QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Connell, C. L. Articles by Yadrick, K. Search for Related Content PubMed PubMed Citation Articles by Connell, C. L. Articles by Yadrick, K.

---

Community and International Nutrition

Food Security of Older Children Can Be Assessed Using a Standardized Survey Instrument¹

Carol L. Connell^*,2, Mark Nord, Kristi L. Lofton^* and Kathy Yadrick^*,**

^* Delta Nutrition Intervention Research Initiative, University of Southern Mississippi, Hattiesburg, MS 39406-5054; Economic Research Service, U.S. Department of Agriculture, Washington, DC 20036-5831; and ^** Department of Nutrition and Food Systems, University of Southern Mississippi, Hattiesburg, MS 39406

²To whom correspondence should be addressed. E-mail: carol.connell{at}usm.edu.

	ABSTRACT

TOP ABSTRACT SUBJECTS AND METHODS RESULTS DISCUSSION LITERATURE CITED

 
Cognitive interviewing methods were used to adapt questions from the U.S. Food Security Survey Module for administration to children. Individual concurrent probing techniques using standardized probes were utilized to assess understanding of the items with 20 African American children (10 males, 10 females, aged 11–13 y). Item wording and response sets were revised, and small groups of boys (n = 5) and girls (n = 14) aged 12–15 y were asked to complete the 9-item survey. Retrospective probing techniques were then used to assess comprehension of items and response sets. Nine items were then piloted in a middle school using a self-administered format. Three hundred forty-five surveys were returned. The majority of the students were between 12 and 15 y (n = 215). Scaling analysis of the 345 completed surveys using statistical methods based on the Rasch measurement model indicated that the module measured a single underlying phenomenon (food insecurity) with sufficient reliability to be a useful tool. The measurable range of food insecurity was about 6 times the estimated measurement error, indicating that the scale could identify 3 categories of food security with reasonable reliability. A survey instrument that reliably measures food security status of individual children can provide researchers with an important tool to assess more accurately the individual-level effects of food security on nutritional status and mental and physical health among this population.

KEY WORDS: • food security • hunger • children • Rasch model • cognitive testing

Data from the December 2002 Current Population Survey Food Security Supplement (CPS-FSS)³ indicate that 16.5% of households with children under 18 y were food insecure some time during the previous 12 mo. Of these households, 3.8% experienced food insecurity with hunger at some time during that same year (1). A growing body of literature reports adverse effects of food insecurity on children’s nutritional status and mental and physical health (2–8). However, the U.S. Food Security Survey Module (US FSSM) administered in the CPS-FSS elicits information on household-level conditions for adults as well as for children from an adult in the household. The food security scales derived from answers to the US FSSM represent the food security status of household members as a group or children in the household as a group, rather than the condition of any particular individual within the household (9). Furthermore, it represents the food security status of children in the household as perceived by an adult. Therefore, limitations exist in identifying associations between food security, measured at the household level, and nutritional status and mental and physical health of children, measured at the individual level. A tool that could measure food security as experienced by individual children would enable researchers to assess more accurately effects of food insecurity on children within a household. Furthermore, a module that can be administered directly to children may make possible the assessment of children’s food security in survey applications when an adult proxy is not practical.

This article reports on cognitive testing and scaling analyses conducted to develop and assess a food security survey module for direct administration to children. The specific aim of the cognitive interviews was to evaluate children’s understanding of the US FSSM items and to modify the items as needed to use language and formatting appropriate for children, in essence to establish face validity. The specific aim of the scaling analysis was to establish the internal validity and reliability of the items in measuring the underlying construct of food security.

