Nonresponse Bias Examination
Wave 3 nonresponse bias examination
Non-response bias is usually small when the response rate is high or moderate, and weight adjustments can eliminate most of the bias. In Wave 3, PEELS was fortunate to have moderately high response rates, reducing the overall level of concern about response bias. Table 1 presents the Wave 1, Wave 2, and Wave 3 response rates for PEELS. The response rate for the teacher questionnaires is below 80 percent in Wave 1, which is a concern. However, it has been improved in later waves.
To explore the possibility of response bias in the Wave 3 collection, Westat generated tables showing the number and percentage of responses by child’s race/ethnicity and household income. Those data are presented in tables 2 and 3.
Response rates suggest some variation by child’s race/ethnicity. On the parent interview, response rates ranged from 82 percent for parents of Black or African American children to 93 percent for parents of non-Hispanic White children. On the child assessment, response rates ranged from 89 percent for American Indians to 97 percent for Asians, Native Hawaiians or other Pacific Islanders. Finally, on the teacher questionnaires, rates ranged from 76 percent of Black children’s teachers to 85 percent of non-Hispanic White children’s teachers (see table 2).
The data on response rates suggest an association between household income and response probability on the parent interview, with lower income families less likely to complete interviews. The response rates for all income groups were between 83 and 95 percent. Response rates for the child assessment and teacher questionnaires showed little variation by household income (see table 3).
This level of variation does not present a serious threat of nonresponse bias.
Table 1. PEELS Response rates, by wave and instrument
|
Wave 1 |
Wave 2 |
Wave 3a |
SEA questionnaire |
100% |
-- |
-- |
LEA questionnaire |
87% |
-- |
-- |
Parent interview |
96% |
93% |
88% |
Child assessment |
96% |
94% |
93% |
Teacher questionnaire |
76% |
86% |
81% |
Principal/program director questionnaire |
76% |
-- |
-- |
a Wave 3 response rates are subject to change until the data have been finalized.
SOURCE: PEELS Study Management System. Previously unpublished tabulations, August 2006.
Table 2. Number and percentage of PEELS Wave 3 respondents, by race/ethnicity
|
Total Sample |
Parent |
Assessment |
Teacher |
|||
n |
n |
% |
n |
% |
n |
% |
|
Total |
3,098 |
2,719 |
87.8 |
2,889 |
93.3 |
2,509 |
81.0 |
American Indian or Alaskan Native |
19 |
17 |
89.5 |
17 |
89.5 |
15 |
78.9 |
Asian, Native Hawaiian or Other Pacific Islander |
59 |
52 |
88.1 |
57 |
96.6 |
49 |
83.1 |
Black or African American |
277 |
228 |
82.3 |
258 |
93.1 |
211 |
76.2 |
Hispanic |
557 |
468 |
84.0 |
526 |
94.4 |
439 |
78.8 |
White, Non-Hispanic |
1,730 |
1,604 |
92.7 |
1,653 |
95.5 |
1,467 |
84.8 |
More than one race |
161 |
138 |
85.7 |
145 |
90.1 |
129 |
80.1 |
|
|
|
|
|
|
|
|
Missing |
295 |
212 |
71.9 |
233 |
79.0 |
199 |
67.5 |
NOTE: Study withdrawals manually added to missing data line, column 1.
SOURCE: PEELS Study Management System. Previously unpublished tabulations, August 2006.
Table 3. Number and percentage of PEELS Wave 3 respondents, by household income
|
Total Sample |
Parent |
Assessment |
Teacher |
|||
n |
n |
% |
n |
% |
n |
% |
|
Total |
3,098 |
2,719 |
87.8 |
2,889 |
93.3 |
2,509 |
81.0 |
Less than $20,000 |
723 |
599 |
82.8 |
677 |
93.6 |
574 |
79.4 |
$20,001 to $30,000 |
424 |
365 |
86.1 |
399 |
94.1 |
354 |
83.5 |
$30,001 to $40,000 |
359 |
330 |
91.9 |
341 |
95.0 |
296 |
82.5 |
$40,001 to $50,000 |
306 |
283 |
92.5 |
293 |
95.8 |
249 |
81.4 |
More than $50,000 |
960 |
913 |
95.1 |
918 |
95.6 |
815 |
84.9 |
Missing |
326 |
229 |
70.2 |
261 |
80.1 |
221 |
67.8 |
NOTE: Study withdrawals manually added to missing data line, column 1.
SOURCE: PEELS Study Management System. Previously unpublished tabulations, August 2006.
Wave 4 nonresponse bias examination
Non-response bias is usually small when the response rate is high or moderate, and weight adjustments can eliminate most of the bias. In Wave 4, PEELS continued to have moderately high response rates, reducing the overall level of concern about response bias. Table 1 presents the Wave 1, Wave 2, Wave 3, and Wave 4 response rates for PEELS. Lower response rates in Wave 4 may be partially attributed to efforts to reduce the Wave 4 data collection budget.
