Adjusting means for different sample sizes


I have calculated the mean score for several variables. Because there is a big difference in the number of respondents for each variable it might be misleading to compare the means directly. I am wondering if there is some way to adjust the means to account for the diffeent size of respondents and therefore make direct comparisons more valid. So that is to say to have an adjusted mean that would be the mean had the variables all had the same number of people responding. So in the table below you can see that for var1 there were 141 people but for var6 there were only 3 people. Can I adjsut the means to account for the different sample sizes?

Variable N Mean
Var1 141 10.86
Var2 76 0.18
Var4 41 0.34
Var6 3 2.00
Var10 25 0.12


I really don't believe there is a way to 'adjust' as it were the means that you already have for your different variables. What you could do is using the standard deviations to examine how adequate your use of this measure of central tendency (mean) is. Sometimes is is better to use the mode if there is a wide dispersion in your data (shown by a large standard deviation value). It also might seem you need more data for var6.


Less is more. Stay pure. Stay poor.
I was also thinking you may present the mean with either n-value or standard error (or confidence interval).


Fortran must die
Why do you want to compare the means? I am not sure why the number of respondents [as compared to the variability or perhaps the scale of the variable] is signficant in comparing anything.