	SUBJECTS AND METHODS

TOP ABSTRACT SUBJECTS AND METHODS RESULTS DISCUSSION LITERATURE CITED

 
Cognitive testing

Cognitive testing of survey items prior to administration can improve data quality, analysis, and interpretation by assisting in the development of items that are interpreted consistently by respondents (10–12). We selected 12 items from the US FSSM for cognitive testing (Table 1). Nine of the items, which together represent the range of food security covered by the US FSSM, were chosen based on earlier qualitative interviews regarding children’s descriptions of food insecurity (13). The other 2 items related to the frequency of occurrence of specific behaviors. Items were modified slightly to focus on individual respondents rather than households. Based on advice from an expert in child interviews for marketing research and our own experience conducting dietary interviews with children, 2 further modifications were made to the original US FSSM items. First, response sets were modified for the food security items to "all/most of the time," "some of the time," and "none of the time." Second, a specific event was added to the time frame reference for each item to assist children to "anchor" the previous 12 mo. Since cognitive interviews were held during a summer camp the event phrase added was "since last summer." Cognitive testing included 2 phases, individual concurrent interviews and group retrospective interviews, described below.

    Group retrospective cognitive interviews. The second phase of cognitive testing utilized retrospective probing techniques to assess comprehension of the items and response sets in a group setting similar to a focus group (10). Others proposed that focus group discussions are less intimidating for children than individual interviews and that the group dynamic helps to stimulate thinking and encourage respondents to build on each other’s responses (14). Small same-gender groups of children (n = 4–5/group), recruited from the same sports camp, completed the 9-item survey in a classroom setting. Five AA boys and 14 AA girls 12–15 y participated in these interviews. They completed the survey as a researcher read the items aloud and marked those questions they felt would be confusing to other kids their age. Then surveys were collected and 1 researcher led participants in group discussions of each item using the standard probes while a second researcher recorded notes on the discussions. We could not tape record group discussions due to the large classroom and distance between participants. These interviews resulted in retention of the 9 items, further described under Results.

Pilot survey

We then piloted the 9 items in a local middle school (grades 5–8) with a student population of 467 and a 50% rate of student eligibility for free and reduced-price school meals. Racial/ethnic groups represented in the school were 85.9% Caucasian, 12.6% AA, and 1.5% Hispanic. All students who had not been denied permission by a parent/guardian to participate and who were in attendance on the day of the pilot study were asked to complete the survey during 1 class. In order to reduce disruption to class, teachers distributed the survey to students, who then completed it without assistance. Three hundred forty-five surveys were returned.

Data analysis

    Cognitive testing. For individual interviews, we reviewed field notes, observations of children’s behavior, and tape-recorded transcripts to evaluate children’s interpretation of items and need for wording or format changes. For group interviews, only field notes and observations were used since tape-recording was not feasible.

    Pilot survey. The scaling analysis assessed internal validity and reliability of the set of items for measuring food insecurity by examining the relations among item responses. Data from the 345 completed surveys were entered into SPSS version 11.5 (SPSS) and exported to SAS version 8 (SAS Institute) for scaling analysis. Responses to the food security items were fitted to a single-parameter logistic (Rasch) model using joint maximum likelihood methods implemented by SAS programs developed by USDA’s Economic Research Service (ERS).⁴ The scaling analysis assessed the suitability of each item and the reliability of the set of items for measuring food insecurity.

Each item in the scale is expected to be associated with food insecurity as experienced by the child. The strength and consistency of these relationships, and thus the reliability of the set of items as a measure of food insecurity, were assessed by examining the interrelationships among item responses. The analysis takes into consideration the specific characteristics of items measuring food security, which are dichotomous and each sensitive to a different level of severity of food insecurity. The differences in severity are observed in the response patterns of survey participants. More severe items are less frequently affirmed than less severe items, and a respondent who affirms an item usually affirms most items that are less severe, while a respondent who denies an item usually denies most items that are more severe. The Rasch model assumes a specific mathematical formulation of this severity ordering and provides statistical methods to estimate the relative severity of each item and to assess the extent to which the response patterns observed in the data are consistent with the model assumptions (15–19). Rasch methods are the accepted standards in assessing the performance of food security items and scales (9,20–22).

