to $25,000" components on the basis of the relative numbers of such returns observed in the sample, taking into account the comparative number of blocks ending in "0" and "5" received from each district. The foregoing estimating procedures with respect to the partitioning of the "Under $25,000" universes differed from prior years when precise universes with respect to all income stratification levels among agent returns were determined as part of the sampling process. Altogether, three distinct income groups among Form 1040 returns were recognized in collectors' offices in the segregation and blocking procedures-namely, (1) returns with adjusted gross income under $7,000 and total receipts from business, if any, under $25,000, designated herein as collector returns, (2) returns with adjusted gross income from $7,000 to $25,000, and returns with adjusted gross income under $7,000 and total receipts from business $25,000 or more, designated as agent returns, and (3) returns with adjusted gross income $25,000 or more, also designated as agent returns. Analysis of the sample received disclosed that a number of returns were segregated and blocked erroneously-that is, returns properly classifiable in a specific group (1), (2), or (3) were classified in either of the other two groups. Throughout the sample selection, tabulation, and weighting operations, such returns were processed according to the strata in which they were blocked. After extension, the data on erroneously classified returns were merged with the strata with which their income sizes were identified, so that the size distributions reflect the income reported and not the classes into which returns were sorted for administrative purposes. The degree of overlapping between collector and agent returns, as indicated by analysis of the sample, is relatively negligible in occurrence and in effect on the final data. The largest body of erroneously segregated returns comprises approximately twenty-nine thousand returns with adjusted gross income under $25,000, classified among those with adjusted gross income $25,000 or more. Since they are completely represented in the sample, their effect on the final data for the lesser income returns is in the direction of reducing sampling error. The most consequential body of erroneously segregated returns comprises an estimated total in the universe of about fourteen thousand returns with adjusted gross income $25,000 or more, erroneously classified among the lesser income agent returns. The effect of this group is to introduce an estimated element into the data for the larger income area, which otherwise are based upon a complete count. The aggregates of reported stratum universes for all collection districts, after such adjustments as were clearly indicated by secondary sources, and after the partitioning of reported universes of agent returns into adjusted gross income strata, provided the basis for weights applied uniformly to the sample data from all collection districts for purposes of the national distributions. The separate collection district stratum universes provided the basis for a series of independent collection district weights for purposes of the State income distributions in table 11 and the State selected aggregates in table 12. WEIGHTING PROCEDURES Although the sampling pattern for 1948 called for 16 distinct strata sible to achieve a substantial degree of simplification in the tabulating and weighting operations by combination of multiple strata, where the data in the tables are composites of strata subject to the same sample selection ratio and where the percentage representations attained for the strata were in close agreement. For example, in the case of the Form 1040A and collector Form 1040 returns, the proportions sampled for the seven separate strata were in close approximation to each other, and all such returns were jointly processed. Similarly, combined processing of the assessable, overpayment, and even elements among agent returns, Form 1040, was accomplished for each of two income ranges-adjusted gross income under $10,000 and adjusted gross income $10,000 to $25,000, since representation of the various year-end adjustment statuses corresponded closely for each income range. In total, four strata were differentiated for tabulating and weighting purposes. These comprise (1) Form 1040A and collector returns, Form 1040; (2) agent returns, Form 1040, with adjusted gross income under $10,000; (3) agent returns, Form 1040, with adjusted gross income from $10,000 to $25,000; and (4) agent returns, Form 1040, with adjusted gross income $25,000 or more. The table below presents, for each of the four estimating strata, the number of returns in the universe and the number of returns in the sample. Individual returns for 1948: Number of returns filed and number of returns in sample by estimating strata [Number of returns in thousands] Estimating strata Form 1040A and collector returns, Form 1040.. Agent returns, Form 1040, with adjusted gross income under $10,000. Total, all returns...... 1 An estimated fourteen thousand returns with adjusted gross income $25.000 or more were erroneously classified among returns with adjusted gross income under $25,000. The sample of these returns was tabulated and weighted as part of the $10,000 to $25,000 estimating stratum, and the total number of such returns filed and the number in the sample are included among corresponding data for the "$10,000 to $25,000" stratum in this table. However, in statistical tables distributing returns by size of adjusted gross income, they are classified according to size of income reported. 'Approximately twenty-nine thousand returns with adjusted gross income under $25,000 were erroneously classified among returns with adjusted gross income $25,000 or more. These returns were all included in the sample and were tabulated as part of the "$25,000 or more" stratum. However, in statistical tables distributing returns by size of adjusted gross income they are classified according to size of income reported. SAMPLING VARIABILITY Insofar as data in this volume are tabulated from samples, they are subject to sampling variability. The degrees of variability shown on page 50 in terms of relative errors relate to specific frequency levels and not to money amounts. Each of the various income areas constituting an independent estimating stratum has its own variability pattern; accordingly, the relative errors are presented separately for three distinct income areas, as follows: (1) returns with adjusted gross income under $7,000, (2) returns with adjusted gross income from $10,000 to $25,000. Group (1) is a composite, collector returns contributing about 99 percent of the total population and agent returns contributing about 1 percent. Relative errors for group (1) are based on the collector component, since the effect on relative error of agent returns with adjusted gross income under $7,000 is generally negligible. In computing the limits of variation and relative sampling error of a given frequency, a range of two standard errors was used; chances are somewhat over 19 out of 20 that the frequency as estimated from the sample tabulation differs from the actual frequency, which would have resulted from tabulation of the entire universe, by less than twice the standard error. Variation beyond the two-error limit would occur less than 1 time in 20 and would be sufficiently rare to justify a two-error range in defining sampling variability. Thus, all limits of variation are determined on the basis of two standard errors, and the degrees of variability are expressed in