1
$\begingroup$

Deeply sorry as this wasn't really covered in my statistics classes. I am current comparing datasets for people divorcing vs. dissolving their marriages. I have two variables, length of marriage, and age of the parties, that are not normally distributed. length is skewed positively. The Age is a little flatter than the bell. I can get normal distribution using a Box-Cox, but I want to compare the means of the two populations, to see if there's a difference. Do i need to pick a Lambda to compare them, and if so, how? the range and distributions for both populations for the variables are similar, they just aren't normal. It's entirely possible I am going about this the wrong way, and should do a different test all together, but I only have access to Excel with the XLstats add on.

Improve this question
$\endgroup$
3
  • $\begingroup$ Welcome to CV, Sophie. You likely don't need to transform the data. But exactly how are you comparing the datasets? How large are they? BTW, you almost surely have access to free, powerful statistical software of all kinds, ranging from R to specialized applets that abound on the Web. $\endgroup$
    – whuber ♦
    Commented May 9, 2023 at 18:57
  • $\begingroup$ One data set has 270 observations, but can be split into two subcategories, one with 108 and one with 160ish, and a third one that's not super relevant. The other dataset has 130 observations. I was trying to see if the characteristics of age and length of marriage differ in a drastically significant way between the populations. I also have some binary data attached to each observation, that I wanted to compare, but I was taking it one set at a time. $\endgroup$
    – Sophie
    Commented May 9, 2023 at 19:03
  • $\begingroup$ Age is unlikely to be a problem for a standard t-test. Length of marriage might be. Researching our posts about t-tests and non-Normality will uncover a great deal of information that I hope is helpful. $\endgroup$
    – whuber ♦
    Commented May 9, 2023 at 19:05

0

Reset to default