The association of each item with respondents’ food insecurity was assessed by examining item-infit and item-outfit statistics. These standard Rasch-model statistics compare the observed proportions of affirmative responses by respondents at each level of severity with the proportions expected under model assumptions. The values of both statistics are 1 if responses fit model assumptions perfectly. Values above 1 indicate a disproportionate share of "out-of-order" responses (i.e., affirmative responses by respondents with severity scores below that of the item or denials by respondents with severity scores above that of the item), while values below 1 indicate a smaller proportion of such responses than would be expected. The infit statistic is an information-weighted statistic that is sensitive to responses by households with severity scores in the range near the severity level of the particular item. Outfit is an unweighted squared-error statistic that is sensitive to highly improbable or erratic responses, that is, to unexpected responses from households with severities much higher or lower than that of the item. The Rasch model assumes that all items are related to the measured construct by the same logistic function, so fit-statistic values (especially infit) that are far from unity call into question the suitability of the item for use in the scale. As a general rule, infits in the range of 0.8 to 1.2 are considered good, and 0.7 to 1.3 may be acceptable. Similar standards may be applied to outfit statistics, but in practice outfits are very sensitive to a few highly unexpected observations.

Overall model fit was assessed in four ways: 1) The SD of a subset of the items was compared with the SD of the equivalent adult-referenced items in the US FSSM (9). The SD of equivalent items in the 2 data sources is proportional to the mean item discrimination in the 2 data sources. Mean item discrimination is a measure of how strongly the items are related to food insecurity. 2) Reliability of respondent measurement was calculated using standard Rasch methodology, which calculates reliability as the proportion of variance in respondent scores that is not due to measurement error. 3) The ratio of the measurable range of respondent scores to the measurement error was divided by 1.96 to approximate the number of categories of severity that could be meaningfully distinguished by the measure. 4) Classification accuracy of a scale based on the items in the Child Food Security Survey Module (CFSSM) was assessed by calculating standard measures of sensitivity, specificity, and positive predictive value at 2 thresholds: food insecure and food insecure with hunger. The calculation of classification accuracy assumed a hypothetical population with true food insecurity distributed approximately as observed among survey respondents. Classification of respondents based on their true level of food insecurity in this hypothetical distribution was taken as the "gold standard" against which their classification based on their measured food insecurity was compared. Taking into account respondent measurement error as estimated under Rasch-model assumptions, the proportions of respondents at each level of severity that would be correctly and incorrectly classified were then calculated. Aggregated across all levels of severity, these proportions were then used to calculate the proportions of false positives, false negatives, and the standard measures of classification accuracy.

The comparability of food insecurity as measured by the CFSSM with that measured by the US FSSM as administered to adults was assessed by comparing severity scores of items estimated from the child food security pilot data with severity scores of equivalent adult items in the US FSSM. Equal levels of severity of food insecurity on the 2 scales were identified by adjusting the metric of the child food security scale to that of the U.S. food security scale to equate the mean and SD of the equivalent items. Thresholds for food insecurity and food insecurity with hunger were specified based on corresponding thresholds in the US FSSM, and the prevalence of food security, food insecurity, and food insecurity with hunger was estimated for the child pilot sample.

Measurement models were estimated separately for boys and girls as well as for younger (<12 y) and older (12 y) children. Overall model fit, item-fit statistics, the relative severity of item scores, and the prevalence of food insecurity and food insecurity with hunger were compared between the 2 groups in each analysis. The research was approved by the Institutional Review Board at the University of Southern Mississippi.

	RESULTS

TOP ABSTRACT SUBJECTS AND METHODS RESULTS DISCUSSION LITERATURE CITED

 
Cognitive testing

    Concurrent cognitive interviews. When asked "what were they asking in that question?" for each item, most respondents stated the items meant running out of food because there was no money. However, most children had to reread each item several times in order to respond to this probe. When asked "was there anything you think kids your age would find confusing or hard to understand," a few children responded with "the commas." When asked "how would you ask this question if you had my job?" a few children suggested "take out the commas." Several children restated the items in much shorter phrases. Two survey items, eating a few kinds of low-cost foods and not being able to eat a balanced meal, generated suggestions for revision to make them easier to understand. The word "cheap" was suggested as a replacement for "low-cost." Suggestions for replacing the term "balanced meal" were not consistent. A few children suggested "full meal" whereas others suggested "good meal." However, we chose to retain both phrases for further testing during the retrospective cognitive interviews.

