I have a data on the sales of two companies for 10 years and there is continuous growht in the sales for the 10 years in both the companies.
I want to test whether there is any significant difference between the two companeis in terms of their sales.
is it appropriate to use independent sample 't' test?
will the results of the test be spurious because of non-stionarity of the data series? if so, what is the alternative solution for this problem?
Look for "ANCOVA" (analysis of covariance).
This is a test on the difference in slopes of two regression lines.
It may be neccesary to transform the sales to get a linear increase over time (could be a logarithm as I suspect that the growth is exponential).
Hi Srikanth,
A t-test would not be appropriate for this data. An assumption of the t-test is that the observations in each population are independent samples from the larger population. Since this time-series data seems to be auto-correlated (the value of 2015 sales depends on 2014 since there is a pattern of growth) this assumption is violated.
Best,
Aran
I would like you to use cumulative frequency approach by adding the sales cumulatively since the sales depends on research and advertisements but l am afraid your n (the sample size) is very small for time series analysis.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis