Statistic do not speak for themselves
Milton Friedman
Statistics is the grammar of science
Karl Pearson
Statistics : The Mathematical Theory of Ignorance
Morris Kline
join di youtube Mellyna Eka Yan Fitri ya...
kunjungi : https://www.youtube.com/@mellynayan
Selasa, 15 April 2014
Minggu, 13 April 2014
TALK A BIT ABOUT STATISTIC
Statistic is divided by
two methods, there are descriptives statistic and inference statistic. In here,
descriptive statistic, we learn about how to describe or display collection
data by using table, graphic and diagram. We make pie chart, frequency table,
stem and leave diagram, histogram, ogive, etc. Descriptive statistic is
statistical methods that summarize and describe a collection of data without
general conclusion about population. And, inference statistic is statistical
methods that give conclusion about population, in here we use hypothesis
statistic, alpha (significance level), degree of freedom, etc.
In thesis or journal or
research for quantitative, we often use this two methods, descriptive to describe
our collection data, may about respondent like gender, education, or some
specific of respondent. And, for inference method we want to make conclusion
about population from sample that we are taken.
Inference methods, it
divides by parametric and nonparametric statistic.
What are they difference?
The difference between
parametric and nonparametric are data characteristic. Parametric, data
collection must have the normal distribution, it has interval or ratio scale.
But for Nonparametric is not like that. Nonparametric statistic methods,
collection data has free distribution (not the normal distribution) and it has
nominal or ordinal scale. Parametric method is more complicated than
nonparametric .
Parametric method, we need
to fulfil the assumption of data for the main test, such as regression,
correlation, difference mean test (t-test), etc. There are classical assumption
such as normality data, heterocesdastisity, independence error,etc. If we cannot
fulfil this assumption, we can do more alternative method such as
transformation of collection data or we must change to use nonparametric
method.
In research, we focus to
want to know about description, association, and comparison of one, two or more
variables. And we have more statistical test to know about it.
For example of description
are Binomial test, Run Test, t-test, etc. And, the association are Regression
Linear, Regression Non Linear, Pearson Correlation, Spearman Correlation, Structural
Equation Model, Path analysis, etc. The comparison : ANOVA (one way or two
way), Kolmogorov Smirnov, Median Test, t-test independent, t-test paired
sample, etc.
And to do that analyze we
can do manual by using formula of statistic or use statistical software. For
efficient and accurate result, we use statistical software. If we have more
data, we need more time and more analyze
to do it. And, maybe our analyze will not correct or bias or more error if we use
manual analyze.
We can get data from
observation, interview or surveys. We can get quantitative and qualitative
data. Quantitative data is numeric data, but qualitative is nonnumeric data.
When I choose statistic in
bachelor degree as my subject, the reason was not because I like statistic, but
I hated statistic. When I was in Mathematical Department to study, I like pure
mathematic that study more about Calculus, Algebraa, Analysis, etc. Statistic
made me curious and I want to know why I didn't like it. But Alhamdulillah, I
can pass it with good grade.
My last thesis in bachelor degree was about Customer Satisfaction in
Government Company. I used combine of three analyses, there was factors
analysis to know what factor influence customer satisfaction, discriminant
analysis to know how was different of customer satisfaction in each area and
the last is Parasuraman theory about Gap Analysis to know what percent of
customer satisfaction about its service.
To know more you can
download in here.
Postingan