Children also found the response sets to be problematic. They were not able to distinguish between "all/most of the time" and "some of the time." When the word "most" was removed and the response restated as "all of the time" children said there was a definite difference between it and "some of the time." For example, 1 boy stated, "some is half of all." However, investigators felt that this significantly changed the meaning of the question because some children said "all of the time" meant "constantly." We modified the response set to 3 more simple choices, "a lot, sometimes, and none," for further cognitive testing.

A similar finding surfaced for the response sets to the 3 frequency follow-up items. Respondents had difficulty distinguishing between "almost every month" and "some months." When asked how the item could be written to make it easier to answer, most simply said "I don’t know" or "just ask them how much it happened." Therefore these items were dropped from further testing and the response set from above was used for all items.

When asked whether there was anything about each item that would make it embarrassing for others to answer, children noted that running out of money or food would be embarrassing. When asked their opinion about the best way to administer a similar survey to other children, participants indicated that an anonymous survey or private interview would be best. As a result of these interviews, item wording was simplified, and item format was shortened to a single question with a response set consisting of 3 simple choices put into the structure of a multiple-choice test.

    Group retrospective cognitive interviews. During retrospective cognitive interviews, children indicated that they understood the intent of the items and the response sets were appropriate. However, the "few kinds of low-cost foods" item again generated wording change suggestions and as a result we changed the term "low-cost" to "cheap." Children indicated that the items seemed to be asking the same thing over and over. However, some pointed out there were minor differences in the items that made each unique. Therefore, modifications were made to improve the emphasis of the range of severity of food insecurity covered by the set of items. Boldface type was used on specific words in each item to help respondents see that the items were not redundant (Table 1). The resource constraint phrase, "didn’t have enough money for food" or "because your family was running out of money... " evoked unease and discussion among participants and also during concurrent cognitive interviews. Children indicated that the phrase made the items embarrassing to answer. However, a few participants pointed out that some people go hungry by choice and therefore the phrase was necessary. Participants stated that repetition of the time anchor phrase at the beginning of each item was annoying and made them feel they were not "smart enough" to remember it after it was stated in the introduction. In addition, they expressed difficulty recalling information from 12 mo prior even though the anchor for the time frame was specific. They recommended asking only about the past month and only in the introduction to the survey.

Again, children stated that these items would best be self-administered in a confidential setting where others could not hear or see the answers. A few mentioned being particularly suspicious if this survey was administered over the telephone. In 2 separate interview groups, children stated others would "hang up on you because they’ll think you’re a bill collector." Therefore, items were reworded in a manner that would be less offensive and not assume food insecurity. For example, "how often did the food that your family bought run out and you didn’t have money to get more" became "did the food your family bought run out and you didn’t have money to get more?" In addition, the response "none" was changed to "never" (Table 1).

Pilot survey

Missing responses were rare in the 345 returned middle-school surveys. Only 8 children (2.3%) missed any responses, and 7 of those missed only a single item. No item was missed by >3 respondents (0.9%). Responses to each item were dichotomized as affirmative (responses of "a lot" or "sometimes") and negative (responses of "never"). The expected difference in the severity of items was reflected in the proportions of respondents that affirmed each item (Table 2).

