Household Food Security in the United States, 1999. By Margaret Andrews, Mark Nord, Gary Bickel, and Steven Carlson. Fall00

This is the latest in the series of reports Measuring Food Security in the United States.

Abstract

This report provides the most recent data on the food security of American households. Preliminary estimates indicate that 89.9 percent of American households were food secure in 1999, up 0.6 percentage point from 1995. Some 31 million Americans were food insecure--they did not have assured access at all times to enough food for an active, healthy life. In 3 percent of all households, one or more household members were hungry, at least some time during the year, because of inadequate resources. Between 1995 and 1999, the number of food-insecure households fell by 12 percent, and the number with hunger due to inadequate resources fell by 24 percent. Households with incomes between 50 and 130 percent of the poverty line were the only household types among the 30 subgroups studied to show a higher rate of food insecurity in 1999 than in 1995.

About the Authors

Margaret Andrews is Assistant Deputy Director for Food Stamp Research in the Food Assistance and Nutrition Research Program and Mark Nord is Team Leader for Food Stamp and Food Security Research; both are with the Food and Rural Economics Division, Economic Research Service, U.S. Department of Agriculture. Gary Bickel is Project Officer for Food Security Measurement and Steven Carlson is Director of Family Programs Staff; both are in the Office of Analysis, Nutrition, and Evaluation, Food and Nutrition Service, U.S. Department of Agriculture.

Acknowledgments

The authors would like to acknowledge the contribution of James Ohls, Abhijay Prakash, Larry Radbill, and Allen Schirm of Mathematica Policy Research, Inc. in edit- ing and preparing the data for 1995-97. The 1998 and 1999 data were edited and pre- pared by Mark Nord.

Introduction

Nearly 90 percent of all U.S. households, 240 million Americans, were food secure during the 12 months ending in April 1999, according to preliminary data. "Food secure" means they had assured access, at all times, to enough food for an active, healthy life. The remaining 10.1 percent of U.S. households, 31 million Americans, were food insecure, meaning that at some time during the previous year they were uncertain of having, or unable to acquire, adequate food sufficient to meet basic needs at all times due to inadequate household resources for food. Of these, about 3 million households were food insecure to the extent that one or more household members were hungry due to inadequate resources at least some time during the year.

This report presents the most recent available information on the changing level of food security among U.S. households over the period 1995-99 and estimates of the extent of hunger and food insecurity in 1999. While these data are subject to revision as independent analyses and reviews are completed, they are being released to provide timely information on the current food security status of the U.S. population.

The incidence of food insecurity and hunger in the United States has declined significantly since food security data were first collected in 1995. The number of food-insecure households fell by 12 percent between 1995 and 1999. The prevalence of hunger fell even more sharply. The number of households where at least one member was hungry because of insufficient resources in the previous 12 months fell by 24 percent between 1995 and 1999. 1 Substantial decreases in food insecurity and hunger were observed, over this period, for all household types. However, even though there were fewer food-insecure households with incomes between 50 and 130 percent of the poverty line in 1999, a higher proportion of households in this income range were food insecure in 1999 than in 1995.

Method

The preliminary results presented in this report are based on data collected in the Food Security Supplements to the Current Population Survey (CPS) for the years 1995-1999. The report provides an update to its predecessor in this series, (Bickel, Carlson, and Nord, 1999), released in July 1999 by USDA's Food and Nutrition Service. Briefly, the measurement method uses responses to a series of questions about conditions and behaviors known to characterize households having difficulty meeting basic food needs. 2 Each question asks whether the condition or behavior occurred during the previous 12 months and specifies a lack of money or other resources to obtain food as the reason for the condition or behavior. 3 (Response frequencies for the 18 items used to classify households are provided in Appendix A. Full-question wordings are presented in Hamilton et al., 1997a, Price et al., 1997, and Bickel et al., 2000.) Each respondent household is classified into one of three categories (food secure, food insecure without hunger, food insecure with hunger) based on the household's overall pattern of response to all items. Appropriate weighting factors are then applied to the surveyed households to make nationally representative prevalence estimates.

