff child already being interviewed separately returns home to live with panel parents for a period of time un- til resuming life on his or her own. Occasionally, siblings who are each being interviewed move in together to share an apartment, or aging panel parents go to live with their panel children. We continue to interview each of these smaller groups separately, as if they were living apart. To allow users the opportunity of utilizing the data set for such situations, several link variables have been created since 1982. For 1986 17 through 1988, a new variable, Current Household Composition (1988: V14816), was coded to flag cases where this sharing occurred. In 1982-1985, the sharing was indicated by Current Family Composition (1985: V11117). Be- cause some information on family composition was lost by doing this, we decided to code family and household compositions separately. The household code for 1988 differentiates between both panel and non- panel primary and secondary families living in the same households. Codes 4, 6, and 8 indicate that such an arrangement exists between panel families. As many as five panel families lived under the same roof at the time of the 1988 interview, and thus there are four sets of linking vari- ables so that all panel families living together are linked in every pos- sible direction. V16195, V16198, V16201 and V16204 contain the values for 1988 interview number or ID number (V14802) of each of these other families. In addition, variables describing kinship ties between these families (V16196, V16199, V16202, and V16205) and the family size (V14889) of each of the other families was included with these linking variables (V16197, V16200, V16203, and V16206). Another variable was added beginning in 1985--Household ID Number (1988: V16207). This variable has as its value the 1988 ID Number (V14802) for all data records where no sharing with other panel families occurred. If sharing with other panel families did occur, then Household ID Number equals the value for the 1988 ID Number of the family with the lowest value among each group of families. Because the identification of such household sharing for waves II-XIV (1969-1981) is difficult, similar variables were retroactively created in 1986 for each of these waves (1969: V1015; 1970: V1766; 1971: V2345; 1972: V2979; 1973: V3310; 1974: V3730; 1975: V4231; 1976: V5113; 1977: V5681; 1978: V6220; 1979: V6814; 1980: V7456; 1981: V8110). Taxes This year and each year since the 1980 wave, taxes of Head, Wife/ "Wife" and other earners have been generated by computer. In previous years they were constructed during the editing process. The 1985 documentation summarizes in detail the procedures used for calculation of taxes in that wave (1984 tax year) on pages 91-100, the 1986 documentation notes changes for 1985 tax year on pages 81-83, and the 1987 documentation lists changes for 1987 (1986 tax year) on pp. 76-78. We men- tion here only changes for 1988 (1987 tax year). 1) Adjusted Gross Income The procedure for calculation of adjusted gross income (AGI) is similar to last year's, but there are two minor changes. The taxable por- tion of Head's (V14954) and Wife's/"Wife's" (V14975) retirement income ex- clusive of Social Security and Veterans Administration pensions is set at 72 percent of its total.4 Unemployment benefits are now taxable, so ___________________________________________________________________________ 4This was the ratio, over all income classes, of pension and annuity income in AGI to total pension and annuity income, using the data from a 18 Head's and Wife's/"Wife's" unemployment compensation (V14956 and V14976) have been added to AGI for 1988. 2) Itemized Deductions From the 1984 interviewing year through the present, Heads were asked whether they itemized on their federal tax returns (1988: V15772). For those answering yes, we estimate itemized deductions as equal to a given percentage of their AGI, with the percentage varying by the size of AGI. The percentages used this year follow below. __________________________________________________________________________ -------------------------------------------------------------------------- Itemized Deductions AGI as a Percent of AGI -------------------------------------------------------------------------- less than $15,000 52 $15,000-$19,999 35 $20,000-$24,999 28 $25,000-$29,999 28 $30,000-$34,999 22 $35,000-$39,999 22 $40,000-$44,999 22 $45,000-$49,999 22 $50,000-$54,999 20 $55,000-$59,999 20 $60,000-$74,999 20 $75,000-$99,999 20 $100,000-$199,999 16 $200,000-$499,999 16 $500,000-$999,999 16 $1,000,000 or more 16 -------------------------------------------------------------------------- These percentages were calculated as the aggregate amounts, by AGI class, of itemized deductions divided by AGI, and are taken from Statistics of In- come - 1986, Table 2.1. These are updated from those used for the 1986 wave, but the data are from 1985 