Mean item discrimination, as measured by the SD of items that may be considered equivalent between the CFSSM and adult items in the US FSSM (Q1, Q4, Q5, Q8, and Q9), was 51% lower for the CFSSM than for the US FSSM (analysis not shown). Reliability based on the observed distribution of respondents across raw score groups was 0.48, meaning that just over half of the observed variance in respondent scores within the scaling sample was attributable to measurement error. Reliability based on a hypothetical distribution with equal numbers of respondents in each score group was somewhat higher at 0.69. Despite the modest reliability, the measurable range of respondent scores was about 6 times the mean measurement error (analysis not shown), suggesting that 3 categories of food insecurity can be meaningfully distinguished by the measure. Simulation of classification accuracy based on a population with "true" food insecurity distributed approximately as observed in the sample gave the following results: at the food-insecure threshold (raw score 2, discussed below), sensitivity was 0.89, specificity 0.93, and positive predictive value 0.82; at the food-insecure-with-hunger threshold (raw score 6, discussed below), sensitivity was 0.77, specificity 0.98, and positive predictive value 0.71.

The 5 equivalent items in the CFSSM and US FSSM are in the same severity order in both surveys and only 1 item differs substantially in severity between the 2 surveys after adjusting for the difference in mean item discrimination. (This analysis is not shown, but details are available on request from the authors.) Q4 is more severe in the CFSSM (P = 0.036), meaning that children are less likely to report this condition than adults with similar responses to the other equivalent items.

Food security scale scores and food security status of respondents with each raw score (Table 3) were calculated from the item scores in Table 2 after adjusting them to the metric of the US Food Security Scale. They are, therefore, directly comparable with the standard household scale scores based on the US FSSM, Exhibit C-2 "Standard Computational Metric" column (9). That is, equal scores on the two scales represent the same level of severity of food insecurity. The following ranges of food security are proposed (discussed below): food secure (raw score 0 and 1), food insecure without hunger (raw score 2–5), and food insecure with hunger (raw score 6 or higher). Based on these ranges, prevalence rates for food security, food insecurity, and food insecurity with hunger were estimated for the pilot survey sample (Table 4).

	DISCUSSION

TOP ABSTRACT SUBJECTS AND METHODS RESULTS DISCUSSION LITERATURE CITED

 
The first aim of this study was to develop a food security survey module with face validity for children using cognitive testing methodology to evaluate and revise the US FSSM items. Concurrent probing permitted respondents’ interpretation of the items to be captured immediately after the survey item was read. It also allowed observation of important behaviors such as rereading the item several times before answering; something that would have been missed otherwise. Since some children were not as articulate in explaining their interpretation of an item, combining observations of their behavior with their responses helped us improve question wording and formatting.

While the children were able to interpret the US FSSM items, the length of the items and the punctuation caused the respondents to hesitate in their answers. When given the opportunity to restate the US FSSM items in their own words, children consistently paraphrased the items into much shorter sentences. Harrison et al. (23) reported that low-income Spanish-speaking adults preferred simpler versions of a Spanish-language household food security survey module. We shortened the items into the form of a question and changed the response sets for the items to retain the intent of the original US FSSM items, but used language more children recognized as representing the concepts of "all/most of the time" and "none of the time." The results were items in a format that most children are familiar with, a multiple-choice test, and efforts to read and answer the items were improved.

Retrospective probing in a group discussion format provided a setting that allowed children to answer the items unassisted, similar to what might be encountered in a self-administered survey situation. In addition, children appeared to be more comfortable when asked probing questions in a group format that was similar to focus groups. These interviews revealed that even the simplified items were viewed as redundant by the children and that repeating the time anchor phrase was offensive and unnecessary in a self-administered survey. In addition, group interviews re-enforced the sensitive nature of the items and the need for privacy and anonymity when interviewing children about food security. We used these findings to further simplify and improve the questions and to reduce respondent burden associated with reporting food insecurity. The rarity of missing responses in the pilot survey is at least weak evidence that the items were generally comprehensible and not considered unreasonably offensive.

