CAPÍTULO I. MARCO TEÓRICO DE LA INVESTIGACIÓN
1. Formación del profesorado. Conceptos, teorías y modelos
1.5. El continuum de la formación: el desarrollo profesional del
1.5.1. La formación inicial
I merge information from each interview with the Princeton Meteorological Forcing Dataset, henceforth the Princeton Data. The Princeton Data is a reanalysis dataset, mean-ing that it combines a host of observational weather data from sources such as weather balloons, weather stations, and satellite images with a physics-based weather model. The model allows coverage to extend to areas with sparse observational data such as developing
countries.16 These data are available through 2010 every three hours for each 0.25 degree latitude/longitude increment. The dataset includes information on temperature, specific hu-midity, and pressure, which I use to calculate relative humidity and wet bulb temperature using standard meteorological formulas.17
Wet bulb temperature is a nonlinear function of dry bulb temperature (dry bulb tem-perature is the temtem-perature reading usually displayed on daily weather forecasts) and relative humidity. A wet bulb reading corresponds to the temperature reading of a thermometer that has been wrapped in a wet cloth: the faster the moisture in the cloth evaporates, the lower the reading. To help the reader visualize the relationship between wet bulb temperature, dry bulb temperature, and relative humidity, Figure B2 displays the relationship between the three variables along with a few illustrative examples. The figure displays isometric dry bulb lines, with each line representing a fixed dry bulb temperature at varying relative humidities, represented by the horizontal axis. The combination of these two variables can be read on the vertical axis as a wet bulb temperature. Wet bulb temperature is always lower than dry bulb temperature, except at 100% relative humidity where the two readings are equal. The difference between wet bulb and dry bulb temperature is larger at higher temperatures.
The red points in the figure plot two examples of the differences between wet and dry bulb temperature. The August, 2017 average of daily average weather conditions are plotted for Las Vegas, NV and Houston, TX. The desert heat of Las Vegas manifests as a relatively high dry bulb temperature (over 90 degrees Fahrenheit), but, due to the low relative humidity, a comparatively low wet bulb temperature. The more muggy heat in
16Appendix B.1 gives more detail on the construction of this dataset.
17See Appendix B.1 for more details.
Houston creates a higher wet bulb temperature than Las Vegas despite the fact that the dry bulb reading is lower. The emphasis on relative humidity means that the ranking of temperatures in wet bulb and dry bulb temperatures are very different, with humid hot days in places like South Asia rated as more severe than the desert heat of places like the Arabian Peninsula or Sub-Saharan Africa by wet bulb temperature, but as less severe using dry bulb temperature. As I show in Section 2.5, wet bulb temperature is more predictive of worker productivity, which means that the effects of weather on productivity may be concentrated in more humid areas of the world.
I merge weather data to DHS interviews by locating the four grid points in the Princeton Data surrounding each DHS cluster (each cluster in this sample has 20 selected households on average and is represented in the DHS data by a single latitude and longitude) and then creating an average of the four grid points weighted by inverse distance between the cluster and each grid point. The weather variables are then specific to a day of interview/DHS cluster. For the main analysis, I collapse the eight daily readings into a daily average wet bulb temperature, but in Section 2.8 I show the results for alternative specifications of weather exposure. To more thoroughly control for local average climate, I also create 10-year averages of wet bulb temperature in a given month in a given cluster (2000-2010 averages).
Figure 2.3 displays summary statistics on the weather data merged with the DHS data. Panel A shows the average wet bulb temperature in each survey cluster sampled in the DHS surveys in the analysis, where the wet bulb temperature is divided into the same bins as in the main regression specification. The average is of daily average wet bulb temperatures during the interviews in the sample. The variation used in the analysis is within a region of the country, so identification comes from areas where there was more variation in wet
bulb temperature across interviewing days within a geographical area in the main regression specifications.
Panel B presents this graphically; it shows the regional distribution of interviewer work days in each wet bulb bin used for the main analysis. As the figures make clear, a bit over half of all observations are in Sub-Saharan Africa (SSA). SSA’s climate is often quite hot, but in most places it is a dry heat, which means that the wet bulb distribution in Africa is largely compressed to the middle bins. The days in the highest wet bulb bins are concentrated in humid locations of the world, such as South Asia and Latin America.
Days in the highest wet bulb bin are quite rare, accounting for less than 0.1 percent of days in the sample, and these occur only in South Asia and Latin America in this sample. These days are projected to become more common with climate change, particularly in areas such as South Asia (Steven C. Sherwood and Matthew Huber, 2010).