terms of relative errors, or percents of the numbers to which they relate. Specific cell frequencies in national distributions in this volume, unless otherwise footnoted, are subject to maximum variation of less than 30 percent. Frequencies which are subject to maximum variation of more than 30 percent, but not more than 100 percent, are footnoted to indicate their great variability. Frequencies which are subject to maximum variation of more than 100 percent and associated data are not separately shown since they are considered too unreliable for general use; they are, however, included in the totals. Sampling variability at selected frequency levels Frequencies by size of specific source in table 4 are footnoted within each adjusted gross income class to indicate sampling variability in accordance with the foregoing rules. However, the totals embracing all adjusted gross income classes are not footnoted. These totals are composites of the various estimating strata, and the sampling variabilities at specific levels are not constant, but depend on the interrelationship of the contributing strata. As previously stated, uniform sample selection ratios within each in view of the block sampling technique applied, some differences in percentage representation were noted between the various collection districts. In general, the differences were minor, and no appreciable bias is believed to result from the tabulation and extension of the sample data for all collection districts combined for the national distributions. For purposes of deriving the State data in basic tables 11 and 12, separate weights were devised for each district, taking into account the particular district universes and sample sizes. The differences in data attributable to the dual weighting system are indicated in the discussion of State aggregates in a subsequent paragraph. The preceding analyses with respect to variability have been confined to cell frequencies. Specific consideration has not been given to associated money amounts; however, the homogeneity within the strata employed for sampling purposes, and the large number of returns included in the samples, together with the progressively increased sample sizes in the higher income areas, generally serve to limit the sampling variability with respect to such amounts. An exception occurs in the case of returns with no adjusted gross income. Stratification by size of adjusted gross deficit was not administratively feasible, nor was it possible to establish separate estimating strata for returns with adjusted gross deficit, distinct from returns with adjusted gross income. Returns, Form 1040, with adjusted gross deficit, regardless of size of deficit, and with total receipts from business, if any, under $25,000, were classified along with the much larger group of returns, collector Form 1040, with adjusted gross income. Returns, Form 1040, with adjusted gross deficit and with total receipts from business of $25,000 or more were classified among agent returns with adjusted gross income under $25,000 and were processed statistically among agent returns with adjusted gross income under $10,000. In view of the facts that (a) the returns with adjusted gross deficit are extremely heterogeneous and were not stratified by size, (b) the samples of such returns are generally inadequate, and (c) they did not constitute an independent estimating stratum, money amounts associated with such returns in this volume may be subject to marked sampling variability. STATE AGGREGATES Data by States are confined to returns with adjusted gross income since returns with adjusted gross deficit were too few, and the sampling variability too great, to permit presentation on a State basis. The distributions in basic tables 11 and 12 were derived from the basic sample of returns with adjusted gross income. Independent weighting factors were established for each of the estimating strata processed for each collection district. Despite the fact that the same sample served as a basis for both the national and the State distributions, and the fact that the national stratum universes to which sample data were extended equal the total of the State universes, slight discrepancies exist between items, as aggregated in the State tables, and corresponding items associated with returns with adjusted gross income in the national tables. These discrepancies are the result of (a) the dual system of weighting, districts for the national distributions and an independent series of weights for each collection district for the State distributions, and (b) the use of rounded weighting factors. A difference between the two series in the $25,000 or more adjusted gross income area, results from the late receipt of a few hundred returns. These were included in the State distributions in tables 11 and 12, but were not available in time to be incorporated in the national distributions. HISTORICAL DATA INDIVIDUAL RETURNS AND FIDUCIARY RETURNS, 1913–1943; INDIVIDUAL RETURNS, 1944-1948 A résumé of significant data from the individual returns and returns of fiduciaries, 1913-1943, and from individual returns, 1944-1948, is presented in basic tables 13 through 16. These tables are prepared mainly from data in the reports, Statistics of Income, published annually since 1916, in accordance with provisions of the various revenue acts, and from data in the Annual Report of the Commissioner of Internal Revenue for 1914-1916. The number of returns and tax tabulated for income years 1913-1915 are as reported by the Commissioner for the fiscal year ended June 30, immediately following, and the net income is estimated from the number of returns filed and the average net income for each income class. Historical data for 1916-1936 are tabulated for individual and fiduciary returns, Forms 1040 and 1040A, including individual returns with no net income for 1928-1936. Data for 1937-1940 are tabulated from individual returns, Forms 1040 and 1040A, with net income and with no net income, and from taxable fiduciary returns for estates and trusts, filed on Form 1041 or filed (improperly) on Form 1040, including for 1937 nontaxable fiduciary returns for estates and trusts which were filed (improperly) on Form 1040. Data for 1941-1943 are tabulated from individual returns, Form 1040, with net income and with no net income, optional returns, Form 1040A, with gross income not over $3,000, and taxable fiduciary returns, Form 1041 (or Form 1040). Data are tabulated for 1944-1947 from individual returns, Form W-2 and Form 1040, and for 1948 from individual returns, Form 1040A and Form 1040, with adjusted gross income or with no adjusted gross income. Basic table 13 shows historical data for the number of taxable and nontaxable returns, the income, tax, and tax credits; basic table 14 shows by income years the number of returns, the income, tax, and effective tax rate tabulated by income classes; table 15 shows for each year sources of income, deductions, and net income or deficit for all returns; and table 16 shows the number of returns, the income, and tax for each State and Territory annually for the 10-year period ending with the current year. Changes in the provisions of the Federal income tax laws under which the returns are filed interfere with the comparability of the statistical data over a period of years. The major provisions of the Federal tax laws from 1913-1948 are summarized on pages 426-447. Revisions in the income tax return form to incorporate changes in the law often eliminate, combine, or introduce items which make it impracticable or impossible to present comparable data from year to |