Due to changes in the administration of the Food Security Supplement in the initial years of data collection, statistics presented in this report are based on two different data-editing protocols. One protocol adjusts the data so that prevalence estimates are consistent across all years, as well as comparable with the same adjusted data series reported in Household Food Security in the United States, 1995-1998: Advance Report. This provides the most accurate picture of trends in the prevalence of food insecurity and hunger over the past 5 years. The second data-editing proto- col makes full use of all households' responses and makes no adjustments for cross-year comparability. Thus, it provides the most accurate picture of the cur- rent prevalence of food insecurity and hunger. The adjustments required for cross-year comparability result in understating the prevalence of food insecurity. For example, the best estimate of food-insecurity prevalence in 1999, using the unadjusted data, is 10.1 percent (table 2), while the estimate adjusted for cross- year comparability (table 1) is 8.7 percent. The estimated prevalence of hunger, the more severe measure, is less affected by the screening change. The unadjusted estimate for 1999 is 3.0 percent, compared with an estimated prevalence of 2.8 percent, based on the data adjusted for comparability. For future data collections, continued use of the 1998/1999 survey administration procedures is anticipated. Thus, food-security and hunger prevalence statistics based on the unadjusted data will be directly comparable from 1998 onward. The present report provides the bridging comparison, with table 1 and appendix table C-1 being comparable to earlier years and table 2, table 3, and appendix table C-2 being comparable for 1998, 1999, and future years. 4

Trends in Prevalence of Food Insecurity and Hunger

Prevalence estimates presented in this section are based on data from the 1995-1999 surveys adjusted for cross-year comparability due to differences in survey administration in 1995-1997.

Figure 1 shows the pattern of change in the prevalence of food insecurity and hunger between 1995 and 1999. 5 In general, food insecurity and hunger declined over the period with year-to-year deviations from trend. However, the slight upturn in 1996 and the more sizable increase in 1998 warrant further study. One possibility is that the data are subject to a seasonal influence, as food-security data collection occurs in April in odd-numbered years and August or September in even years.

Examples of Questions from the Survey

"We worried whether our food would run out before we got money to buy more." Was that often, sometimes, or never true for you in the last 12 months?
"The food that we bought just didn't last and we didn't have money to get more." Was that often, sometimes, or never true for you in the last 12 months?
In the last 12 months did you or other adults in the household ever cut the size of your meals or skip meals because there wasn't enough money for food?
In the last 12 months were you ever hungry, but didn't eat, because you couldn't afford enough food?
(For households with children) In the last 12 months did any of the children ever not eat for a whole day because there wasn't enough money for food?

The declines in food insecurity and hunger from 1998 to 1999 were in the opposite direction of the changes observed from 1997 to 1998 (table 1). While food- insecure households accounted for 8.7 percent of households in the 1997 and 1999 data (as adjusted), this proportion had risen to 10.2 percent (adjusted) in 1998. A similar change occurred in the proportion of households that were food insecure with hunger. In 1998 the proportion of households experiencing hunger rose 0.5 percentage point above the 1997 rate, but in 1999 fell 0.8 percentage point to the lowest level (2.8 percent, adjusted) since these data have been collected.

Between 1995, when food security was first measured at the national level, and 1999, the incidence of food insecurity declined for nearly all household types (fig. 2). 6 Large percentage decreases were observed for female-headed households with children, Hispanic households, households living in central cities, and households in the Midwest. However, for households with income between 50 and 130 percent of the poverty line, the percentage of households experiencing food insecurity increased, even though there were 715,000 fewer food-insecure households in this category in 1999 than in 1995. For households with income above 185 percent of the poverty line, the percentage experiencing food insecurity was unchanged.

For most household types, the declines in hunger prevalence between 1995 and 1999 were similar to the declines in food insecurity (fig. 3). However, among low-income households, the prevalence of hunger declined even as the prevalence of food insecurity increased. In other words, while the overall prevalence of food insecurity rose for households with incomes below 130 percent of the poverty line in this period, the average severity of that food insecurity declined. The distribution within this group shifted away from the more severe range (food insecure with hunger) to the less severe range (food insecure without hunger).

Prevalence of Food Insecurity and Hunger in 1999

Prevalence estimates presented in this section are based on the unadjusted data from the 1999 survey. Although they cannot be compared with 1995-1997 estimates, they are comparable to unadjusted 1998 statistics presented in Appendix C and with planned future data collections.

