A Guide to Longitudinal Program Impact Evaluation

Ángeles, Gustavo; Cronin, Christopher; Dk, Guilkey; Pm, Lance; Sullivan, Briana M.

doi:10.2139/ssrn.2796689

Cited by 6 publications

(7 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…GLLAMM are a class of multilevel models that estimate fixed and random effects of various types of outcome variables including continuous, count, binary, ordered and unordered categorical variables [ 41 – 43 ]. In this study, two-level multinomial logistic regression models were run with discrete factor approximation, which give maximum likelihood semi-parametric estimators that are consistent and asymptotically efficient under the model’s assumptions [ 44 ]. An intercept-only (null) model was run to test the null hypothesis that there was no between-cluster variation in the modern contraceptive use pattern during the study period.…”

Section: Methodsmentioning

confidence: 99%

Gender norms and modern contraceptive use in urban Nigeria: a multilevel longitudinal study

et al. 2018

View full text Add to dashboard Cite

BackgroundEvidence suggests that gender equality positively influences family planning. However, the evidence from urban Africa is sparse. This study aimed to examine the association between changes in gender norms and modern contraceptive use over time among women in urban Nigeria.MethodsData were collected in 2010/2011 from 16,118 women aged 15–49 living in six cities in Nigeria (Abuja, Benin, Ibadan, Ilorin, Kaduna, and Zaria) and again in 2014 from 10,672 of the same women (34% attrition rate). The analytical sample included 9933 women living in 480 neighborhoods. A four-category outcome variable measured their change in modern contraceptive use within the study period. The exposure variables measured the changes in the level of gender-equitable attitudes towards: a) wife beating; b) household decision-making; c) couples’ family planning decisions; and d) family planning self-efficacy. Multilevel multinomial logistic regression models estimated the associations between the exposure variables at the individual and neighborhood levels and modern contraceptive use controlling for the women’s age, education, marital status, religion, parity, household wealth, and city of residence.ResultsThe proportion of women who reported current use of modern contraceptive methods increased from 21 to 32% during the four-year study period. At both surveys, 58% of the women did not report using modern contraceptives while 11% reported using modern contraceptives; 21% did not use in 2010/2011 but started using by 2014 while 10% used in 2010/2011 but discontinued use by 2014. A positive change in the gender-equitable attitudes towards household decision-making, couples’ family planning decisions, and family planning self-efficacy at the individual and neighborhood levels were associated with increased relative probability of modern contraceptive use (adoption and continued use) and decreased relative probability of modern contraceptive discontinuation by 2014. No such associations were found between the individual and neighborhood attitudes towards wife beating and modern contraceptive use. Accounting for the individual and neighborhood gender-equitable attitudes and controlling for the women’s demographic characteristics accounted for 55–61% of the variation between neighborhoods in the change in modern contraceptive use during the study period.ConclusionInterventions that promote gender equality have the potential to increase modern contraceptive use in Nigerian cities.Electronic supplementary materialThe online version of this article (10.1186/s12905-018-0664-3) contains supplementary material, which is available to authorized users.

show abstract

Section: Methodsmentioning

confidence: 99%

Gender norms and modern contraceptive use in urban Nigeria: a multilevel longitudinal study

et al. 2018

View full text Add to dashboard Cite

show abstract

“…In robust regressions, the so-called sandwich estimator is used to estimate the standard errors (Angeles, Cronin, Guilkey, Lance, & Sullivan, 2014;Colin Cameron & Miller, 2015).…”

Section: Regression Modelsmentioning

confidence: 99%

Convergent validity of several indicators measuring disruptiveness with milestone assignments to physics papers by experts

Bornmann

Tekles

2021

Journal of Informetrics

View full text Add to dashboard Cite

“…This option specifies that the citation rates of the papers are independent across subject categories but are not necessarily independent within the same subject category (Hosmer & Lemeshow, , section 8.3; Long & Freese, ). The cluster option in Stata leads to robust standard errors in the regression model (with further corrections for the effects of clustered data) so that for each subject category (or category combination) a set of papers exists whose citations arise from a similar discipline‐specific citation culture (Angeles, Cronin, Guilkey, Lance, & Sullivan, ). In general, robust standard errors should be used if the assumptions of a model are not correctly specified (here, independent observations).…”

Section: Methodsmentioning

confidence: 99%

“…Predictive margins are presented in this study with confidence intervals. The cluster option in Stata leads to robust standard errors in the regression model (with further corrections for the effects of clustered data) so that for each subject category (or category combination) a set of papers exists whose citations arise from a similar discipline-specific citation culture (Angeles, Cronin, Guilkey, Lance, & Sullivan, 2014). There is something deeply inappropriate about scientists using a measure of performance without considering its precision."…”

Section: Data Set and Statisticsmentioning

confidence: 99%

How much does the expected number of citations for a publication change if it contains the address of a specific scientific institute? A new approach for the analysis of citation data on the institutional level based on regression models

Bornmann

2015

Asso for Info Science & Tech

View full text Add to dashboard Cite

Citation data for institutes are generally provided as numbers of citations or as relative citation rates (as, for example, in the Leiden Ranking). These numbers can then be compared between the institutes. This study aims to present a new approach for the evaluation of citation data at the institutional level, based on regression models. As example data, the study includes all articles and reviews from the Web of Science for the publication year 2003 (n = 886,416 papers). The study is based on an in-house database of the Max Planck Society. The study investigates how much the expected number of citations for a publication changes if it contains the address of an institute. The calculation of the expected values allows, on the one hand, investigating how the citation impact of the papers of an institute appears in comparison with the total of all papers. On the other hand, the expected values for several institutes can be compared with one another or with a set of randomly selected publications. Besides the institutes, the regression models include factors which can be assumed to have a general influence on citation counts (e.g., the number of authors).

show abstract

A Guide to Longitudinal Program Impact Evaluation

Cited by 6 publications

References 9 publications

Gender norms and modern contraceptive use in urban Nigeria: a multilevel longitudinal study

Gender norms and modern contraceptive use in urban Nigeria: a multilevel longitudinal study

Convergent validity of several indicators measuring disruptiveness with milestone assignments to physics papers by experts

How much does the expected number of citations for a publication change if it contains the address of a specific scientific institute? A new approach for the analysis of citation data on the institutional level based on regression models

Contact Info

Product

Resources

About