tax year because the new version of Statistics of Income was not yet in print. The percentage used for those with AGI of less than $15,000 was calculated using only the aggregates for the $10,000-$14,999 group in Table 2.1, as too many cases of itemizers with AGI below $10,000 were suspect. The IRS no longer refers to zero bracket amounts in its tax tables, so we now subtract the larger of itemized deductions or the standard deduc- tions. Standard deduction amounts for those under 65 and not blind are: $2,540 for single persons and heads of households and $3,760 for married couples filing jointly. ___________________________________________________________________________ preliminary draft of Table 1.4 of Internal Revenue Service, Statistics of Income - 1989, Individual Income Tax Returns, Washington, D.C.: U.S. Government Printing Office, 1989. 19 For the blind and those 65 or older, the basic standard deduction is $3,000 if single and $4,400 if a head of household. An extra amount of $750 is added to this for each status; thus, a single person age 65 or older has a standard deduction of $3,750 ($3,000 basic deduction + $750 ex- tra amount). If he or she is also blind, the deduction increases by another $750 to $4,500. Married couples filing jointly can claim a basic standard deduction of $5,000 if either spouse is age 65 or older or blind. The extra amount is $600 for each status for either person. The maximum extra deduction is $2,400 for a total standard deduction of $7,400. The standard deduction for a person who is claimed as a dependent by someone else is limited to $500 or the person's earned income, whichever is greater, but no more than the standard deduction allowed for the depend- ent's filing status. The probabilities of itemizing, however, which can be expected to vary significantly not only with income but also for homeowner, were generated from the 1988 family-level data. The probability distributions for itemization by AGI class are as follows: ________________________________________________________________________ ------------------------------------------------------------------------ Homeowners Homeowners with without AGI Renters Mortgages Mortgages ------------------------------------------------------------------------ less than $1 .7% 2.9% 3.7% $1-$4,999 4.5 16.0 11.6 $5,000-$9,999 7.6 41.4 35.1 $10,000-$14,999 9.0 39.7 32.2 $15,000-$19,999 11.2 54.5 36.7 $20,000-$24,999 23.8 57.9 43.3 $25,000-$29,999 25.0 76.1 59.2 $30,000-$34,999 38.5 80.4 48.0 $35,000-$39,999 40.0 83.9 63.0 $40,000-$44,999 40.7 88.4 67.6 $45,000-$49,999 45.1 88.1 72.7 $50,000-$54,999 57.9 93.0 63.2 $55,000-$59,999 69.2 88.1 69.6 $60,000-$74,999 65.5 93.8 90.0 $75,000-$99,999 72.7 93.6 100.0 $100,000-$199,999 100.0 95.2 88.5 $200,000-$499,999 100.0 92.3 100.0 $500,000-$999,999 100.0 92.3 100.0 $1,000,000 or more 100.0 100.0 100.0 ------------------------------------------------------------------------ The percent itemizers was calculated as the number answering yes to V14503 divided by the sum of the number answering either yes or no. The denominator excluded those not answering or who didn't know whether they had itemized. The percent for those with AGI of less than $5,000 was based only on those whose AGI was also above zero. 20 3) Dependents and Exemptions The allowance per exemption was increased to $1,900 for 1988 (1987 tax year). The IRS changed its exemption rules for the 1987 tax year to eliminate the double-counting of dependents as exemptions for both the dependent's claimers and the dependent's own tax forms. Hence, the number of exemptions can now equal zero. 4) Filing Status No changes were made to this procedure. 5) Marginal Tax Rates and Tax Payments Before Credits If other family members were present in the family for only part of the year, their incomes represent only that portion for the time they were present. In the 1985 wave, their marginal tax rates were assigned and tax payments calculated by our program using the part-year incomes and, because of the progressivity of the income tax, were somewhat distorted. For 1986 through 1988, we incorporated an adjustment to tax liability by applying the percent proration variables (1988: V15060, V15065, V15070, V15075, V15080) to the PSID "Taxable Income" variables. Otherwise, our procedure is identical with that used in 1985. 6) Tax Credits Earned income credit. This credit increased over last year's. It is now equal to 14% of the first $6,080 of earned income, less 10 percent of income over $6,920. This credit then cannot exceed $851 and falls to zero at an earned income or AGI of $15,432. Tax credit for the elderly. This credit remained the same as last year's. Institutionalization To facilitate analysis regarding family units living in institutions, a variable (1988: V14823) was coded for 1985 through 1988. This variable indicates the type of institution in which the family resides. Thus, the analyst will no longer have to employ oblique and inaccurate methods to isolate such cases. Families who are not in institutions but who have some member in educational facilities, the Armed Forces, prison, or health care facilities might be incurring some financial responsibility for such members. Thus, four variables (V16174-V16177) count the number of members in each of the above-mentioned types of institutions. A very few families may have some members in other types of institutions, such as religious houses. No counting variable was generated at the family level because the number of such cases is negligible. 