PSM

Biostatistics

Question
56 out of 76

2 drugs were given to population and response was noted are follows.
 Drug Cured Not cured A 510 100 B 110 50 Total 620 150

Which study design is most suitable?

 A Student T test B Paired T test C Chi square test D Bland and Altman test

Ans. C Chi square test Ref: High Yield Biostatistics, Pg: 40 - 51

Since the data is qualitative and the variable is dichotomous nominal as the response noted is in terms of cured and not cured chi square test is most suitable because the question has given the proportions for comparison.

Chi-square test:

Used for qualitative data, which is in the form of proportions (percentages). e.g. – comparing incidence of a disease in 2 groups of children with one group vaccinated and the other not, OR com­paring incidence of anemia in 2 groups of pregnant women, one taking iron supplements and the other not.

Tests of Significance:

Once a hypothesis is made, the next step is to test it by means of various tests (known as tests of significance). These can be broadly classified as:

Parametric tests: These are - t test, z-test and ANOV A (or F-test) These share few common features:­

i. These hypothesis refers to certain population parameters which is the population mean (in case of t and z tests) or the population variance (in case of F test)

ii. These hypothesis concern interval or ratio scale data (e.g. weight, height, BP etc)

iii. They assume that the population data are normally distributed.

Nonparametric tests: e.g. - Chi square test

i. They do not test hypotheses concerning parameters.

ii. They do not assume that the population data is normally distributed hence - distribution free tests

iii. They are used to test nominal or ordinal scale data.

T -test: (student t-test)

As for all parametric tests, it is used for quantitative data which is normally distributed in the population

(i.e. continuous data such as weight/height etc).

Steps:

1. State the null and alternate hypotheses

2. Select decision criteria α (level of significance)

3. Establish critical values of 't' associated with this criteria (using degree of freedom and compare t value from table of t scores)

4. Draw a random sample from the population and calculate it's mean

5. Calculate standard deviation and standard error of the mean.

6. Calculate value of t that corresponds to mean of the sample.

7. Compare the calculated value of 't' with the critical value selected above and then accept or reject the null hypothesis.

T -test can be unpaired or paired:

Paired t-test is a special type of t-test used when the data is "paired" e.g. - blood glucose before and after taking a drug OR

Pulse rate before and after exercise (in the same group).

Z-Test:

1. Involves the same steps as t-test
2. Used when the sample is large (n> 100)
3. T-test though is more important as there is no ­situation where as Z test can be used and a t-test cannot be but the vice versa can occur.

ANOVA:

Whereas a T test is appropriate for making just one comparison (between two sample means, or between a sample mean and a hypothesized population mean), when more than one comparison is being made (i.e. when means of> 2 groups are being compared), ANOVA (analysis of variance) is the appropriate technique.

ANOV A can be:

1. One way ANOVA - when the groups differ in terms of only one factor at a time.

e.g.: 3 different drugs given to 3 groups of people to compare their effects.

2. Two way ANOVA - If the groups differ in 2 factors at a time

e.g. - If in the above example if each group consists of male and female patients, the researcher may also want to make comparisons between the sexes, making a total of 6 groups.