Independent Variables. Measure Description Source Staff size FTEs of the agency; control variable Annual budget declarations, agency websites Budget Annual budget of each agency; control variable Annual budget declarations, agency websites Media salience Number of newspaper articles that appear when using the agency name as a search term on the Factiva newspaper database for the time from 01/01/2014- 01/01/2018 Factiva Legal mandate Responsibility for regulatory or non-regulatory tasks Coded based on Rimkute (2019) coding scheme, verified by checking agency websites. Policy field Responsibility for economic or social policy Coded based on Rimkute (2019) coding scheme, verified by checking agency websites. tweeting frequency Average number of posted tweets per month Twitter (Share of ) informational tweets (Percentage of) tweets that have an informational character Twitter, machine learning classification of Tweets (Share of ) engaging tweets (Percentage of) tweets that have an engaging character Twitter, machine learning classification of Tweets Age Age of each agency Twitter account in days until October 1, 2019. Twitter This research used a machine learning classifier to automatically classify the tweets of agencies according to whether they are informational or engaging in nature (or none of these two options). Using machine learning techniques is increasingly common in public administration and political science research and has also been applied to agency communication (Xxxxxxxxxxxxxx and Xxxxxxxx 2018; Xxxxxxxxxx 2018; Xxxxxx and Xxxxxxxxx 2020). A subset of account descriptions and agency tweets was hand-coded into different classes (see table 2) and then automatically classified by the Bidirectional Encoder Representations from Transformers (XXXX) algorithm (Xxxxxx et al. 2018) provided by Google AI (for detailed information on how XXXX operates, see appendix B). XXXX is a language model that is based on some of the most recent developments in natural language processing and outperforms many of the standing benchmarks in language recognition and text classification. It is therefore well suited for application in this research context. Amongst the over 91000 tweets of the EU agencies, 1000 were randomly sub-sampled and hand-coded whether they belonged to one or two of the non-exclusive tweet categories (for coding examples, see table 2). Tweets were classified as engaging if they encouraged the reader to become active, either in everyday life or by participating in events...
Independent Variables robots’ parameters We examined the influence of three independent variables on the selected self-reported measures.
Independent Variables. ANC variables included receipt and quality measures. ANC receipt was measured by a timing variable (Early enrollment (1st trimester) vs. late enrollment (2nd or 3rd trimester) of first ANC visit) and a frequency variable (<4 ANC visits vs. ≥4 ANC visits) defined by WHO standards27. ANC quality variable comprised 18 practices and counseling topics measured by the survey instrument. These topics did not address IFA administration or counseling but were included as an overall quality measure. Nine covered ANC practices (weight, height, blood pressure, blood test, urine test, breast exam, abdomen exam, sonogram/ultrasound, and delivery date given) and nine reviewed ANC counseling topics (advice regarding delivery, nutrition, breastfeeding, keeping the baby warm, cleanliness at delivery, family planning for spacing and limiting, improved maternal and child nutrition, and importance of institutional delivery). We removed 5 variables (blood pressure, urine test, delivery advice, keeping the baby warm, and family planning for limiting) from the analysis because they were too highly correlated, causing a singular matrix. We used the remaining variables to conduct an exploratory factor analysis using polychoric correlation matrices. These have been shown to result in more accurate correlations between categorical variables28. We then extracted factors from a principal components analysis and rotated them orthogonally using the varimax method.
