|
|
LETTER TO THE EDITOR |
|
|
|
Year : 2013 | Volume
: 45
| Issue : 2 | Page : 205-206 |
|
Author's reply
Ganesh Dakhale
Department of Pharmacology, Indira Gandhi Govt. Medical College, Nagpur, India
Date of Web Publication | 11-Mar-2013 |
Correspondence Address: Ganesh Dakhale Department of Pharmacology, Indira Gandhi Govt. Medical College, Nagpur India
 Source of Support: None, Conflict of Interest: None  | Check |

How to cite this article: Dakhale G. Author's reply. Indian J Pharmacol 2013;45:205-6 |
Sir,
This is with reference to the comments on our article. [1] We, thank the reader for their queries. Our response to these queries follow:
- Statistics has a wide field of application and it is not possible to condense all information in a single article. Our aim was to sensitize post graduate students in pharmacology, all information in clinical and experimental trials and to the basics of this topic. The term "Type of Data" is not absolute. [2],[3] Both terms can be used synonymously. As far as ordinal data is concerned, the differences mentioned by the authors depend on the investigator, the objectives of the study or the variable tested. An example was quoted by us in this regard.
- We used the word " converted" as synonymous to "transformed" for better understanding but scientifically speaking, it must be 'transformed' which is correctly mentioned in the later part of the paragraph. We also mentioned that this transformation can be done by taking logarithm, square root or reciprocal.
- We have neither assumed nor written that if sample size is more than 30, then it assumes normal distribution. The distribution can be tested easily by different softwares, irrespective of the sample size.
- In correlation coefficient, we have mentioned that this test measures the degree of linear relationship between two continuous variables. We agree that if variables are normally distributed then Pearson correlation test can be used and if variables do not follow normal distribution, Spearman correlation coefficient test isused.
- As for clarification regarding skew in normal distribution, what we referred in text is applicable to only ideal symmetrical curve, where mean, median and mode are all equal to each other. The first half of the curve or distribution below mean is the mirror image of the other. [3],[4] Values lying between +1 to -1 can be considered to follow normal distribution. Even small amount of skewness or kurtosis do not make the sample non normal.
- We agree with your observtions made in points 6 to 9.
- We confirmed after a literature review that even if the sample size is less than 30, Chi-square test is used after applying Yate's correction but frequency in any cell should not be less than 5. [5] If frequency in any cell is less than 5 and sample size is less than 40, then Fisher's test is more reliable.
» References | |  |
1. | Dakhale GN, Hiware SK, Shinde AT, Mahatme MS. Basic biostatistics for post-graduate students. Indian J Pharmacol 2012;44:435-42.  [PUBMED] |
2. | Nanivadekar AS, Kannappan AR. Statistics for clinicians. Int J Assoc Physicians India 1990;38:853-6.  [PUBMED] |
3. | Medhi B, Prakash A. Biostatistics in pharmacology. Practical Manual of Experimental and Clinical Pharmacology. 1 st ed. New Delhi: Jaypee Brothers Medical Publisher ltd.; 2010. p. 123-33.  |
4. | Rao KV. Measures of variation, skewness and kurtosis. In: Rao KV, editors. Biostatistics: A manual of statistical methods for use in health, nutrition and anthropology. 2 nd ed. New Delhi: Jaypee Brothers Medical publisher Pvt. Ltd; 2007. p. 51-7.  |
5. | Mahajan BK. The Chi Square Test. Methods in biostatistics. 7 th ed. New Delhi; Jaypee Brothers Medical publisher Pvt. Ltd; 2010. p. 154-69.  |
|