21 FIPS State and County Codes and Beale's Urbanicity Code Population density of the area in which the family lives is a very im- portant item. The urbanicity code (1988: V16155) devised by Calvin Beale and Peggy Ross of the USDA has been added to 1985 through 1988 data. Also, the FIPS system of coding state and county (1988: V16153-V16154) as used by Beale to assign urbanicity was added to the data. We retain our usual state and county codes with 1988 (V14803-V14805). Appendix 1, pages 701- 721 of A Panel Study of Income Dynamics, Procedures and Tape Codes, 1985 Interviewing Year, Vol. I, lists the FIPS codes and the ways in which they differ from the PSID's codes. Marriage and Fertility Histories--Family-Level Variables No marriage history variables are included at the family level for 1988. Marriage and Fertility Histories--Individual-Level Summary Variables The individual-level marriage and fertility data available on the 1968-1988 cross-year file are not current for 1988. Due to lagging processing in 1987, 1988 marriage and fertility updates, and our desire to release the remainder of the 1968-1988 cross-year data, we have included only marriage and fertility data as collected in 1985 and 1986. We have not updated V32009-V32049 with 1987 and 1988 information. Each individual from whom we have collected information about a parent for either 1985 or 1986, whether the individual is currently response or nonresponse, has the individual identifiers of his or her parents (V32009, V32010 for mother; V32016, V32017 for father) on his or her record. The total number of children of the parent (V32012 for mother, V32019 for father), birth weight of the individual (V32014), and birth order with each parent (V32013 for mother and V32020 for father) are also included. The data record for each individual whose marital and fertility his- tories were collected, that is, a 1985 or 1986 Head, Wife, "Wife", or other FU member then age 12-44, contains birth dates of the four youngest and the oldest children (V32023-V32032), as well as the total number of births (V32022). Marriage data include the total number of marriages (V32033), month and year dates, and outcomes of the first and last marriages (V32034- V32047). The marital status of the individual (V32048) is included, as is a variable for the wave (1985 or 1986) in which the data were most recently gathered or updated (V32049). 1988 Parent and Help Supplement Data collection for 1988 included a large supplement asking PSID respondents about dollar and time help given and received with special em- phasis on the parents of Head and Wife/"Wife." The help variables (dollar help given: V15779-V15803, dollar help received: V15918-V15942, time help given: V15943-V15967, and time help received: V15968-V15992) contain totals from all sources, which are then broken into seven subcategories by the giver's or receiver's relationship 22 to the Head or Wife/"Wife." These relationship subcategories include parents, children, siblings, ex-spouses, other relatives, nonrelatives, and those whose relationship is unknown. Each of these categories contains three pieces of information: the number of reports of this type of help, the known dollar or hour amount, and a data quality code, which attempts to document missing data problems. See the tape codes for details. The parent information includes birth and death dates, health measures, marital status, living situation, state or country of residence, and the distance from parent's to sample family unit's dwelling. This in- formation is available for Head's and Wife's/"Wife's" father and mother (Head: V15804-V15831, Wife/"Wife": V15861-V15888). The 1988 family iden- tifier of each parent is included as a linking variable (Head's father: V15805, Head's mother: V15819, Wife's/"Wife"s/father: V15862, and Wife's/ "Wife's" mother: V15876). Users may thus analyze parents and adult children who are both in responding families. Information was also collected about parents' assets, housing status, and income. We have three sets of these variables for each pair of parents, depending on whether the parents share assets and housing. Table 7 summarizes the available information. Parents may share assets but not housing, or vice versa. We expected, from preliminary anecdotal reports, that we would encounter a number of such cases, and hence made processing plans that