A significant and a meaningful pattern of differences was again source of information. RocaglamideMedChemExpress Roc-A students from schools with the lowest percentages of individuals PD98059MedChemExpress PD98059 possessing certificates mentioned schools as their primary information source– Kwakadogo (45.6 of time), Mwambalazi (33.7 ). This compared to lower mention of schools as an information source among students at schools with higher percentage of students possessing certificates (e.g., Kinango, 1.4 ; Ndohivyo, 15.3 ). On the other hand, mention of chiefs was lower among schools whose students showed lowest rates of possessing certificates–Kikoneni, Mwambalazi–compared to schools with higher percentages of students in possession of certificates (see Table 2). Finally, to better understand the combined contribution of the above factors, we conducted a binary logistic regression. This again used all of the factors (listed on the left side of Table 3) which had shown significance in regards to possession of birth certificates as predictor variablesPLOS ONE | DOI:10.1371/journal.pone.0149925 March 3,10 /How Would Children Register Their Own Births?Table 3. Knowledge and demographic factors significantly tied to possession of birth certificate among students in Kwale, Kenya. Univariate analyses (left) and Binomial Logistic Regression (right). have certificate Univariate Analyses X2 (df, N) Grade level eight seven six School Kinango Lukore NdoHivyo Mwambalazi Kikoneni KwaKadogo Sibling has certificate? Yes No Religion Catholic Protestant Muslim Know other children with cert.? Yes No Definition of cert. Doc. of ID/citizenship Doc. used for healthcare Doc. with vital/birth statistics Doc. used in “future” Other (general card or important) Doc. for school/exam registration Don’t know Source of information Relatives Church Chief SART.S23506 Hospital Midwife/ sibling birth School Community Don’t know 56.0 55.6 51.7 48.5 40.0 37.0 34.1 23.5 83.3 50.0 45.4 46.7 38.9 36.9 45.0 14.0 (7, N = 473)* 5.1 49.7 39.1 16.7 (6, N = 473)* 4.4 54.2 53.9 39.5 5.0 (1, N = 473)* 1.9 70.3 33.6 9.8 (2, N = 473)** 75.7 54.4 52.8 40.5 31.4 19.4 56.5 (1, N = 473)*** 63.9 32.9 38.6 58.9 (5, N = 473)*** 36.2 (2, N = 473)*** Binomial Logistic Regression a Wald b 40.7*** —36.9*** 22.8*** 29.3*** —4.6* 5.1* 4.3* 8.3** 27.6*** 38.0*** ——4.3 1 0.2 —[0.12?.34] 1 0.41 0.38 0.36 0.26 0.08 —[0.18?.93] [0.16?.88] [0.14?.94] [0.11?.65] [0.03?.21] 1 0.17 0.24 —[0.10?.30] [0.13?.43] Odds ratio c CI (95 )Notes: All respondents j.jebo.2013.04.005 were elementary school students in grades six to eight.aRegression model (right side) performed on all factors shown in left column which had shown significance in individual univariate analyses. All factorstreated as categorical. Model significant at p < .001. 2 (24, N = 473) = 165.97, Nagelkerke R2 = .40, correctly predicted cases = 74.2 . b *** significant at p < .001, ** p < .01, * p < .05.cOdds ratios and CIs are shown for significant factors only, and show comparison to topmost category in group.Odds ratios should not be interpreted as approximated relative risk. Estimated odds as shown in the Table will be closer to 1 than the ratio change of all odds (which cannot reliably be estimated via this approach) doi:10.1371/journal.pone.0149925.tPLOS ONE | DOI:10.1371/journal.pone.0149925 March 3,11 /How Would Children Register Their Own Births?(treated as categorical data). The model results are reported on the right side of Table 3, with Odds Ratios and 95 Confidence Intervals also shown for factors that w.A significant and a meaningful pattern of differences was again source of information. Students from schools with the lowest percentages of individuals possessing certificates mentioned schools as their primary information source-- Kwakadogo (45.6 of time), Mwambalazi (33.7 ). This compared to lower mention of schools as an information source among students at schools with higher percentage of students possessing certificates (e.g., Kinango, 1.4 ; Ndohivyo, 15.3 ). On the other hand, mention of chiefs was lower among schools whose students showed lowest rates of possessing certificates--Kikoneni, Mwambalazi--compared to schools with higher percentages of students in possession of certificates (see Table 2). Finally, to better understand the combined contribution of the above factors, we conducted a binary logistic regression. This again used all of the factors (listed on the left side of Table 3) which had shown significance in regards to possession of birth certificates as predictor variablesPLOS ONE | DOI:10.1371/journal.pone.0149925 March 3,10 /How Would Children Register Their Own Births?Table 3. Knowledge and demographic factors significantly tied to possession of birth certificate among students in Kwale, Kenya. Univariate analyses (left) and Binomial Logistic Regression (right). have certificate Univariate Analyses X2 (df, N) Grade level eight seven six School Kinango Lukore NdoHivyo Mwambalazi Kikoneni KwaKadogo Sibling has certificate? Yes No Religion Catholic Protestant Muslim Know other children with cert.? Yes No Definition of cert. Doc. of ID/citizenship Doc. used for healthcare Doc. with vital/birth statistics Doc. used in "future" Other (general card or important) Doc. for school/exam registration Don't know Source of information Relatives Church Chief SART.S23506 Hospital Midwife/ sibling birth School Community Don’t know 56.0 55.6 51.7 48.5 40.0 37.0 34.1 23.5 83.3 50.0 45.4 46.7 38.9 36.9 45.0 14.0 (7, N = 473)* 5.1 49.7 39.1 16.7 (6, N = 473)* 4.4 54.2 53.9 39.5 5.0 (1, N = 473)* 1.9 70.3 33.6 9.8 (2, N = 473)** 75.7 54.4 52.8 40.5 31.4 19.4 56.5 (1, N = 473)*** 63.9 32.9 38.6 58.9 (5, N = 473)*** 36.2 (2, N = 473)*** Binomial Logistic Regression a Wald b 40.7*** —36.9*** 22.8*** 29.3*** —4.6* 5.1* 4.3* 8.3** 27.6*** 38.0*** ——4.3 1 0.2 —[0.12?.34] 1 0.41 0.38 0.36 0.26 0.08 —[0.18?.93] [0.16?.88] [0.14?.94] [0.11?.65] [0.03?.21] 1 0.17 0.24 —[0.10?.30] [0.13?.43] Odds ratio c CI (95 )Notes: All respondents j.jebo.2013.04.005 were elementary school students in grades six to eight.aRegression model (right side) performed on all factors shown in left column which had shown significance in individual univariate analyses. All factorstreated as categorical. Model significant at p < .001. 2 (24, N = 473) = 165.97, Nagelkerke R2 = .40, correctly predicted cases = 74.2 . b *** significant at p < .001, ** p < .01, * p < .05.cOdds ratios and CIs are shown for significant factors only, and show comparison to topmost category in group.Odds ratios should not be interpreted as approximated relative risk. Estimated odds as shown in the Table will be closer to 1 than the ratio change of all odds (which cannot reliably be estimated via this approach) doi:10.1371/journal.pone.0149925.tPLOS ONE | DOI:10.1371/journal.pone.0149925 March 3,11 /How Would Children Register Their Own Births?(treated as categorical data). The model results are reported on the right side of Table 3, with Odds Ratios and 95 Confidence Intervals also shown for factors that w.