The prevalence of food insecurity and hunger varied considerably among household types (table 2). As in previous years, some groups experienced rates of food insecurity greater than the national average: n 36.7 percent of households with incomes below the official poverty line ($16,895 for a family of four in 1999),

Overall, households with children experienced food insecurity at double the rate for households without children (14.8 versus 7.4 percent). Among households with children, those with married-couple families showed a lower rate of food insecurity (9.6 percent) than the national average (10.1 percent) for all house- holds.

In terms of individuals, about 31 million persons lived in food-insecure households, with 12 million (39 per- cent) of them being children (table 2). In addition, some 5 million adults and 2.7 million children lived in households where someone in the household had experienced hunger in the previous year due to inadequate resources.

The prevalence of food insecurity for households located in central cities (13.8 percent) and nonmetro areas (10.1 percent) substantially exceeded that of households in suburbs and other metropolitan areas outside central cities (7.7 percent). Regionally, the food-insecurity rate was above the national rate in the South and West (11.1 and 11.8 percent, respectively) and below the national rate in the Midwest and Northeast (8.3 percent in both cases).

Hunger prevalence varies with household characteristics in a pattern similar to that observed for food insecurity. Hunger prevalence is much higher than average among families headed by a single woman (8.1 per cent), black and Hispanic households (6.4 and 5.5 per- cent, respectively), and households below the poverty line (12.2 percent). Geographically, hunger is more common in central-city households (4.2 percent) and in those in the South and West (3.3 and 3.8 percent, respectively).

Figure 2 Prevalence of food insecurity (with or without hunger), 1995 versus 1999
See notes to tables.

Calculated by ERS based on data from Current Population Survey Food Security Supplements as of April 1995 and April 1999, adjusted for cross-year comparability.

Figure 3 Prevalence of hunger, 1995 versus 1999
See notes to tables.

Table 2--1999: Prevalence of food security, food insecurity, and hunger by selected characteristics of households
(Based on unadjusted data) See notes to tables.

Hunger rates below the national average were observed in households with elderly (1.6 percent), married-couple families with children (1.6 percent), multiple-adult households with no children (1.8 percent), white non- Hispanic households generally (2.1 percent), and households at or above 185 percent of the poverty line (1.0 percent).

Prevalence of Food Insecurity and Hunger in Low-Income Households

Prevalence estimates presented in this section are based on the unadjusted data from the 1999 survey.

Food insecurity and hunger are conditions that result from insufficient household resources. Food insecurity is six times as prevalent, and hunger eight times as prevalent, in households with annual income below 185 percent of the poverty line as in households with income above that line (table 2). However, many factors that might affect a household's food security (e.g., job loss, divorce, or other unexpected events) are not captured by an annual income measure. A few house- holds experience episodes of food insecurity, or even hunger, even though their annual income is well above the poverty line. On the other hand, many low-income households (including almost two-thirds of those with income below the official poverty line) manage to remain food secure.

Table 3 presents food security and hunger statistics for households with annual incomes below 130 percent of the poverty line. About one-third of these low-income households were food insecure, and in 10.7 percent of those households, people were hungry at times during the year. Low-income households with children were more vulnerable to food insecurity than were house- holds without children (40.3 percent versus 26.1 per- cent), and 42 percent of children in low-income house- holds lived in food-insecure households. The prevalence of hunger, however, was slightly lower in low- income households with children than in those without children. Low-income single mothers with children were especially vulnerable; 44.3 percent of these households were food insecure, and in 12.9 percent of them, members were hungry at times during the year because of lack of money or other resources for food.

Caveats

The preliminary findings presented in this report are based on application of a consistent methodology to the five data sets for 1995-1999. Analysis of all these data sets is still in progress for presentation in subsequent reports. These ongoing analyses may produce further refinements to the prevalence estimates for these years, or in their form of presentation. Consequently, final estimates may differ in minor detail from those presented in the present report.

The measure of children in food-insecure households with hunger is not, as such, a valid estimate of the number of children directly experiencing hunger, but an upper bound for this figure. In most households, children are shielded from food deprivation until the level of deprivation among adult members is quite severe. Work is currently under way to develop a more accurate estimate of children's hunger.

Other sources of possible estimation bias in the prevalence estimates include: the omission of homeless persons from the CPS household-based sample, a probable underreporting bias of unknown size, and a potential over-estimation bias resulting from the highly skewed distribution of households across the range of severity measured by the scale.

