Hypothesis testing

Refer to the prothrombin time data on 1000 infants (PROTHROM). Here, 500 infants will be Full-Term
and 500 infants will be Pre-Term (Preemie). Select independently a simple random sample of size 30
from each sub-group of size 500. We could also say “from each population” — treat the size 500 Large
Data Set samples as two populations to sample from.
On your two samples of size 30, conduct an
appropriate hypothesis test to determine whether one could conclude (can infer about the populations
from the two samples) that the two populations differ with respect to their mean prothrombin time. Let
alpha = 0.05. What assumptions are necessary for the validity of the test? What statistics are important to report?
Consider the difference, the confidence interval, the variance or standard error, and the p-value in each
test. What are the strengths and weaknesses of your test?

\ Get descriptive statistics of the sample, assess the type or normality of the distribution (for the assumptions to be used), and check for outliers. Determine if you need to remove any outliers. In general, be conservative about removing or adjusting outliers (policies may differ among institutions). Keep a note of whatever you change. Or you could create a new variable in a separate column that will have the data with the changes. Changing outliers usually involves marking them as missing. descriptive statistics for the data should include CI, especially for the samples being analyzed, not just
for all observations. Conclusions are based on the sample.
Histograms, Box plots, and Individual Value Plots, visualizes data and are good for providing comparison
of the samples and displaying any outliers.
Plot of value by its observation number also helps to visualize data and its outliers.
Scatterplot of one population against another helps visualize potential relationships.
Use of Outlier test:
Minitab Stat > Basic Statistics > Outlier test, and for Options, choose the type of test.
Set certain outlier observations to missing — the symbol is an Asterisk (*) in Minitab
Clearly identify the observed outliers and action taken in your report.
Statement of CL (confidence level), alpha, CI, t and critical t, p-value, comparison of p-value to alpha,
and comment on whether or not 0 is inside the CI, give your interpretation.
General observation that all samples drawn are different — as evidenced by their graphic and descriptive
statistics — yet the data conclusions are mostly similar.
Based on the 95% confidence level, about 5% of the samples would not contain the true mean or the
true difference in means.
\

Take a random sample of the data (the subject number serves as the row number from the row it came from). The sample size here is usually 30 (confirm this). Sample all the variables you need at the same time, so the data in a row of a sample can be matched against the original dataset. NOTE: The sample columns first have to be given new names. They are stored in the same Minitab worksheet as the full dataset.

Again,
MUST INCLUDE discussion of the data, objectives of the analysis summary and conclusions, as well as supporting numeric and graphic software results in Minitab.

Last Completed Projects