M17 Parameter Estimation SD Video Lecture | Mintab Tutorials for Beginners - Data & Analytics

I've moved on to a job with a
pharmaceutical company where I have been
asked to find out there's a new drug
used to control fevers and they want to
know the variability of how long it
controls the fever for so not the mean
amount of time that it controls the
fever for but really the standard
deviation of that time how much
variation is there in how long it
controls the fever in an adult so I've
given it to a sample of adults I've
given the drug to a sample of adults who
had fevers and measured how long in
hours that fever was kept down how long
it effectively curbed the fever and now
I want to find an estimate a confidence
interval estimate of the standard
deviation of how much variation there is
so we are going to find find a 95%
confidence interval of the standard
deviation of the time of effectiveness
of this drug alright so how are we going
to do that now in Minitab I'm going to
be doing my parameter estimation by
going to stat basic statistics and I
have to find the right thing to use
you'll notice there's nothing here about
standard deviation standard deviation is
not an unbiased estimator it is not
something that I can estimate in the
same way however variance is and
standard deviation is the square root of
variance or variance of standard
deviations squared so if I can estimate
things about variance then I can take
square roots to estimate things about
standard deviation so here I'm actually
going to be working on variances
variance is an unbiased estimator
I have only one sample so I'm going to
one variance all right again much like
other hypothesized parameter estimation
that you can do in Minitab you could
have your samples actually in a column
or you could just give it certain sample
information some summarized data you
could give it either the standard
deviation of the variance right now I
actually have my data in a column itself
so I'm going to say I have one or more
samples each and a column perfect what
column is it it's time of effectiveness
and I'm going to click on my options I
want to work at 95% great and again
since I'm doing parameter estimation I
want to leave this as a not equal that
too in the alternative hypothesis
that'll actually come up when you do
hypothesis testing I'm going to say okay
there okay here and what happens is that
Minitab now has calculated those
confidence intervals for me so the
standard deviation itself was 0.133
hours that would be my best point
estimate but as far as the time of
effectiveness goes Minitab was actually
done it by two methods the chi-squared
method and the bonnet method chi-squared
is what we're learning in our book
chi-squared can only be used when you
know your data is normally distributed
which is actually it's telling you right
there the chi-squared method is only for
the normal distribution the bonnet
method is for any continuous
distribution so sometimes we use the
bonnet method but we like the
chi-squared method when we know
something's normally distributed let's
assume the time of effectiveness is
normally distributed we can use the
chi-squared method then we're going to
say with 95% confidence the standard
deviation of the time of effectiveness
is between zero point one zero four
hours and zero point one eight five
hours and we have done it we have
estimated our standard deviation