The Field Representative's Manuals for 2001 and 2003-2004 directed interviewers that, when an item included two or more behaviors linked by an "or":

The Manuals for 2001 and 2003-2004 also provided directions on how interviewers should respond if parents indicated that the child was taking medication:

As the Appendix on SDQ in the 2001 and 2003-2004 Codebook of the NHIS public use files explains:

• emotional symptoms;

• conduct problems;

• hyperactive behavior;

• peer relationships;

• prosocial behavior.

The survey forms for 2001 and 2003-2004 also acknowledged the debt to Dr. Goodman, as follows:

This information was included for legal reasons and was not shared with survey respondents. More information on the SDQ is available at www.sdqinfo.org.

The other elements of this subscale on conduct problems are:

• Often loses temper, past 6 months (BADTEMPER)

• Generally well behaved, past 6 months (WELLBEHAVED)

• Often fights or bullies kids, past 6 months (BULLIES)

• Often lies or cheats, past 6 months (LIECHEAT)

Valid responses for these questions were "not true," "somewhat true," and "certainly true." A response of "not true" for BADTEMPER, BULLIES, LIECHEAT, and STEALS (code 0 in IPUMS NHIS) implies the lowest level of conduct problems; a response of "certainly true" for these variables (code 2 in IPUMS NHIS) implies the highest level of conduct problems; and a response of "somewhat true" for these variables (code 1 in IPUMS NHIS) implies an intermediate level of conduct problems. A response of "not true" for WELLBEHAVED (code 2 in IPUMS NHIS) implies the highest level of conduct problems; a response of "certainly true" for this variable (code 0 in IPUMS NHIS) implies the lowest level of conduct problems; and a response of "somewhat true" for this variable (code 1 in IPUMS NHIS) implies an intermediate level of conduct problems. Researchers may choose to use a single variable from this set, but they can also sum the scores across the 5 variables. Summing these elements yields a total score for conduct problems ranging from 0 (the lowest level) to 10 (the highest level).

Summing the values for these variables yields valid totals only if the analyst excludes not in universe cases (persons other than sample children age 4-17, code 6 in IPUMS NHIS) and cases with missing information (codes 7, 8, and 9 in IPUMS NHIS).

The SDQ Scoring Guide provides guidelines for interpreting subscale scores (with Not in Universe and unknown cases excluded) for the SDQ-EX. For "Parent Completed" results (as in the NHIS), the guide suggests a three-band or four-band approach for interpreting the summed score for "Conduct Problems." Under the original three-band categorization, the summed Conduct Problems score can be identified as normal, borderline, or abnormal: a score of 0-2 is normal; a score of 3 is borderline; and a score of 4-10 is abnormal. Under the newer four-band categorization, the summed score can be identified as close to average, slightly raised, high, or very high: a score of 0-2 is close to average; a score of 3 is slightly raised; a score of 4-5 is high; and a score of 6-10 is very high.

Variables from the SDQ receive codes in IPUMS NHIS that facilitate this scoring process. More specifically, the response that corresponds to the lowest likelihood of a psychological problem is coded as 0 in IPUMS NHIS; the response that corresponds to the highest likelihood of a psychological problem is coded 2 in IPUMS NHIS; and the response that corresponds to an intermediate level likelihood of a psychological problem is coded 1 in IPUMS NHIS.

This coding strategy to facilitate scoring means that a given parental response of "not true" or "certainly true" may be sometimes coded as 0 and sometimes coded as 2. (Responses of "somewhat true" are always coded 1.) For example, the question, "During the past 6 months, has [child] often complained of headaches, stomach aches, or sickness?" presupposes that "not true" (coded 0 in IPUMS NHIS) implies less emotional symptomology than a response of "certainly true" (coded 2 in IPUMS NHIS). However, with the question, "During the past 6 months, has [child had] at least one good friend?," a response of "not true" implies poorer peer relationships than a response of "certainly true." In the latter case (for the variable HASFRIEND), responses of "not true" are coded 2 in IPUMS NHIS and responses of "certainly true" are coded 0 in IPUMS NHIS.

Once analysts exclude not in universe cases (persons who are not sample children age 4-17, IPUMS NHIS code 6) and cases with missing information (IPUMS NHIS codes 7, 8, and 9), they may sum the values for the following SDQ-EX variables for an overall score of 0 (least likely to have psychological problems) to 40 (most likely to have psychological problems). In addition to the 5 variables (including STEALS) used to measure "Conduct Problems," as described in the previous section, the variables whose values may summed are:

• Complains of headaches/stomach-aches or sickness, past 6 months (STOMACHE)

• Often seems worried, past 6 months (WORRIED)

• Often unhappy, depressed, or tearful, past 6 months (UNHAPPY)

• Nervous or clingy in new situations, past 6 months (CLINGY)

• Many fears or easily scared, past 6 months (FEARFUL)

• Restless or overactive, past 6 months (OVERACTIVE)

• Constantly fidgeting, past 6 months (FIDGETY)

• Easily distracted, past 6 months (DISTRACTED)

• Thinks before acting, past 6 months (THINKSFIRST)

• Good attention span and finishes tasks, past 6 months (GOODATTEN)

• Prefers to be alone, past 6 months (SOLITARY)

• Had at least 1 good friend, past 6 months (HASFRIEND)

• Liked by other kids, past 6 months (KIDSLIKE)

• Was picked on or bullied, past 6 months (PICKEDON)

• Gets along better with adults than kids, past 6 months (GETALONGAD)

From 1997-2000, the NHIS used questions from the Child Behavioral Checklist (CBCL) developed by Dr. Thomas Achenbach to measure children's emotional and behavioral problems. For 2001 forward, the NHIS retained the CBCL questions for children age 2-3, but replaced the CBCL survey with the Strengths and Difficulties Questionnaire (SDQ) for older children.

For the most part, the public use files of the NHIS data include only summary recodes for the CBCL questions, calculated separately for male and female children age 2-3 (MTODMHI, FTODMHI), for male and female children age 4-11 (MKIDMHI, FKIDMHI), and for male and female children age 12 to 17 (MTEENMHI, FTEENMHI). The one exception to this generalization is that variables from the CBCL relating to whether the child had been unhappy or depressed in recent months (TODDEPRES, KIDDEPRES, TEENDEPRES) appear in the NHIS public files.