would cover these unusual situations. Notes are added to the variable descriptions in the tape codes explaining how such cases were to be treated. Table 8 shows actual case counts of the possible asset/housing sharing/splitting combinations for Head's and Wife's/"Wife's" parents (Head's parents: V15832-V15833, Wife's/ "Wife's" parents: V15889-V15890). Part 6: The 1985 Ego-Alter, Demographic Event History and 1968-1985 Rela- tionship Files The 1985 Ego-Alter file is a special file containing more detail about marital and fertility histories than is available on the main file. Be- cause the number of children and marriages reported varied, ranging from no children or marriages for an individual to seven marriages or twenty children, because not all users will be interested in the full detail, and because the size of the dataset already creates considerable constraints for users, summary variables regarding children and marriages were added to the main file data at the individual level. These should suffice for much analysis and are described above. However, we are aware that some users will need all of the detail, and so it is available as a separate dataset-- the 1985 Ego-Alter file--that can be merged with our family-individual data. The Ego-Alter file is so named because each of its records contains one relationship--one marriage between a pair of individuals, a natural parent-child relationship, an adoption, or substitute parenting. The "ego" is the individual who is the reporting spouse or parent. The "alter" is the individual about whom the relationship was reported. The user can merge data based on either the ego or the alter identifiers. These iden- tifiers are those always used in the PSID: the 1968 ID number (V30001) and 23 Table 7 ECONOMIC VARIABLES FOR HEAD'S AND WIFE'S/"WIFE'S" PARENTS __________________________________________________________________________ -------------------------------------------------------------------------- Head's Parents | Wife's/"Wife's" Parents -------------------------------+------------------------------- Combined Father Mother | Combined Father Mother for (and (and | for (and (and Father Step- Step- | Father Step- Step- and mother, father, | and mother, father, Mother if any) if any) | Mother if any) if any) ------------------------------------------+------------------------------- | Total 1987 V15834/ V15843/ V15852/ | V15891/ V15900/ V15909/ income V15835 V15844 V15853 | V15892 V15901 V15910 V15843*/ | V15900*/ V15844* | V15901* | Own/rent V15836 V15845 V15854 | V15893 V15902 V15911 V15845** | V15902** | House V15837/ V15846/ V15855/ | V15894/ V15903/ V15912/ value V15838 V15847 V15856 | V15895 V15904 V15913 V15846**/ | V15903**/ V15847** | V15904** | Whether V15839 V15848 V15857 | V15896 V15905 V15914 mortgage V15848** | V15905** | Whether V15840 V15849 V15858 | V15897 V15906 V15915 net assets V15849* | V15906* positive | | Amount of V15841/ V15850/ V15859/ | V15898/ V15907/ V15916/ net assets V15842 V15851 V15860 | V15899 V15908 V15917 V15850*/ | V15907*/ V15851* | V15908* -------------------------------------------------------------------------- Note: * Applies to situations in which parents do not share housing (V15832 or V15889=5, 8 or 9) but do share assets (V15833 or V15890=1). ** Applies to situations in which parents share housing (V15832 or V15889=1) but do not share assets (V15833 or V15890=5, 8 or 9). the person number (V30002), which uniquely identify any individual ever in a panel family. All egos were present in a 1985 family, but the alters may have been in the same or a different family, nonresponding, or may never have been included in a panel family. We issued a documentation volume for the 1985 Ego-Alter file that is available by request through ICPSR. That volume describes the data in much more detail, including tape codes. 24 Table 8 PARENTS SHARING ASSETS AND HOUSING ________________________________________________________________________ ------------------------------------------------------------------------ Head's Parents Wife's/"Wife's" Parents ------------------------------------------------------------- One or Do Not Miss- One or Do Not Miss- Both Share Share ing Both Share Share ing Deceased Housing Housing Data Deceased Housing Housing Data ------------------------------------------------------------------------ One or both deceased 4,088 0 0 0 5,162 0 0 0 Share as- sets 0 2,105 3 0 0 1,483 4 0 Do not share as- sets 0 0 895 0 0 0 451 0 Missing data 0 0 0 23 0 0 0 14 ------------------------------------------------------------------------ The length of each Ego-Alter record is short. It includes variables for month and year of births, marriages, deaths of children, children moving in and out, divorces, separations, and widowhood; where a natural, raised or adopted child was at the time the birth information was asked; relationship of an adoptive parent to a child; relationship of a substitute parent to a child; and the mother's participation in public programs when pregnant. Information and updates of marital and birth histories have been col- lected for each wave since 1985, but are as yet unreleased. Our current plan is to create merged files for data collected in years 1985 through 1989, and to release them shortly after we release the 1968-1989 cross-year files. All of the demographic events will not be included in one file, as was the case for 1985. Instead smaller files will be created, one for each type of event. Thus, three demographic event history files will be released: a 1985-1989 childbirth file (comprised of both adoptions and natural children), a 1985-1989 marriage history file, and a 1985 substitute parenting file (e.g., parenting step- and foster-children who have not been legally adopted). The Relationship file is a publicly released file that shows the blood, marital, or cohabitational relationship between pairs of individuals for each year from 1968 to 1985. Its expected release date is early summer 1991. The file contains two records for each pair of individuals who were members of family units that descended from a common, original 1968 family unit. It contains a total of 426,608 records of such relationship pairs. 25 The records contain variables that identify the relationship between the pair, that describe their living relationship, and that identify whether each individual was present in an interviewed household for each year 1968 through 1985. They also contain a summary gender variable and year-of- birth variable to aid in selection of records. The file is designed to be used in combination with the PSID's 1985 cross-year files. The Relation- ship file will be especially useful for clarifying relationships in the years prior to 1983, when the "relationship-to-head" variable was coded as a single instead of a two-digit item. It helps distinguish, for example, stepchildren from biological children. It can be used to identify rela- tionships of individuals living in separate family units but sharing the same dwelling. It will also be useful for identifying an individual's ex- tended kin, whether or not the individual has ever lived with them. The information on the Relationship file offers the opportunity for rich analysis of living-arrangement patterns of "family," broadly defined. Part 7: The Work History Supplement File The employment histories with event dating include a fairly large amount of data for some Heads and Wives/"Wives" who experienced several job changes. Thus, some cases do not include complete histories, from January 1983 until the time of the 1988 interview, on the cross-year tape. A separate tape will be created for all Heads and Wives/"Wives" in any year from 1984 through 1988. This tape will contain the additional employment history data, if any, as well as the complete family-level record. Each Head and each Wife/"Wife" have complete records. For married couples, the family-level data are duplicated for the two persons. Work history questions were extensively revised for 1988. Their orientation is employer-based rather than position-based, as in 1984-1987. Current year information on weeks spent working, unemployed, and out of the labor force is no longer available. Extra job questions are more exten- sive, with the addition of month and year the extra employment began and ended. The documentation volume for 1984-1988 work history will contain more detail on comparability of the 1988 information with that of prior waves. Part 8: Data Available For each year of this study, both an individual-level and a family- level tape have been created. In addition, the family-level tape has been merged with the previous years' family tapes so that there are two through twenty-one-year merged family-level tapes. The individual-level data from year five through twenty-one were merged. These tapes contain all in- dividuals with data in the current year. Additionally, for 1984 through 1988 we have created much expanded versions of the data that include all individuals ever in the study; that is, data for all nonresponse in- dividuals are available as a separate file that can be concatenated with the usual data. See the introduction to the individual-level tape code in Section II, Part 2 of this volume. For a more detailed description of all of the tapes, see the User Guide to the PSID. Two tapes were also created using the 1967 S.E.O. data for the part of the sample that was originally interviewed by the Census. 26 Employment histories for 1984-1987 and the 1985 Ego-Alter data on fer- tility and marriage histories are also available. Refer to Parts 6 and 7 above for brief descriptions. All inquiries for information about this study should be made in writ- ing to: Member Services, Inter-University Consortium for Political and So- cial Research, Institute for Social Research, University of Michigan, P.O. Box 1248, Ann Arbor, Michigan 48106. Refer to ICPSR study number 7439, and please specify which datasets you need. Machine readable documentations for the 1985 through the 1988 waves are available upon special request. They are comprised of three files for each wave: Volume I, identical to our published Volume I, and Volume II, which includes the Family Alphabetical Index and the Individual Alphabeti- cal Index on two separate files. These three files contain everything from the bound versions except the pages containing each year's questionnaires and editing worksheets. The questionnaires and worksheets can also be ob- tained through ICPSR by special request. Part 9: Creating Family-Level Data from the Cross-Year Family-Individual Tape Since the twenty-one-year individual-level data tape is very unwieldy (with 37,528 cases for both currently responding and nonresponding in- dividuals and 30,169 tape locations), and users might well be interested in analyzing the data largely on a family-level basis, we make suggestions on the creation of family files from the cross-year family-individual tapes. The structure of the individual-level tape combines family-level data for each person in the family unit together with information unique to that person. That is, each member of a family has family-level data identical with the data of all the other family members in that family. The follow- ing instructions remove the duplication of these data. Each individual is assigned a unique sequence number, which indicates the person's position and status on the 1988 list of family members. Thus, the first person listed, the Head of the family, is 01, the second person listed is 02, and so on. To create a family-level file, it is necessary only to write onto a new tape those cases where V30591 (1988 Sequence Number) is equal to 01, since each family must have at least one member, although it may or may not have two or more.5 One may also, when creating the family-level tape, truncate the data at V16208, since most individual-level variables for the Head are also present at the family level. These instructions create a merged 1968-1988 family-level file for currently responding families. ___________________________________________________________________________ 5It is suggested that V30591, Sequence Number, be used instead of V30592, Relationship to Head, because although each family has one and only one current Head (i.e., where V30591 = 01-20 and V30592 = 10), it is pos- sible that the prior year's Head of the family has moved out since the previous interview and a new Head is present for the current interview. Relationship to Head for movers-out is coded with reference to the previous year's Head, so for both the current Head and the previous Head, V30592 = 10. 27 For other years' merged family-level files, the family-individual response and nonresponse files must be concatenated, and the Sequence Num- ber variable for the latest desired year of data should be used. Again, this produces a file of families who were response in this latest year and eliminates families who had already become nonresponding. See the User Guide for more detail. Part 10: PSID User Guide The PSID staff has completed a User Guide to the panel study. The volume was designed to supplement, but not replace, the documentation volumes issued for each year's data. It is published in a loose-leaf form, so that updates can easily be made. Chapters in the User Guide include PSID history, sample composition and weighting, how to deal with family composition and change, structure of the data tapes, study content, and other topics of interest to users. The User Guide is included with the set of documentation volumes that accompany an order for PSID tapes. It can be ordered separately as well. The Guide is being rewritten, and we expect to complete the new version in Fall 1991. For information on obtaining the User Guide, and other PSID products, contact Janet Vavra, Inter-University Consortium for Political and Social Research, P.O. Box 1248, Ann Arbor, Michigan, 48106 (313) 763-5010. Part 11: PSID CD-ROM Historically, PSID data files have been released through the Inter- University Consortium for Political and Social Research (ICPSR) on magnetic tape. Beginning with cross-year waves 1968-1987, the ICPSR has released a field-test CD-ROM version of the rectangular cross-year family-individual response and nonresponse files. An ASCII version of these two files have been mastered onto two CD-ROMs: the response file on one, and the non- response file on the other. SAS and SPSS-X program files have also been placed on the CD-ROMs to facilitate retrieval of data for users of those statistical software systems. 