to maryam/any1...plz...nbme3bl3q49 - shathabisha

[ArchivalUser]

as part of study to determine factors influencing mortality in ICU,ser crea is measured...THERE IS A GRAPH WHICH I CANT SHOW HERE....asking abt mean,median,mode...
which of th efollowing is correct....

A.mean is larger than median
B.median is larger than mean
C.median is larger than mode
D.mode is larger than median
E.mean,median,mode cant b determined coz this is not gaussian distribution...


some1 plz kindly open ur files....look at the -ve shew distribution looking graph and plz answer....thanks

[ArchivalUser]

bb
median is larger than mean .negative skew distribution
its not d because mode is least affected by changes in skew

[ArchivalUser]

If u are approaching it as How skewed it look? as the tail is pointing towards left then yes it look like negatively skewed.But according to "Frequency Distribution Graph" the y-axis represent the "FREQUENCY of cases or values" or in other words so and so # of patients(or % of a group or population) that correlate with the "MEASUREMENT VALUES" given on the x-axis.

But on the graph given in the stem the y-axis representing the "Measurement values"(serum creatinin) rather than the Frequency of appearing of those serum creatinin values and x-axis representing the # of Observations (# of patients) rather than Measurement values.

So in order to decide just by looking at the shape of the graph whether the Distribution is normal,+ skewed or - skewed we need to be looking at "Frequency Distribution Graph" and which is not the case in this stem.

Now if calculate this:

Patient #----------- 1---------2-------3------4-------5-------6------7-------8-------9--------10
creatinin values:--0.5------0.6-----0.7----0.8------1-------1------1-------2-------4---------5

MEDIAN:
Since total # of patients is an EVEN #i.e. 10 ,we get middle two values and divide them by 2
patient 5th and 6th got creatinin values of 1 and 1.so it will be 1+1 divided by 2=1 (Median)

Median= 1
Mode = 1 (most frequently appearing value)

Mean = (.5 +.6 +.7 +.8+ 1 +1 + 1 + 2 + 4 + 5) divide by 10 ( total# of patients)
........= 16.6/10
........= 1.6

so
Median=1
Mode =1
Mean =1.6

Mean>Mode

Now if we were to place Measurement values(creatinin) in a Frequency Distribution graph then the serum creatinin value of 1 mg/dl appeared in 3 patients and all other values appeared only once ,so there will be one Tall Bar(representing 1 mg/dl appearing 3 times) and remaining seven Smaller Bars of same height cz all seven have different creatinin values but each appeared only once.

Quick approach to the above Headache:

look for Median as it is easy to figure out which is = 1
look for total # of patients =10
start adding measurement values starting from higher number as in this case 5+4+2+....
and once we cross 10 (which is the total # of Observation) we know the MEAN is going to be more than 1 (cz Mean=add all values divide by total # of observation)


it is posted by sarim