Using chi-square or Fisher's exact test for differences between groups [message #23395] |
Mon, 06 September 2021 13:07 ![Go to next message Go to previous message](/theme/default/images/down.png) |
TD
Messages: 3 Registered: July 2021
|
Member |
|
|
I am currently working on the 2018 Nigeria DHS and the 2019 Sierra Leone DHS using R for my statistical analyses.
I'm carrying out cross tabulation to compute the proportion of children that had received their full basic childhood vaccination between different exposure groups. I'm also using chi-square test and Fisher's exact test (if the conditions for chi-square test are not met i.e., more than 80% of expected values are less than 5 and some are less than 1) to assess the statistical significance of the differences between the different exposure groups (e.g., if 80% of fully vaccinated children live in urban areas and just 20% live in rural area how significant is this difference). However, for some of my variables Fisher's exact test does not work (an example of an error message I received is "LDSTP=18240 is too small for this problem")
Just wanted to clarify if Fisher's exact test can be used for DHS survey data or whether the svychisq function in R (which takes into account the survey design) should be used instead even if the conditions for the chi-square test are not met?
If not, please kindly suggest any further steps that can be taken.
|
|
|