What Type of Test should I use? Help Needed

#1
Hello,
I am doing research on the perceived rate vs actual rate of hallmark symptoms that accompany a certain disease. Essentially, I want to know if it is possible to compare this perceived vs actual rate of symptoms and see if this contributes to a later diagnosis of said disease. Is this statistically possible, and if so, what test should I use?
Thanks to anyone who is willing to help.
 

Karabiner

TS Contributor
#2
Please tell us the sample size, and the range of the dependent variable (rate of hallmark symptoms).

With kind regards

Karabiner
 

Karabiner

TS Contributor
#4
I want to know if it is possible to compare this perceived vs actual rate of symptoms
You can create a Band-Altman plot for analysing the agreement. You can calculate the 103
differences between peceived and acual rate, then describe mean, standard deviation,
minimum, maximum and median of these differences, and perform a dependent samples
t-test (answering the question whether the mean perceived rate is significantly higher or
lower than the mean actual rate).

With kind regards

Karabiner
 
Last edited:
#5
You could create a Band-Altman plot for analysing the agreement. You can calculate the 103
differences between peceived and acual rate and describe mean, standards deviation, median
of these differences, and perform a dependent samples t-test (answering the question whether
the mean perceived rate is significantly higher or lower than the mean actual rate). And you
can create an X-Y scatterplot between the 2 measurements.

With kind regards

Karabiner
This is very helpful. Thank you so much.
 

Karabiner

TS Contributor
#6
(...) And you can create an X-Y scatterplot between the
2 measurements and calculate a Pearson correlation coefficient.

and see if this contributes to a later diagnosis of said disease.
You say that you want to know if the comparison contributes to a later diagnosis.
I am not sure what that means. Could you perhaps explain this a little bit more?
Do you have diagnostic data which are to be included?

With kind regards

Karabiner
 
#7
(...) And you can create an X-Y scatterplot between the
2 measurements and calculate a Pearson correlation coefficient.


You say that you want to know if the comparison contributes to a later diagnosis.
I am not sure what that means. Could you perhaps explain this a little bit more?
Do you have diagnostic data which are to be included?

With kind regards

Karabiner
What I wish to do, if possible, is to statistically determine whether this theoretical gap in between perceived and actual rate of symptoms causes doctors to diagnose this disease later because they rely on the presence of these symptoms before actually pushing for diagnostic tests. My diagnostic data would include the prevalence of these symptoms in patients who have been diagnosed with this disease, but from there on, I am not sure what to do. Is there perhaps another set of data I can collect to determine whether or not these perceived differences lead to doctors not realizing the disease is present and therefore failing to diagnose it? I am not quite sure how to get from my collected data (perceived and actual presence of symptoms) to my endpoint goal (seeing whether or not this disparity can cause a late diagnosis).

Best,
Sean
 

Karabiner

TS Contributor
#8
I am afraid that I have no clear idea abiout your study
data and your study goals, so unfortunately I am of
no help here.

With kind regards

Karabiner