Without knowing the issue that you are looking at, 3.6% and 5% seem not to deviate very much. But you may consider calculating the 95% C.I. within your population to get a better sense of the data.

Thank you very much for your answer!! I think that they don't deviate that much, too, I was just trying to find a maximum acceptable percentage somehow. Either way, thank you for your response!

Kalliopi,

There is not specific reason to expect that every single group show similar epidemiological profiles. Comparisons of epidemiological analyses with previous studies must be performed in recognition of differences/similarties in contextual factors (ie. health system characteristics, risk profile, demographic structure, etc.).

The health outcome being analyzed is also relevant. For example a 5% difference in prevalence between two populations would mean different things depending on considerations as if it is letal, if it is easily preventable, or its impact on burden of disease, etc.

Those considerations lead to an epidemiological analysis of prevalences (for example), beyond the quantitative comparison of two proportions (even if prevalence is an specific kind of proportion).

Diego Fernando Rojas Gualdrón Thank you so much for your response. I finally continued to Discussion section with contextual factors, as you said. Thank you again for your time!!

Dear Kalliopi:

Without having enough information from you, it seems that you are trying to compare your result with other people results, in terms of prevalence.

Being so, the first thing to know is that there is not so call "acceptable deviation". The second point is that when you work with a sample, normally there is a random variation that create differences among study results, and finally, you have the opportunity to estimate the confidence interval for your prevalence, and compare this estimation with others people estimation to see how close or far is your results from them.

Some things that you need to consider if you are interested in comparing the prevalence of two or more populations are the standard population used to adjust the prevalence and the statistical and biological significance of those differences.

The difference in the prevalence between populations can be affected by differences in the distribution of factors like age or gender between those populations, that is the reason why age and sex adjusted prevalence are reported in the articles. Nevertheless, to make an effective comparison between populations, the prevalence of each population should be adjusted using the same standard population.

Another important aspects to consider are the statistically and biologically significance of the difference between the prevalence. To evaluate if your prevalence is statistically different from other you can calculate the confidence interval of those proportions and if they doesn’t overlap you can say that the prevalence are significant statistically different. In the other hand, the biologically significance of the difference in the prevalence is related to the effect that this difference is going to have in the health status of the population, in long term projections, in money saved, etc.

Dear All:

I hope this article helps.

Regards,

It depends of the seriousness of the Type I Error associated to the problem of research. Usually 5% is accepted.

Kalliopi Tsakpounidou,

Depending on the seriousness of Type I Error, the level of significance varies. Usually it is 5%. But it may be 1% or less if the degree of seriousness of Type I Error becomes higher.

Thank you for recommending my earlier answer.

Dear Kalliopi Tsakpounidou,

Depending on the seriousness of Type I Error, the level of significance varies. Usually it is 5%. But it may be 1% or less if the degree of seriousness of Type I Error becomes higher.

Thank you for recommending my earlier answer.