Independent Variables. ANC: ANC variables included receipt and quality measures. ANC receipt was measured by a timing variable (Early enrollment (1st trimester) vs. late enrollment (2nd or 3rd trimester) of first ANC visit) and a frequency variable (<4 ANC visits vs. ≥4 ANC visits) according to WHO standards92. ANC quality variable comprised 18 practices and counseling topics measured by the survey instrument. These topics did not address IFA administration or counseling but were included as an overall quality measure of ANC services. There were 9 ANC practices (weight, height, blood pressure, blood test, urine test, breast exam, abdomen exam, sonogram/ultrasound, and delivery date given) and 9 ANC counseling topics (advice regarding delivery, nutrition, breastfeeding, keeping the baby warm, cleanliness at delivery, family planning for spacing and limiting, improved maternal and child nutrition, and importance of institutional delivery). We removed 5
Independent Variables. Additionally, the QLTE collected demographic information (gender, race/ethnicity, and years of teaching experience) from each respondent. To account for the influence of individual identity and group membership on evaluations of working conditions within the school, these data were included as individual-level control variables. Table 3 provides a complete list of variables. The survey instrument is in Appendix A. To incorporate measures of NCLB policy and the implementation of sanctions in the sampled schools, I linked the QLTE data with School District reports to the Georgia Department of Education. Specifically, I collected data from the School Report Cards filed and publicized on the Georgia Department of Education Office of Student Achievement website.9 I drew two types of data from these reports. First, I used the 2004-2005 reports to draw demographic and compositional data during the time of QLTE data collection. These data items include: percentage of students with disabilities, percentage on free and reduced price lunch, percent minority enrollment, percent limited English proficiency, total school enrollment, total number of full time teachers, and Title I status. These items serve as organization-level control variables to account for the 9 School report cards are available for examination on the Georgia Department of Education Office of Student Achievement website. This site can be found at xxxx://xxxxxxxxxx0000.xxxxx.xxx/k12. impact of school context and composition on teachers‘ reports of working conditions. Second, I used the 2003-2004 Adequate Yearly Progress (AYP) reports to the Georgia Department of Education. I used the prior year‘s data since they indicate the AYP designations in effect during the 2004-2005 school year when QLTE data collection occurred. From these reports I included: previous year AYP status, number of years in Needs Improvement status, and an indicator for each sanction currently imposed on the school. These measures of NCLB‘s influence served as the primary independent variables under analysis. I assessed the impact of NCLB in two ways in this statistical model. First, I measured NCLB using the total number of years that a school has been in Needs Improvement status. 10 From this perspective, I examine the cumulative impact of NCLB on teachers. This approach assumes that NCLB may intensify over time and the prolonged exposure to sanctions may produce distinct interpretations of working conditions by teachers. Second,...
Independent Variables. FWS State Lands
Independent Variables. HOMETOWN OF FLEMISH MINISTER
Independent Variables. Imperfections in electoral markets
Independent Variables. Independent variables are identified from associated individual and household level factors. Individual-level factors include sex of child, size of child at birth, children’s age, birth order, maternal education attainment, and husband’s education level. Household-level factors include type of place of residence, family wealth index, and region. Among these factors, birth size, children’s age, and maternal education attainment have been re-coded as new variables and have small categories combined to avoid bias. New variable “mother education attainment” is created using variable v149 (maternal educational attainment) and categorized as no education, incomplete primary, complete primary, incomplete secondary and higher. New variable “birth size” is determined using variable m18 (size of child at birth) and categorized as very large, larger than average, average, and smaller than average. New variable “child age” is created using variable v013 (age in 5-year groups) and categorized as 15-29 and 30-49 months. Lastly, new variable “birth order” is created using variable bord (birth order number) and categorized as first, second and third plus. Any missing values are recorded as “.” and excluded from the data analysis. The 2014 data has 9 missing maternal weight values, 1 missing birth size value, and 1 missing partner education level value. For mother-child weight pairs, concordant weight pairs have 1,861 missing, while discordant weight pairs have 214 missing. In 2010, there are 118 missing values for birth size and 60 missing values for partner education. There are 103 missing values for partner education and 35 missing values for birth size in the 2005 data. Lastly, there are 42 missing values for partner education and 119 missing values for birth size in the 2000 data.
Independent Variables. My key independent variables are operationalized as follows. Veto players are actors whose assent is required for change to the policy status quo to occur. I limit the 10Very large firms may sometimes find it in their interest to engage in activities with positive externalities that will benefit the entire sector. These activities may include creating horizontal and vertical organizations that serve the interests of the sector. Although such behavor is initiated by large firms for self-interested reasons, I count it as evidence of coordinative interfirm institutions when the actions of the resulting institutions serve the interests of the horizontal or vertical bodies broadly. scope of this project to single, non-alternating veto player regimes and countries with a very high number veto players.11 Following XxxXxxxxx (2003a,b), I chose not to code ideology, but rather focused on the number of discrete, partisan actors who have formal veto authority over the particular policy space. Where factions are potentially important players within collective veto players, I use the level of direct, zero-sum competition among factions to determine whether to code them as veto players. So, though military factions may be important actors within a junta, they are only be counted as separate veto players where they are actively trying undermine one another via promotion channels or a coup. Likewise, a faction within a political party will only be counted as a separate veto player if it competes for votes with other factions in the same party. I am distinguishing between macro and micro economic policy stability because, while both are important for actors’ perceptions of the risks involved in forming coordinating institutions, they may be decided by different numbers of veto players. Additionally, they may be of unequal importance. It may be that the presence of particularistic interests in macro policy is more likely to result in highly hierarchical institutions than in micro policy.12 Though some have suggested that additional veto players may have non-linear effects on a variety of outcomes (XxxXxxxxx, 2003a,b; Cox & XxXxxxxxx, 2001), none have specified precisely where such a threshold would be. The observations I use to 11I do not assume all authoritarian governments are single veto player regimes. Tsebelis (2002; 1995) notes that powerful and distinct actors may be sharing power through institutionalized means with or without mass participation or protections of ci...
