# Statistical significance

#### Jerryjosh1

##### New Member
There are 42 patients;

23 have respiratory allergy
By 12 patients allergy is excluded
7 are suspect

Is the frequency of respiratory allergy statistically significant?

What type of test should we use in this example?

#### Karabiner

##### TS Contributor
I am sorry, but the question does not make sense.
One cannot say that a parameter or some
aggregated data are by itself "statistically significant".
You have to specify a comparison for the working
hypothesis first, such as "the means of the 2 status
groups are the same [in the population]" or "the
correlation coefficient is zero [in the population]"
or "the proportion of black-haired people is higher
than the proportion of people with other hair
colours". So, which comparison(s) do you want
to make here?

With kind regards

Karabiner

Last edited:

#### Jerryjosh1

##### New Member

It's kind of difficult for me to properly use statistics language because of my medical education. I have recently studied only the basics, but usually I recognize type/way of testing hypothesis (frequencies chi xtest, paired group t test etc.). Is there any chance to simply use these numbers in formula for any specific test or we need more data? Maybe "statistically significant" was bad choice of words.

#### Karabiner

##### TS Contributor
You can use these data if you wish, but n=42 is a bit sparse.

With kind regards

Karabiner

#### noetsi

##### Fortran must die
How is the data measured. What is your hypothesis you are testing, your null. You will have a hard time generalizing with only 42 cases and low power, but that is a different issue.

#### hlsmith

##### Less is more. Stay pure. Stay poor.
What do you hope to do with your results?

#### css

##### Member
I agree with all previous comments... there is no proper answer to an undefined question. In any case, since you have count data and 3 different situations (allergy, suspected, no-allergy), a chi-squared should provide you an answer (although, as mentioned in previous answers, the small sample size compromises any reached conclusion)