28 SECTION II TAPE CODES FOR WAVE XXI Part 1: Twenty-first-Year Family-Level Tape Code The following is the codebook for the twenty-first wave of family- level data from the interview schedule. The twenty-one-year individual- level codebook can be found in Part 2 of this section. The variable num- bers and tape locations refer to those on the l968-1988 cross-year tape. For family-level codes for the first five waves of this study, see A Panel Study of Income Dynamics, Volume II, Section II. The remainder of the family-level codes for Waves VI through XX will be found in successive volumes entitled A Panel Study of Income Dynamics: Procedures and Tape Codes. The distributions for the following variables are weighted and based on all families interviewed in 1988. To generate distributions on field amounts, percent nonzero and mean nonzero values are provided for relevant variables. Tape Code Information The example below illustrates the information contained in this codebook for a typical variable. The numbers in brackets do not appear in the codebook, but refer to the explanations which follow this example. [1] V14809 [2] 'MODE OF INTERVIEW 88' [3] TLOC= 26044 [4] MD=9 [5] Mode of Interview in 1988 [10]----- [11]---- [6] [7] [8] [9] 589 5.6 0. Personal interview 6,510 94.1 1. Telephone interview 14 0.2 2. Mail interview 1 0.0 9. NA [1] Indicates the cross-year variable number. A variable number is as- signed to each item in the study. (See the introduction to the numerical index, Section III, Part 1 of this volume, for a list of the range of variable numbers specific to each year.) [2] Indicates the abbreviated variable name (maximum of 24 characters) used in the OSIRIS system to identify the variable for the user. This abbreviated variable name is identical to the variable name listed in the OSIRIS dictionary for this variable. It is also 29 30 listed as a subheading of the printout when a variable is accessed in an OSIRIS program. When used in this manner, the abbreviated variable name can be useful as a cross-reference tool, as well as a way to avoid errors. Refer to the list of abbreviations below for some help in translating the names into sensible English. [3] Indicates the starting location and ending location for this vari- able when the data are stored on a magnetic tape in the OSIRIS for- mat. [4] Indicates the code value for missing data. In this example, code values equalling nine are missing data (MD=9). Alternative state- ments for other variables are "MD=0 or GE 8" or "MD=GE 7." In cases where nothing is printed in this space, missing data are not per- mitted for the variable; values were either assigned for such cases, or missing data were impossible. Some analysis software packages (including the OSIRIS software package) require that certain types of data which the user desires to exclude from analysis can be designated as "missing data," e.g., inappropriate, unascertained, or ambiguous data categories. Al- though these codes have been defined by the PSID staff as missing data categories, this does not mean that the user should not or can- not use them in a substantive role if so desired. [5] Indicates the full question number that was used in the question- naire, as well as the exact wording of the questionnaire item; for variables not coded directly from the questionnaire, such as generated data, an appropriate title appears here. [6] Indicates the unweighted N for each code value. Blanks indicate that no cases have this value. [7] Indicates the weighted percentages for each code value. Blanks in- dicate that no cases have this value. [8] Indicates the code values occurring in the data for this variable. For variables containing field amounts, refer to the notes appearing directly below items [10] and [11] for the range of data values. [9] Indicates the textual definitions of the codes. Abbreviations com- monly used in the code definitions are "DK" (Don't Know), "NA" (Not Ascertained), and "Inap." (Inappropriate). [10] Indicates the "% nonzero" value, where specified. These are weighted. [11] Indicates the "mean nonzero" value, where specified. These are weighted. 31 List of Standardizations of Common Abbreviations Used by the Panel Study in Generating Variable Names with OSIRIS (see item 2 above) ACC . . . . . Accuracy ACCT . . . . . Account ACUM . . . . . Accumulate ANN . . . . . Annual BEG . . . . . Began; Begin(ning) BEN . . . . . Benefit(s) BUS . . . . . Business CHKPNT . . . . Checkpoint CNTY . . . . . County COMM . . . . . Commission COMP . . . . . Composition; Compensation COMPL . . . . Complete(d) CONDITN . . . Condition CONTR . . . . Contribut(e/ion) CORP . . . . . Corporation CR . . . . . . Care CS . . . . . . Coversheet CVD . . . . . Covered DED . . . . . Deduct(ed/ion) DEF . . . . . Deferred DEG . . . . . Degree DIGT . . . . . Digit DIV . . . . . Dividends E . . . . . . Section of questionnaire (Sec- tion C or F) applying to those who are currently employed EARNR . . . . Earner EDUC . . . . . Education ELIG . . . . . Eligible EMP(R) . . . . Employed; Employer EXC . . . . . Except EXEMP . . . . Exemption(s) FA . . . . . . Father FAM . . . . . Family FD ST . . . . Food Stamps FORML . . . . Formula FR . . . . . . From GOVT . . . . . Government H or HD . . . Head