In addition to assessing face validity of the items in the CFSSM, the second aim of this study was to assess the internal validity and reliability of a scaled measure of food insecurity based on the items. Individually, all items fit Rasch model assumptions reasonably well, indicating that the strengths of the associations of all the items with the underlying construct of food insecurity were similar. Only Q3 had outfit statistics indicating a somewhat higher proportion of erratic responses than expected. This, along with a slightly high infit, may indicate that the item is not consistently understood or that the experience is less consistently associated with food insecurity than is true for the other items. Since the findings from the cognitive interviews also suggested possible comprehension problems with this item, further testing of the item may be prudent. In particular, placing the wording emphasis (boldface) on the phrase "few kinds" in that item may better convey its original intent.

Overall model fit was not as good as that of adult response data from the US FSSM. Lower mean item discrimination (poorer model fit) is reflected in the relatively low reliability as calculated from respondent scores. In practical terms, low item discrimination reflects less consistent ordering of responses. This may reflect less consistent understanding of the items or less consistent ordering of the behaviors indicated by the items. The lower than expected discrimination in this sample may have resulted in part from self-administration of the pilot survey. Lower discrimination in the CFSSM may also result from children’s experiences of specific effects of household food insecurity being, in fact, less consistent than that of adults. In most cases, adults manage the household food resources while children are affected by the decisions of adults but may have less influence on the management of food resources.

Lower mean item discrimination reduces the precision of the measure somewhat. Nevertheless, the measurable range of food insecurity was about 6 times the estimated measurement error, indicating that the scale can identify 3 categories of food security with reasonable reliability. The simulation of classification accuracy indicates that the CFSSM has better sensitivity at the food-insecure (with or without hunger) range than at the food-insecure with hunger range (0.89 vs. 0.77). Specificity is modest at both levels of food insecurity (0.93 and 0.98, respectively). Prevalences of food insecurity as measured by the CFSSM will likely be biased upward somewhat in typical U.S. populations, while prevalences of food insecurity with hunger will be estimated with relatively small bias.

The similarity of the relative severity of items in the CFSSM and the US FSSM suggests measured levels of severity of food insecurity can be compared between respondents to the 2 modules. Respondent scale scores for the CFSSM indicates, as nearly as possible, the same food insecurity conditions as those of households with the same scores based on the US FSSM (Table 3). To be consistent with the adult scale based on the US FSSM, only children with raw score 3 on the CFSSM should be classified as food insecure. However, we suggest specifying as food insecure the range of raw score 2 on the child scale. Based on the cognitive content of the items and the conceptual definition of food insecurity (24), we consider it inappropriate to classify as food secure those children who have affirmed 2 items, typically Q1 and either Q3 or Q2.

For the range food insecure with hunger, a raw score 6 is consistent with the adult scale based on the US FSSM and with our assessment of the cognitive content of the items. The item score for Q8, after adjusting to the metric of the US Food Security Scale, was about midway between the respondent scale scores for raw score = 5 and raw score = 6. As expected, based on this relation, a majority (60%) of respondents with raw score = 5 denied Q8, while a majority (82%) of those with raw score = 6 affirmed Q8.

The CFSSM appears to have worked equally well for both sexes. The sizes of the scaling samples when disaggregated by gender were, however, quite small (69 boys, 73 girls). Therefore, results should be considered preliminary and the scaling analysis repeated if the module is implemented in a larger survey. Response patterns were significantly different between children < 12 y and those 12 y. Samples sizes for these analyses were also small. However, the less consistently ordered responses of the younger children taken together with results of cognitive testing in which younger children had more difficulty with the items suggest caution in administering this module to younger children.

These results, obtained in the linguistic and cultural environment of south Mississippi, indicate that the items in the CFSSM measure the severity of food insecurity of children, especially children 12 y, with sufficient reliability to be a useful tool. More research is needed in a nationally representative sample to confirm our results. In addition, future research should focus on other aspects of validity because face validity and internal validity are only 2 facets of validation criteria. A survey instrument that could reliably measure food security status of individual children would provide researchers with an important tool to assess more accurately the individual-level effects of food security on nutritional status and mental and physical health among this population. In addition, a module that can be administered directly to children would make possible the assessment of children’s food security in survey applications where an adult proxy is not practical.

