If correlation between two variables for two distinctive groups (for example students and employees) is same, does it mean results from one group can be generalized for other group?
In case, I am thinking on wrong lines, let me put it in this way. When can results from one group be generalized for another group (For example, results from student's survey generalized for general public or employees of an organization)?
Thanking in anticipation.
A result from a sample representing a certain population can only be generalised on the population it represented. Hence, in order to be able to make a generalisation on a bigger population, the sample should be taken from that bigger population either through random, stratified, cluster or other recognised sampling methods.
A result from a sample representing a certain population can only be generalised on the population it represented. Hence, in order to be able to make a generalisation on a bigger population, the sample should be taken from that bigger population either through random, stratified, cluster or other recognised sampling methods.
Aminuddin is right, and it makes life difficult for those who run experiments with students and then claim the results extend to other populations. This does not mean some results would not come out the same or similar, but we cannot claim that without checking with a larger/different sample that includes those to whom we wish to generalize. I tAdd your answerhink of student-based results as "exploratory:" they can give us ideas about what to check for,, much as pilot studies do. But then it is incumbent on us to verify.
Ali -
As stated above, you do need to collect a proper sample from each "distinct group." However, you also mentioned a "...correlation between two variables for two distinctive groups..." and I wondered how you obtained this correlation, and exactly what you meant. That would seem to indicate a link between responses for the two "groups." For continuous data you could obtain scatterplots for these (x,y) data pairs between "groups," and use regression to relate one to the other. However, I do not think that is really what you meant. However, if that is what you meant, more could be said about it.
Cheers - Jim
No. Nothing can be generalized. In my experience, this generalization has resulted in greater fraud, which the government is incapable to solve. For example, no leader is equal from one environment to the other environment or from one organization to another. Each environment is different. It is more difficult when it comes to the government. The contemporary leadership knowledge is generalizing leadership across sector. The impact has been, severe leadership and democratic deficits. In other word, leadership and democracy do not exist in the New Public Management.
The generalization of the findings from one sample can only be done for the population of similar character. However, in recent times statisticians are objecting generalisation of results for any population. They opine that the results for the sample can only be generalised for the sample only.
- Badges
- Science topic
- Similar topics
- Statistics
- Survey