Table 3--1999: Prevalence of food security, food insecurity, and hunger in households with income below 130 percent of the poverty line, by selected characteristics of households
(Based on unadjusted data) See notes to tables.

Appendix A: Households' Response Frequencies for Items in the Food Security Scale

Table A-1--Food security scale item responses, 1995, 1998, and 1999 ¹

¹ Item response frequencies weighted to population totals.

² The actual wording of each item includes explicit reference to resource limitation, e.g., "because (I was/we were) running out of money to buy food," or "because there wasn't enough money for food."

³ Households not responding to item are excluded from the denominator. Households without children are excluded from the denominator of child-referenced items.

Source: Current Population Survey Food Security Supplement, April 1995, August 1998, and April 1999.

Appendix B: Background on the U.S. Food Security Measurement Project

This report of household food security in 1999 is the latest in a series of reports on Measuring Food Security in the United States. The series was inaugurated in September 1997 with the three-volume report Household Food Security in the United States in 1995 (Hamilton et al., 1997a and 1997b; Price et al., 1997). The advance report of findings for 1995-1998 (Bickel, Carlson, and Nord, 1999) was released in July 1999 and a report detailing hunger and food insecurity prevalence by State for the 1996-98 period (Nord, Jemison, and Bickel, 1999) was released in September 1999.

The estimates contained in all of these reports are based on a direct survey measure developed over several years by the U.S. Food Security Measurement Project, an ongoing collaboration among Federal agencies, academic researchers, and both commercial and nonprofit private organizations (Carlson et al., 1999; Olson, 1999.) The measure was developed in response to the National Nutrition Monitoring and Related Research Act of 1990. The Ten-Year Comprehensive Plan developed under the Act specified the following task:

Recommend a standardized mechanism and instrument(s) for defining and obtaining data on the prevalence of "food insecurity" or "food insufficiency" in the U.S. and methodologies that can be used across the NNMRR Program and at State and local levels. 1

Beginning in 1992, USDA staff reviewed the existing research literature focusing on the conceptual basis for measuring the severity of food insecurity and hunger and on the practical problems of developing a survey instrument feasible for use in sample surveys at nation- al as well as State and local levels.

In January 1994, USDA's Food and Nutrition Service (FNS) joined with the U.S. Public Health Service, Centers for Disease Control, and Prevention National Center for Health Statistics (NCHS), in sponsoring a national Conference on Food Security Measurement and Research. The conference brought together leading academic experts and other private researchers with key staff of the concerned Federal agencies. The conference identified the consensus existing among researchers in the field as to the strongest conceptual basis for a national measure of food insecurity and hunger and reached working agreement as to the best operational form for implementing such a measure in national surveys (USDA, 1995).

After extensive cognitive assessment, field testing, and analysis by the U.S. Census Bureau, a food security questionnaire was fielded by the Bureau as a Supplement to the Current Population Survey (CPS) of April 1995. 2 The Food Security Supplement was repeated in September 1996, April 1997, August 1998, and April 1999. Minor modifications to the questionnaire format and screening procedures were made over the first several years and a more substantial revision in screening and format, designed to reduce respondent burden and improve data quality, was introduced with the August 1998 CPS Supplement. However, the con- tent of the "core module" questions upon which the U.S. Food Security Measurement Scale is based has remained constant in all years.

Initial analysis of the 1995 data was undertaken by Abt Associates, Inc., through a cooperative venture with FNS, the interagency working group, and other key researchers involved in developing the questionnaire.

The Abt team used nonlinear factor analysis and other state-of-the-art scaling methods to produce a measure- ment scale for the severity of deprivation in basic food needs as experienced by U.S. households. Extensive testing was carried out to establish the validity and reliability of the scale and its applicability across various household types in the broad national sample (Hamilton et al., 1997a, 1997b). 3 3 The food security scale reported here is based on the Rasch measurement model, an application of maximum likelihood estimation in the family of Item Response Theory models (Wright, 1977, 1983). These statistical measurement models were developed in educational testing, where test items vary systematically in difficulty and the overall score measures the level of difficulty that the tested individual has mastered. In the present application, the severity of food insecurity recently experienced by household members is analogous to the level of test difficulty that an individual has mastered.