HOSP . . . . . Hospital(ized) HR(S) . . . . Hour(s) HSEWRK . . . . Housework HTG . . . . . Heating ILL . . . . . Illness IND . . . . . Industry; Individual INHER . . . . Inherit(ance/ed) IN(S) . . . . Insurance INT . . . . . Interest IW . . . . . . Interview LAB or LAB FRC Labor; Labor Force LF . . . . . . Life LFT . . . . . Left 32 LGST . . . . . Largest LK(G) . . . . Look(ed); Looking LST . . . . . Last LTR . . . . . Later LVG . . . . . Living MARR . . . . . Marriage; Married MED . . . . . Medical MKT GARDN . . Market Gardening MNY . . . . . Money MO . . . . . . Mother; Month MORT or MTG . Mortgage MOS . . . . . Months MTRL . . . . . Maternity Leave NONFU . . . . Non-Family Unit Member OCC . . . . . Occupation OFUM . . . . . Other Family Unit Member (other than current Head/Wife/ "Wife") OPP . . . . . Opportunity OT . . . . . . Overtime OTR . . . . . Other person(s); Other PAY/HR . . . . Hourly rate, in dollars and cents PD . . . . . . Paid PEN(S) . . . . Pension PERM . . . . . Permanent(ly) PLN . . . . . Plan PMNT . . . . . Payment PN . . . . . . Person Number POS . . . . . Position PRAC . . . . . Practice PRES . . . . . Present PREV . . . . . Previous PRIN . . . . . Principal PROF . . . . . Professional PROFT . . . . Profit PROP . . . . . Property PT . . . . . . Part; Portion; Private PY . . . . . . Pay(s) R . . . . . . Section of questionnaire (Sec- tion E or H) applying to those who are currently retired, housewives, or students REC(D) . . . . Receive(d) RELS . . . . . Relatives REM . . . . . Remaining RET . . . . . Retire(d); Retirement RL EST . . . . Real Estate RQ . . . . . . Require(d) SK . . . . . . Sick S/O . . . . . Splitoff SPG . . . . . Spring ST . . . . . . Start(ed); State STAT . . . . . Status 33 STD . . . . . Standard SUBSDY . . . . Subsidy SUPP(S) . . . Support; Supplement(s) SUPR . . . . . Supervis(e/or) TOT . . . . . Total TR Y . . . . . Transfer Income TX . . . . . . Tax TXBL Y . . . . Taxable Income U . . . . . . Section of questionnaire (Sec- tion D or G) applying to those who are currently unemployed and looking for work UN . . . . . . Union UNEMP . . . . Unemployed; unemployment UNINCORP . . . Unincorporated UTIL . . . . . Utility; Utilities VAC . . . . . Vacation W or WF . . . Wife WELFR . . . . Welfare WHR . . . . . Where WK . . . . . . Week WRK(D) . . . . Work(ed) WRKG . . . . . Working WRKHRS . . . . Work Hours WT . . . . . . Weight WTR . . . . . Whether; if X . . . . . . Times; cross (as in cross- year) XPCT . . . . . Expect XTRA . . . . . Extra Y . . . . . . Income # . . . . . . Number (of) 34 1988 FAMILY TAPE CODE V14801 'STUDY NUMBER (717) ' TLOC= 26024-26026 Study Number 714 (Wave 21) V14802 '1988 INTERVIEW NUMBER ' TLOC= 26027-26030 1988 Interview Number The range of values for this variable is 0001-7114. V14803 'CURRENT STATE ' TLOC= 26031-26032 MD=99 State of Residence at Time of 1988 Interview (PSID Code) Please refer to Appendix 1, Wave XIV documentation (1981 data), for state and county codes. 99. NA V14804 'CURRENT COUNTY ' TLOC= 26033-26035 MD=999 County of Residence at Time of 1988 Interview (PSID Code) Please refer to Appendix 1, Wave XIV documentation (1981 data), for state and county codes. 999. NA V14805 'CURRENT STATE+CNTY ' TLOC= 26036-26040 MD=99999 State and County of Residence at Time of 1988 Interview (PSID Code) Please refer to Appendix 1, Wave XIV documentation (1981) data), for state and county codes. V14803 and V14804 are combined here into one variable; the first two digits represent the state code and the last three, the county. V14806 'SIZE LGST CITY/COUNTY 88' TLOC= 26041 MD=9 Size of Largest City in County of Residence 1,679 17.3 1. SMSA: largest city 500,000 or more 1,773 25.8 2. SMSA: largest city 100,000-499,999 743 10.6 3. SMSA: largest city 50,000-99,999 728 13.0 4. Non-SMSA: largest city 25,000-49,999 991 16.1 5. Non-SMSA: largest city 10,000-24,999 1,162 16.7 6. Non-SMSA: largest city under 10,000 38 0.6 9. NA; household is outside USA V14807 'SPLITOFF INDICATOR 88' TLOC= 26042 35 36 - RAW DATA Splitoff Indicator: Color of Coversheet 6,862 96.6 0. Aqua (Main Family) 252 3.4 1. Pink (Splitoff) V14808 'WHETHER REFUSED 88' TLOC= 26043 MD=9 Whether Initially Refused in 1988 7,083 99.6 0. Never refused 30 0.4 1. Refused at least once 1 0.0 9. NA V14809 'MODE OF INTERVIEW 88' TLOC= 26044 MD=9 Mode of Interview in 1988 589 5.6 0. Personal interview 6,510 94.1 1. Telephone interview 14 0.2 2. Mail interview 1 0.0 9. NA V14810 'FAM COMP CHANGE 88' TLOC= 26045 Family Composition Change between 1987 and 1988 Waves Codes 2 through 8 take priority over codes 0 and 1. 5,348 78.1 0. No change; no movers-in or movers-out of the family. 1,059 12.8 1. Change in members other than Head or Wife/"Wife" only. 212 2.6 2. Head is the same person as in 1987 but Wife/"Wife" left or died; Head has new Wife/"Wife"; used also when cohabiting, nonrelative female becomes "Wife." 126 2.1 3. Wife/"Wife" from 1987 is now Head. 108 0.9 4. 1987 female Head