	FOOTNOTES

 
¹ Supported in part by ERS/USDA cooperative agreement no. 43–3AEM-2–80033 and ARS/USDA cooperative agreement no. 6251–52000-0020–000.

³ Abbreviations used: CFSSM, Child Food Security Survey Module; CPS-FSS, Current Population Survey-Food Security Supplement; ERS, Economic Research Service; NYSP, National Youth Sports Program; Q, question; US FSSM, U.S. Food Security Survey Module; AA, African American.

⁴ ERS SAS programs (ERSRasch) were developed specifically to facilitate Rasch-based statistical analysis of food security response data. They implement joint maximum likelihood methods to estimate Rasch parameters and fit statistics based on the same formulae used by Winsteps (MESA Psychometric Laboratory, University of Chicago, Chicago, IL).

Manuscript received 19 May 2004. Initial review completed 28 June 2004. Revision accepted 19 July 2004.

	LITERATURE CITED

TOP ABSTRACT SUBJECTS AND METHODS RESULTS DISCUSSION LITERATURE CITED

 

1. Nord, M., Andrews, M. & Carlson, S. (2003) Household food security in the United States, 2002. Food Assistance and Nutrition Research Report Number 35 2003 United States Department of Agriculture, Economic Research Service Washington, DC.

2. Casey, P., Goolsby, S., Berkowitz, C., Frank, D., Cook, J., Cutts, D., Black, M. M., Zaldivar, N., Levenson, S., Heeren, T. & Meyers, A., Children’s Sentinel Nutritional Assessment Program Study Group (2004) Maternal depression, changing public assistance, food security and child health status. Pediatrics 113:298-304.[Abstract/Free Full Text]

3. Alaimo, K., Olson, C. M. & Frongillo, E. A. (2002) Family food insufficiency, but not low family income, is positively associated with dysthymia and suicide symptoms in adolescents. J. Nutr. 132:719-725.[Abstract/Free Full Text]

4. Casey, P. H., Szeto, K., Lensing, S., Bogle, M. & Weber, J. (2001) Children in food-insufficient, low-income families. Prevalence, health, and nutrition status. Arch. Pediatr. Adolesc. Med. 155:508-514.[Abstract/Free Full Text]

5. Alaimo, K., Olson, C. M. & Frongillo, E. A. (2001) Food insufficiency and American school-aged children’s cognitive, academic, and psychosocial development. Pediatrics 108:44-53.[Abstract/Free Full Text]

6. Cutts, D. B., Pheley, A. M. & Geppert, J. S. (1998) Hunger in Midwestern inner-city young children. Arch. Pediatr. Adolesc. Med. 152:489-493.[Abstract/Free Full Text]

7. Kleinman, R. E., Murphy, J. M., Little, M., Pagano, M., Wehler, C., Regal, K. & Jellinek, M. S. (1998) Hunger in children in the United States: potential behavioral and emotional correlates. Pediatrics 101(1):e3.[Abstract/Free Full Text]

8. Murphy, J. M., Wehler, C. A., Pagano, M. E., Little, M., Kleinman, R. E. & Jellinek, M. S. (1998) Relationship between hunger and psychosocial functioning in low-income American children. J. Am. Acad. Child Adolesc. Psychiatry 37:163-170.[Medline]

9. Bickel, G., Nord, M., Price, C., Hamilton, W. L. & Cook, J. T. (2000) Guide to Measuring Household Food Security, Revised 2000 2000 Office of Analysis, Nutrition, and Evaluation, Food and Nutrition Service, U.S. Department of Agriculture Alexandria, VA.

10. Willis, G. B. (1994) Cognitive Interviewing and Questionnaire Design: A Training Manual. Cognitive Methods Staff Working Paper Series No. 7 1994 Centers for Disease Control and Prevention, National Center for Health Statistics Hyattsville, MD.

11. Alaimo, K., Olson, C.M. & Frongillo, E.A. (1999) Importance of cognitive testing for survey items: an example from food security questions. J. Nutr. Educ. 31:269-275.

