Hypothesis testing: principles and methods

¶ … global warming. One of the arguments in favor of this being an issue is that temperatures have increased over time, since humans began burning fossil fuels. There are a number of ways to look at the issue, but one thing that has not received much consideration is that baseline data, which is taken from urban areas, may be skewed by other factors besides anthropogenic global warming. One factor that has not yet been studied is the effect of cloudy days vs. sunny days on the temperature readings. To investigate whether there is any merit to this idea, we need test two hypotheses. The first is whether or not the weather in the region makes a difference to the mean daily temperatures that are recorded. The second is whether or not there has been a change in weather patterns (for example, if the pollution from our cities has brought about more cloudy days). We have only looked at the first test.

The first test is whether or not cloudiness affects the mean temperature. The null hypothesis for this test is that cloudy days are warmer than sunny days in winter, and cooler than sunny days in the summer. This hypothesis will be tested with 20 years' worth of data from LaGuardia Airport in New York City.

The results of these tests support the null hypothesis for the summer, and reject it for the winter. It has been found that the mean daily temperature on sunny days is 0.2 degrees warmer than the mean daily temperature in the summer, but on cloudy days it is 2.4 degrees cooler, with a p-value of 0.043, which indicates that this value is significant. The winter data did not show a high degree of variation in the mean temperature, so cloudy cover is not a good predictor of temperature in the winter.

The key to understanding the statistical test is to recognize that the data set used was sufficiently large. The population was for the history of the city, but the sample was the last 20 years. By running data with such a large set, one can have confidence in the outputs. There are definitely limits to the application of this data, however, to the original problem. It is unlikely that upholding the significance of the summer data is going to relate back to a case against anthropogenic climate change, for example. Far more research would need to be made, from multiple cities around the world, and then tested against rural areas to show that there is a link between the buildup of urban areas, changes of weather patterns linked that buildup, and that all of this contributes to our current views on climate change. As a first step, this research is limited as well, because it does not take into account all of the possible variables, but it does show that an increase in cloud cover resulting from urbanization of the ecosystem would not have resulted in a temperature increase because there is no link between cloud cover and temperature change in winter.

Please consider funding more studies of this nature in order to better understand how different factors contribute to our understanding of how urbanization affects our climate.

Hypothesis testing requires a data set, and then it requires a null and alternative hypothesis. Such tests can seek to determine a correlation between two variables, known as the dependent and independent variables. An example of a hypothesis to be tested is whether sunny days in New York are correlated with warmth. This test compares days that are sunny with days that are cloudy, adjusted for the season. The null hypothesis will be that sunny days are hotter than cloudy ones in the summer, but colder than cloudy ones in the winter. The alternative hypothesis is the vice versa of the null hypothesis. Seasons will be defined by the calendar seasons, and the temperatures will be the daily mean in Fahrenheit.