No description available.

No description available.

Yes / Low cost carriers (budget airlines) have a significant share of the air travel market, but little research has been done to understand the distributional effect of their operation on key tourism indicators such as length of stay and expenditure. Using data on European visitors to the United Kingdom we demonstrate how counterfactual decompositions can inform us of the true impact of mode of travel. Passengers on low cost carriers tend to spend less, particularly at the upper end of the distribution. Budget airline users typically stay longer, though differences in characteristics of observed groups are important to this result. Counterfactual techniques provide additional valuable insights not obtained from conventional econometric models used in the literature. Illustrating an application of the methodology to policy we demonstrate that enabling respondents to extend their stay generates the greatest additional expenditure at the lower end of the distribution. We also show nationality is a significant characteristic, with important impacts across the expenditure distribution.

This thesis explores the proposal that there is a close link between counterfactual thinking and lying. Although both require the imagination of alternatives to reality, research has yet to establish a direct link. In the first seven studies the relationship between counterfactuals and lies is directly investigated using novel scenario-based and behavioural tasks. In a further four studies we also investigate the role of affect and executive functions as explanatory mechanisms. Results show that individuals with a tendency to think counterfactually are more likely to generate potential lies and to be more successful when lying in front of others (Study 1 and 6). Furthermore, we also show that counterfactual availability influences people’s tendency to come up with lies (Studies 2, and 3) and the extent to which they expect others to lie (Studies 4, and 5). We also find that the saliency of counterfactual alternatives can affect people’s moral standards by motivating them to lie (Study 7). Based on these results we argue that counterfactuals motivate lying by providing information about how things could have been different. We however also investigate alternative explanations. In Studies 8, 9 and 10 we seek to understand whether counterfactually derived affect might also underlie the relationship, but find no such link. Additionally, in Study 11 we investigate the relationship in Parkinson’s disease participants in order to understand if executive function might be an underlying mechanism. We do not find this to be the case and we show that PD patients are able to engage in counterfactual thinking and also lie. The findings in this thesis are the first to provide a direct link between counterfactual thoughts and lies. Overall, we show how counterfactuals can help us mislead others and we reveal that counterfactual thinking is an important cognitive process in deception.

Personalized learning considers that the causal effects of a studied learning intervention may differ for the individual student (e.g., maybe girls do better with video hints while boys do better with text hints). To evaluate a learning intervention inside ASSISTments, we run a randomized control trial (RCT) by randomly assigning students into either a control condition or a treatment condition. Making the inference about causal effects of studies interventions is a central problem. Counterfactual inference answers “What if� questions, such as "Would this particular student benefit more if the student were given the video hint instead of the text hint when the student cannot solve a problem?". Counterfactual prediction provides a way to estimate the individual treatment effects and helps us to assign the students to a learning intervention which leads to a better learning. A variant of Michael Jordan's "Residual Transfer Networks" was proposed for the counterfactual inference. The model first uses feed-forward neural networks to learn a balancing representation of students by minimizing the distance between the distributions of the control and the treated populations, and then adopts a residual block to estimate the individual treatment effect. Students in the RCT usually have done a number of problems prior to participating it. Each student has a sequence of actions (performance sequence). We proposed a pipeline to use the performance sequence to improve the performance of counterfactual inference. Since deep learning has achieved a huge amount of success in learning representations from raw logged data, student representations were learned by applying the sequence autoencoder to performance sequences. Then, incorporate these representations into the model for counterfactual inference. Empirical results showed that the representations learned from the sequence autoencoder improved the performance of counterfactual inference.

This dissertation assesses the causes and consequences of gender wage discrimination in Ethiopia. In the first chapter, we estimate the distribution of Gender Wage Discrimination in the Ethiopian urban labor market using quantile counterfactual decompositions. The literature generally finds a u-shaped distribution suggesting the presence of both a sticky floor effect and a glass ceiling effect. Using repeated cross-section data for the years 2006, 2010 and 2014, we find a strong evidence of a sticky floor effect but not a glass ceiling effect in the Ethiopian urban labor market. Our paper also provides evidence that there is substantial difference in the extent of discrimination between working in private and public jobs. Public jobs are less discriminatory for women relative to the private jobs. In the second chapter, we investigate the determinants of the gender wage gap in the Ethiopian manufacturing sector between the years 1996 and 2010 with a particular focus on the impact of the export orientation. This is done both at the intensive and extensive margin. Accordingly, we find that more export orientation helps reduce the firm level gender wage gap regardless of whether it is at the intensive or extensive margin. Our results also provide evidence on the presence of sectoral variation on the association between export orientation and gender wage gap. Export orientation doesn’t have a significant impact on the gender wage gap in the construction and housing goods sector. Segmenting the data in to two we also find that the impact of export orientation in reducing gender wage gap is much stronger for the period 2003-2010 relative to the 1996-2002 period. Finally, we estimate the impact of gender earnings differentials on the technical efficiency of the firm in the Ethiopian manufacturing sector for the period 1996 through 2010. We adopt a two-step time-variant panel stochastic frontier model using a translog production function. Our results provide fresh evidence on the existence of a significant negative association between gender wage gap and predicted technical efficiencies of firms. Further subdividing the manufacturing sector into four different industries, we find that the negative association is consistent in most industries. Our results are also robust to the inclusion of other firm level explanatory variables at the sectoral level.

Adjusting for baseline pre-intervention characteristics between treatment groups, through the use of propensity score matching methods, is an important step that enables researchers to do causal inference with confidence. This is critical, largely, due to the fact that practical treatment allocation scenarios are non-randomized in nature, with various inherent biases that are inevitable, and therefore requiring such experimental manipulations. These propensity score matching methods are the available tools to be used as control mechanisms, for such intrinsic system biases in causal studies, without the benefits of randomization (Lane, To, Kyna , & Robin, 2012). Certain assumptions need to be verifiable or met, before one may embark on a propensity score matching causal effects journey, using the Rubin causal model (Holland, 1986), of which the main ones are conditional independence (unconfoundedness) and common support (positivity). In particular, with this dissertation we are concerned with elaborating the applications of these matching methods, for a ‘strong-ignorability’ case (Rosenbaum & Rubin, 1983), i.e. when both the overlap and unconfoundedness properties are valid. We will take a journey from explaining different experimental designs and how the treatment effect is estimated, closing with a practical example based on two cohorts of enrolled introductory statistics students prior and post-clickers intervention, at a public South African university, and the relevant causal conclusions thereof. Keywords: treatment, conditional independence, propensity score, counterfactual, confounder, common support / Dissertation (MSc)--University of Pretoria, 2020. / Statistics / MSc / Unrestricted

No description available.

No description available.

No description available.