12. Collins, D. (2003) Pretesting survey instruments: an overview of cognitive methods. Quality Life Res. 12:229-238.

13. Connell, C. L., Lofton, K. L., Yadrick, K. & Rehner, T. A. (2003) Perceptions of food insecurity among low income southern children and youth. J. Am. Diet. Assoc. 103:A-82.

14. Lytle, L. A., Eldridge, A. L., Kotz, K., Piper, J., Williams, S. & Kalina, B. (1997) Children’s interpretation of nutrition messages. J. Nutr. Educ. 29:128-136.

15. Wright, B. D. (1977) Solving Measurement Problems with the Rasch Model 1977 Mesa Psychometric Laboratory, the University of Chicago, College of Education Chicago, IL.

16. Wright, B. D. (1983) Fundamental Measurement in Social Science and Education 1983 Mesa Psychometric Laboratory, the University of Chicago, College of Education Chicago, IL.

17. Fischer, G. H. Molenaar, I. W. eds. Rasch Models: Foundations, Recent Developments, and Applications 1995 Springer-Verlag New York, NY. .

18. Baker, F. B. (1992) Item Response Theory: Parameter Estimation Techniques 1992 Marcel Dekker, Inc. New York, NY.

19. Hambleton, R. K., Swaminathan, H. & Rogers, H. J. (1991) Fundamentals of Item Response Theory 1991 Sage Publications, Inc. Newbury Park, CA.

20. Hamilton, W. L., Cook, J. T., Thompson, W. W., Buron, L. F., Frongillo, E. A., Jr, Olson, C. M. & Wehler, C. A. (1997) Household Food Security in the United States in 1995: Technical Report 1997 Office of Analysis, Nutrition, and Evaluation, Food and Nutrition Service, United States Department of Agriculture Washington D.C.

21. Nord, M. & Bickel, G. (2002) Measuring Children’s Food Security in U.S. Households, 1995–99. Food Assistance and Nutrition Research Report Number 25 2002 United States Department of Agriculture, Economic Research Service Washington, D.C.

22. Nord, M. (2002) A 30-Day Food Security Scale for Current Population Survey Food Security Supplement Data. E-FAN No. 02015 2002:3-17-0004 United States Department of Agriculture, Economic Research Service .

23. Harrison, G. G., Stormer, A., Herman, D. R. & Winham, D. M. (2003) Development of a Spanish-language version of the U.S. Household Food Security Survey Module. J. Nutr. 133:1192-1197.[Abstract/Free Full Text]

24. Anderson, S. A. (1990) Core indicators of nutritional state for difficult-to-sample populations. J. Nutr. 120:1557S-1600S.

This article has been cited by other articles:

				M. Hackett, H. Melgar-Quinonez, R. Perez-Escamilla, and A. M. Segall-Correa Gender of respondent does not affect the psychometric properties of the Brazilian Household Food Security Scale Int. J. Epidemiol., August 1, 2008; 37(4): 766 - 774. [Abstract] [Full Text] [PDF]

				C. Gundersen, B. J. Lohman, J. C. Eisenmann, S. Garasky, and S. D. Stewart Child-Specific Food Insecurity and Overweight Are Not Associated in a Sample of 10- to 15-Year-Old Low-Income Youth J. Nutr., February 1, 2008; 138(2): 371 - 378. [Abstract] [Full Text] [PDF]

				M. Nord and H. Hopwood Recent Advances Provide Improved Tools for Measuring Children's Food Security J. Nutr., March 1, 2007; 137(3): 533 - 536. [Abstract] [Full Text] [PDF]

				C. L. Connell, K. L. Lofton, K. Yadrick, and T. A. Rehner Children's Experiences of Food Insecurity Can Assist in Understanding Its Effect on Their Well-Being J. Nutr., July 1, 2005; 135(7): 1683 - 1690. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Connell, C. L. Articles by Yadrick, K. Search for Related Content PubMed PubMed Citation Articles by Connell, C. L. Articles by Yadrick, K.