Cautions in Interpreting Findings

Readers should remember the following issues when interpreting the findings in this report:

• The purpose of this report is descriptive; as nonexperimental studies, NLTS and NLTS2 do not provide data that can be used to address causal questions. The descriptions provided in this document concern the post-high school experiences of youth. No attempt is made to "validate" respondents' reports with information on their understanding of the survey items or with third-party information on their experiences (e.g., from employers or postsecondary education institutions).
• The analyses are descriptive; none of the findings should be interpreted as implying causal relationships.
• Information about the nature of students' disabilities came from rosters of all students in the NLTS and NLTS2 age ranges receiving special education services in the 1983–84 or 2000–01 school year (respectively) under the auspices of participating LEAs and statesupported special schools. In analyses in this report, each student is assigned to a disability category on the basis of the primary disability designated by the student's school or district. Although there are federal guidelines in making category assignments, criteria and methods for assigning students to categories vary from state to state and even between districts within states, with the potential for substantial variation in the nature and severity of disabilities included in the categories. Therefore, NLTS and NLTS2 data should not be interpreted as describing students who truly had a particular disability, but rather as describing students who were categorized as having that primary disability.
• Data presented are combined youth self-report and parent-report data. If an NLTS Wave 2 or NLTS2 Wave 3 youth interview/survey was completed, youth's responses to these items were used in this report. In both studies, if a youth interview/survey could not be completed for an eligible youth or if a youth was reported by parents not to be able to participate in an interview/survey, parent responses were used. For the subsample of out-of-high school youth included in this report, the youth interview/survey was the source of data for post-high school outcomes for 84 percent of NLTS youth and for 70 percent of NLTS2 youth, and the parent interview was the source for 16 percent of NLTS youth and 30 percent of NLTS2 youth who did not have a youth interview. Combining data across respondents raises the question of whether parent and youth responses would concur—i.e., would the same findings result if parents' responses were reported instead of youth's responses. When both parents and youth were asked whether the youth belonged to an organized community group, currently worked for pay, worked for pay in the past 2 years, and the wages currently employed youth earned per hour, their responses agreed from 70 percent to 91 percent of the time in NLTS and from 69 percent to 80 percent of the time in NLTS2.
• Differences exist between NLTS and NLTS2 that required analytic adjustments to age, disability category, and household income, for comparisons between the studies to be valid. After these adjustments had been made, differences remained between the NLTS and NLTS2 samples in two of the subgroups included in this report: the other health impairment/autism disability category and the high school completion status variable. Consistent with the increasing number of students identified with autism (Volkmar et al. 2004), the NLTS2 sample included significantly more youth in the other health impairment/autism category than the NLTS sample (6 percent vs. 1 percent, p < .01). In addition, as presented in previous reports comparing the experiences of youth in NLTS with those in NLTS2,⁶ youth in NLTS2 were more likely to have completed high school than those in NLTS (85 percent vs. 70 percent, p < .001).
• It is important to note that descriptive findings are reported for the full sample of out-ofhigh school youth; those findings are heavily influenced by information provided for youth with learning disabilities, who constitute 62 percent of the weighted NLTS sample and 64 percent of the weighted NLTS2 sample. Comparisons also were conducted between groups of youth who differed with respect to disability category, high schoolleaving status and timing, gender, race/ethnicity, and household income. These bivariate analyses should not be interpreted as implying that a factor on which subgroups are differentiated (e.g., disability category) has a causal relationship with the differences reported. Further, readers should be aware that demographic actors (e.g., race/ethnicity and household income) are correlated among youth with disabilities and are distributed differently across disability categories. These complex interactions and relationships among subgroups relative to the variables included in this report have not been explored.
• Extensive efforts were made to ensure the comparability of the two studies and that the wording of most NLTS and NLTS2 survey items are identical. A few items have minor wording differences, which may account for different responses. Survey items are included as chapter footnotes and wording differences are described there.
• Several types of analyses were conducted for this report, including between-group means, between-group percentages, and within-subject percentages. Because of the weighted nature of NLTS2 data, equality between the mean values of the responses to a single survey item in two disjoint subpopulations was based on a test statistic essentially equivalent to a two-sample t test for independent samples using weighted data. Sample sizes for each group being compared were never less than 30. For a two-tailed test, the test statistic was the square of the t statistic, which then followed an approximate chisquare distribution with one degree of freedom, that is, an F (1, infinity) distribution.
• Although discussions in the report emphasize only differences that reach a level of statistical significance of at least p < .01, the large number of comparisons made in this report may result in some significant differences mistakenly determined to be significant when they are not (i.e., "false positives" or type I errors). Readers also are cautioned that the meaningfulness of differences reported here cannot be derived from their statistical significance.

Top