Dear researchers,
I am working on a dataset from 4 countries where one of our hypotheses is to examine the relationship between gender role attitudes and how much household chores couples engage in. The problem is, the gender role scale has a highly negative skew in three countries and a fairly normal distribution in the fourth (Statistic = -.434, SD = .174). I tried to use the Log transformation to normalize the datasets, but now the three "highly skewed" countries are normal, and the last country is even more negatively skewed (Statistic = -1.008, SD = .174). Any idea how to sort this out? I thought about only transforming the data for the three countries, but then how do I merge it with the fourth?
Best,
Faith
I guess the first question is, what kind of analysis are you trying to perform? If investigating the relationship between gender roles and household chores, how do the four countries come into play in the analysis?
I'd also recommend looking at histograms of the data, rather than the skewness statistics. A skewness of -1 may not be very problematic. Or a robust analysis may make short work of the discrepancy in distributions.
Now this is a sample effect, either from population or from bootstraps . If you must work with normal distribution for all the countries then d other country with skewed distribution should be dealt with using a stochastic process. This is where you perform a Monte Carlo simulation using descriptive statistics then you bootstrap and choose or select the simulation that gives a normal distribution for your sample.
But on a serious note, you can not merge the two distribution types for that kind of study.
Check https://www.khanacademy.org/math/ap-statistics/random-variables-ap/combining-random-variables/a/combining-normal-random-variables
- Badges
- Science topic
- Similar topics
- Mathematics