To explore the possibility of response bias in the Wave 4 collection, Westat generated tables showing the number and percentage of responses by child’s race/ethnicity and household income. Those data are presented in tables 2 and 3.
Response rates suggest some variation by child’s race/ethnicity. On the parent interview, response rates ranged from 70 percent for parents of Black or African American children to 90 percent for parents of American Indians or Alaskan Natives. On the child assessment, response rates ranged from 79 percent for American Indians and Alaskan Natives to 90 percent for non-Hispanic Whites. Finally, on the teacher questionnaires, rates ranged from 75 percent of Black children’s teachers to 87 percent of non-Hispanic White children’s teachers (see table 2).
The data on response rates suggest an association between household income and response probability on the parent interview, with lower income families less likely to complete interviews. The response rates for all income groups were between 74 and 93 percent. Response rates for the child assessment and teacher questionnaires showed little variation by household income (see table 3).
This level of variation does not present a serious threat of nonresponse bias.
Table 1. PEELS Response rates, by wave and instrument
|
Wave 1 |
Wave 2 |
Wave 3 |
Wave 4a |
SEA questionnaire |
100% |
-- |
-- |
-- |
LEA questionnaire |
87% |
-- |
-- |
-- |
Parent interview |
96% |
93% |
88% |
80% |
Child assessment |
96% |
94% |
93% |
85% |
Teacher questionnaire |
76% |
86% |
81% |
81% |
Principal/program director questionnaire |
76% |
-- |
-- |
-- |
a Wave 4 response rates are subject to change until the data have been finalized.
SOURCE: PEELS Study Management System. Previously unpublished tabulations, July 2007.
Table 2. Number and percentage of PEELS Wave 4 respondents, by race/ethnicity
|
Total Sample |
|
Parent |
|
Assessment |
|
Teacher |
||||
|
n |
|
n |
% |
|
n |
% |
|
n |
% |
|
Total |
3,104 |
|
2,488 |
80.2 |
|
2,632 |
84.8 |
|
2,502 |
80.6 |
|
American Indian or Alaskan Native |
19 |
|
17 |
89.5 |
|
15 |
78.9 |
|
15 |
78.9 |
|
Asian, Native Hawaiian or Other Pacific Islander |
58 |
|
50 |
86.2 |
|
48 |
82.8 |
|
50 |
86.2 |
|
Black or African American |
276 |
|
193 |
69.9 |
|
224 |
81.2 |
|
207 |
75.0 |
|
Hispanic |
545 |
|
418 |
76.7 |
|
480 |
88.1 |
|
435 |
79.8 |
|
White, Non-Hispanic |
1,702 |
|
1,490 |
87.5 |
|
1,528 |
89.8 |
|
1,472 |
86.5 |
|
More than one race |
160 |
|
127 |
79.4 |
|
131 |
81.9 |
|
127 |
79.4 |
|
Missing |
344 |
|
193 |
57.1 |
|
206 |
59.9 |
|
196 |
57.0 |
NOTE: Study withdrawals manually added to missing data line, column 1.
SOURCE: PEELS Study Management System. Previously unpublished tabulations, July 2007.
Table 3. Number and percentage of PEELS Wave 4 respondents, by household income
|
Total Sample |
|
Parent |
|
Assessment |
|
Teacher |
|||
|
n |
|
n |
% |
|
n |
% |
|
n |
% |
Total |
3,104 |
|
2,488 |
80.2 |
|
2,632 |
84.8 |
|
2,502 |
80.6 |
Less than $20,000 |
720 |
|
531 |
73.8 |
|
621 |
86.3 |
|
580 |
80.6 |
$20,001 to $30,000 |
412 |
|
327 |
79.4 |
|
358 |
86.9 |
|
331 |
80.3 |
$30,001 to $40,000 |
357 |
|
287 |
80.4 |
|
308 |
86.3 |
|
287 |
80.4 |
$40,001 to $50,000 |
300 |
|
256 |
85.3 |
|
265 |
88.3 |
|
260 |
86.7 |
More than $50,000 |
948 |
|
879 |
92.7 |
|
852 |
89.9 |
|
828 |
87.3 |
Missing |
367 |
|
208 |
56.7 |
|
228 |
62.1 |
|
216 |
58.9 |
NOTE: Study withdrawals manually added to missing data line, column 1.
SOURCE: PEELS Study Management System. Previously unpublished tabulations, July 2007.
File Type | application/msword |
File Title | Nonresponse Bias Examination |
Author | celia.rosenquist |
Last Modified By | #Administrator |
File Modified | 2008-10-22 |
File Created | 2008-10-22 |