result with partial data not same with full data

[daboer]

Dear all,

I have 91 samples for 10 years, 2 variable with 3 indicators every variable.

for examples:
indicators of independent variable = a, b, c
indicators of dependent variable = d, e, f

when I test with partial data (data to test per year) test results showed indicator "a" has significant value for every year, but when I test with full data (data to test for 10 years), test results showed other indicators (b or c) has significant value.

why results showed like this? what is normal results? what this results mean?

_Thank you very much_

[Hengkov]

Hi Dave,
You have 91 samples for 10 years (it is mean: 91/10 = 9.1 sample per year => very small).
You test partial data per year, indicator 'A' significant (bias result) and full data indicators 'B' and 'C' significant because different sample size.

Because your data set is time series => test full data, no partial.

Regards,
Hengky