Following collection of the September 1996 and April 1997 CPS food security data, FNS awarded a second research contract to Mathematica Policy Research, Inc. (MPR), to reproduce independently the earlier results obtained from the 1995 CPS data, to estimate hunger and food insecurity prevalence for 1996 and 1997, and to assess the stability and robustness of the measure- ment model when applied to the separate data sets. The MPR findings, which will be presented in full in a forthcoming final report, establish the stability of the food-security measure over the 1995-97 period. That is, the relative severity of the items was found to be nearly invariant across years and across major population groups and household types.

In 1998, USDA's Economic Research Service (ERS) assumed sponsorship of the Census Bureau's annual food-security data collection for USDA. ERS has ana- lyzed the 1998 and 1999 data, applying the procedures developed and established for USDA in the Abt and MPR research. The ERS work, which will be presented in a subsequent report, has found continuing stability of the measure in 1998 and 1999.

Appendix C: Comparing Prevalence Rates of Food Insecurity and Hunger Across Years

Due to changes in administration of the Food Security Supplement over the first few years of data collection, prevalence statistics cannot be directly compared between some years. Inter-year comparisons to 1995, 1996, or 1997 require adjustment of data in all years to represent, post hoc, a common methodology. In this appendix, reasons for the inter-year differences are described, and tables of prevalence rates based on separate editing protocols are presented to facilitate comparison both to earlier reports and to anticipated future reports.

The main adjustment required for comparability of prevalence estimates is to apply common screening criteria to each year's data. In collecting the CPS food-security data, an initial screener is applied to higher income households to avoid undue interview burden for households showing little indication of a food adequacy problem. The exact specification of that screener was changed several times in the initial years of data collection. Further, in the 1996 and 1997 surveys a similar screen also was applied to lower income households. In any year, households screened out are deemed to be food secure, based on negative responses to the broad initial screening questions. Analysis of the 1995 data conducted after implementation of the 1996 and 1997 surveys indicated that a small but non-trivial proportion of the low-income households screened out in 1996 and 1997 would have been classified as food insecure if they had been asked the full set of food-security indicator questions. For this reason, screening procedures were again revised in 1998. Under the new protocol, all low-income house- holds (with incomes below 185 percent of the poverty line), as well as higher income households that register indications of food stress on either of two preliminary screening questions, are asked the three least severe core food-security questions (or the first five if there are children in the household.) If any of these less severe items is affirmed, then additional, more severe core questions are asked.

To achieve strict comparability across years, data from 1995, 1998, and 1999 were, in effect, adjusted post hoc to the 1996-1997 screening rules. 1 Households that would have been screened out in any year are deemed to be food secure without reference to their actual responses. This method achieves cross-year comparability, but at the cost of discarding valid information and understating the prevalence of food insecurity. This is apparent in appendix figure C-1, which presents prevalence rates based on both common-screen- adjusted data and on full-sample, or unadjusted, data. For example, the best estimate of food-insecurity prevalence in 1999, using all data available, was 10.1 percent, while the estimate based on the more restrictive screening was 8.7 percent. The original 1995 screening method and the revised method adopted for 1998 and 1999, although different in form, show very little difference in their effect on estimated prevalence of food insecurity. However, the data as collected in 1996 and 1997 are problematic for inter-year comparisons. The screening method used in 1996 exerted a significant downward bias on prevalence estimates; this was reduced somewhat in the 1997 screening, but to an extent which is unknown. Consequently, direct comparisons between the unadjusted data appear to be acceptable only for the years 1995, 1998, and 1999.

The estimated prevalence of hunger is less affected by the screening differences across years because most households with this level of severity of food deprivation screen "in" under any year's protocol. The full- data estimate for 1999 is 3.0 percent, compared with estimated prevalence of 2.8 percent based on the more restrictive common screen.

The revised screening protocol initiated in 1998 was continued in 1999 and is planned for 2000 and future data collections. Thus, food security and hunger prevalence statistics based on the full data as collected will be directly comparable from 1998 onward. The present report provides the bridging comparison. Table 1 and appendix table C-1 present 1999 statistics adjust- ed to be comparable to earlier years' statistics as presented in Household Food Security in the United States 1995-1998: Advance Report. Tables 2 and 3 present 1999 statistics based on full-sample data as collected. These are comparable to unadjusted 1998 statistics and to those for future years. Appendix tables C-2 and C-3 present 1998 statistics based on unadjusted data to complete the baseline of statistics based on the new screening protocol.

Appendix figure C-1 Comparison of prevalences of food insecurity and hunger in U.S. households, 1995-99
Data as collected in 1996 and 1997 are not comparable with 1995, 1998, and 1999.

Source: Calculated by ERS based on Current Population Survey Food Security Supplement data.

Table C-1--1999: Prevalence of food security, food insecurity, and hunger by selected characteristics of Households
(Based on data adjusted for cross-year comparability) See notes to tables

Table C-2--1998: Prevalence of food security, food insecurity, and hunger, by selected characteristics of Households
(Based on unadjusted data) See notes to tables

Table C-3--1998: Prevalence of food security, food insecurity, and hunger in households with income below 130 percent of the poverty line, by selected characteristics of households
(Based on unadjusted data) See notes to tables

References

Andrews, M., G. Bickel, and S. Carlson (1998). "Household Food Security in the United States in 1995: Results from the Food Security Measurement Project" Family Economics and Nutrition Review 11:(1&2):17-28.

Bickel, G., M. Andrews, and S. Carlson (1998). "The Magnitude of Hunger: In a New National Measure of Food Security."_ Topics in Clinical Nutrition 13(4):15- 30.

Bickel, G., S. Carlson, and M. Nord (1999). Household Food Security in the United States 1995- 1998: Advance Report. U.S. Dept. Agr., Food and Nutrition Service, Alexandria, VA. (available: www.fns.usda.gov/oane/MENU/Published/FSP/FSP.htm )

Bickel, G., M. Nord, C. Price, W.L. Hamilton, and J.T. Cook (2000). Guide to Measuring Household Food Security, Revised 2000. U.S. Dept. Agr., Food and Nutrition Service, Alexandria, VA.

Carlson, S.J., M.S. Andrews, and G.W. Bickel (1999). "Measuring Food Insecurity and Hunger in the United States: Development of a National Benchmark Measure and Prevalence Estimates." The Journal of Nutrition 129:510S-516S.

Hamilton, W.L., J.T. Cook, W.W. Thompson, L.F. Buron, E.A. Frongillo, Jr., C.M. Olson, and C.A. Wehler (1997a). "Household Food Security in the United States in 1995: Summary Report of the Food Security Measurement Project." Report prepared for the U.S. Dept. Agr., Food and Consumer Service, Alexandria, VA. (summary available: www.fns.usda.gov/oane/MENU/Published/FSP/FSP.htm ).

Hamilton, W.L., J.T. Cook, W.W. Thompson, L.F. Buron, E.A. Frongillo, Jr., C.M. Olson, and C.A. Wehler (1997b). Household Food Security in the United States in 1995: Technical Report. Report pre- pared for the U.S. Dept. Agr., Food and Consumer Service, Alexandria, VA. (available: www.fns.usda.gov/oane/MENU/Published/FSP/FSP.htm ).

Nord, M., K. Jemison, and G.W. Bickel (1999). Prevalence of Food Insecurity and Hunger by State, 1996-1998, FANRRŒ2, Economic Research Service, U.S. Dept. Agr., September 1999. (available: www.ers.usda.gov/briefing/foodsecurity .)

Olson, C.M. (Ed., 1999). Symposium: Advances in Measuring Food Insecurity and Hunger in the U.S. Sponsored by the American Society for Nutritional Sciences as part of Experimental Biology 98, April 1998, San Francisco, CA. Published as supplement to The Journal of Nutrition 129:504S-528S. (available: www.nutrition.org/content/vol129/issue2/ )

Price, C., W.L. Hamilton, and J.T. Cook (1997). Household Food Security in the United States in 1995: Guide to Implementing the Core Food Security Module. Report prepared for the U.S. Dept. Agr., Food and Consumer Service, Alexandria, VA.

U.S. Dept. Agr., Food and Consumer Service, Office of Analysis and Evaluation (1995). Food Security Measurement and Research Conference: Papers and Proceedings. Alexandria, VA.

Wright, B.D. (1977). Solving Measurement Problems with the Rasch Model. Mesa Psychometric Laboratory, The University of Chicago, College of Education. Chicago, IL. (available: www.rasch.org/memos.htm )

Wright, B.D. (1983). Fundamental Measurement in Social Science and Education. Mesa Psychometric Laboratory, The University of Chicago, College of Education. Chicago, IL. (available: www.rasch.org/memos.htm )