05.07.2013 Views

assessing the transportation needs of welfare-to-work participants in ...

assessing the transportation needs of welfare-to-work participants in ...

assessing the transportation needs of welfare-to-work participants in ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Ano<strong>the</strong>r limitation <strong>in</strong>troduced by <strong>the</strong> use <strong>of</strong> supplemental <strong>in</strong>formation was that <strong>the</strong> GEARS<br />

database did not provide a case name. The case name used for surveys was derived from <strong>the</strong><br />

FOCUS database ma<strong>in</strong>ta<strong>in</strong>ed by DPSS, which <strong>in</strong>dicates <strong>the</strong> <strong>of</strong>ficial person designated <strong>to</strong> conduct<br />

<strong>the</strong> bus<strong>in</strong>ess <strong>of</strong> <strong>the</strong> <strong>welfare</strong> case with DPSS. The case name used for survey<strong>in</strong>g purposes was<br />

derived from a different source than <strong>the</strong> case phone numbers. Therefore, some sampled cases<br />

with phone numbers did not have a case name and vise versa. Ano<strong>the</strong>r issue regard<strong>in</strong>g case<br />

names <strong>in</strong>volved <strong>the</strong> disproportionate presence <strong>of</strong> female case names, even for two-parent (U)<br />

cases <strong>in</strong> which a male is usually present <strong>in</strong> <strong>the</strong> household. Therefore, when <strong>in</strong>terviewers were<br />

contact<strong>in</strong>g two-parent (U) families, <strong>the</strong>y were disproportionately survey<strong>in</strong>g <strong>the</strong> woman<br />

associated with <strong>the</strong> case. Although <strong>the</strong> oldest person on cases sampled was approximately 15%<br />

male (based on FOCUS data), <strong>the</strong> <strong>in</strong>itial male response rate was considerably lower.<br />

Adjustments were made <strong>to</strong> <strong>the</strong> survey<strong>in</strong>g process <strong>to</strong> compensate for this limitation (described <strong>in</strong><br />

<strong>the</strong> “Survey Implementation” section below).<br />

To respond <strong>to</strong> <strong>the</strong> low male response rate <strong>in</strong>troduced by <strong>the</strong> fact that case names are<br />

disproportionately female, 100 supplemental <strong>in</strong>terviews <strong>of</strong> two-parent (U aid type) cases were<br />

conducted <strong>to</strong> <strong>in</strong>crease <strong>the</strong> male response rate. The UCLA Lewis Center sampled additional cases<br />

and funded <strong>the</strong> supplemental <strong>in</strong>terviews, which targeted two-parent (U) cases because males are<br />

most prevalent <strong>in</strong> this type <strong>of</strong> <strong>welfare</strong> case. The process <strong>of</strong> sampl<strong>in</strong>g additional two-parent (U)<br />

cases for <strong>the</strong>se <strong>in</strong>terviews resulted <strong>in</strong> an overall CTNA sample that conta<strong>in</strong>s a disproportionately<br />

large number <strong>of</strong> two-parent (U) cases (described <strong>in</strong> more detail below). Also, <strong>the</strong> sample drawn<br />

for <strong>the</strong>se additional 100 supplemental <strong>in</strong>terviews were not geocoded and assigned a supervisorial<br />

district s<strong>in</strong>ce <strong>the</strong> first wave <strong>of</strong> 1545 <strong>in</strong>terview adequately represented all districts.<br />

Description <strong>of</strong> Stratified Random Sample<br />

The follow<strong>in</strong>g describes <strong>the</strong> randomly sampled GAIN cases with regard <strong>to</strong> <strong>the</strong> data limitations<br />

described above:<br />

Initial random sample: 19,996 (100%)<br />

Cases with addresses: 15,595 (78%)<br />

Estimated cases with valid residential phone number: 12,629 (63%)<br />

Total completed surveys: 1,645 (8%)<br />

Despite <strong>the</strong> limitations <strong>in</strong> data and contact <strong>in</strong>formation described above, <strong>the</strong> overall randomness<br />

<strong>of</strong> <strong>the</strong> sample and survey were preserved. As shown <strong>in</strong> Table 1, <strong>the</strong> composition <strong>of</strong> <strong>the</strong> sample<br />

population and f<strong>in</strong>al survey respondents is representative <strong>of</strong> <strong>the</strong> entire GAIN population. There<br />

are slight differences, primarily <strong>in</strong> terms <strong>of</strong> household aid type and case primary language, which<br />

can largely be attributed <strong>to</strong> <strong>the</strong> over-sampl<strong>in</strong>g <strong>of</strong> two-parent (U) cases <strong>to</strong> <strong>in</strong>crease <strong>the</strong> male<br />

response rate. The over-sampl<strong>in</strong>g <strong>of</strong> two-parent cases may also expla<strong>in</strong> <strong>the</strong> observed differences<br />

<strong>in</strong> primary language. Non-English cases are more prevalent among two-parent (U) cases. S<strong>in</strong>ce<br />

U cases are over-represented among survey respondents, survey tabulations for this report are<br />

weighted <strong>in</strong> order <strong>to</strong> assure that that tabulations are representative <strong>of</strong> <strong>the</strong> <strong>welfare</strong> <strong>to</strong> <strong>work</strong><br />

population <strong>in</strong> Los Angeles County (as described <strong>in</strong> more detail below).<br />

3

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!