Common Errors

Example: There was a 2% difference in the mortality rates of group A (10%) and group B (12%).

Fixing the Problem: There is a 20%, and not a 2%, difference between the mortality rates of group A and B; but there is a difference of 2 percentage points (the difference between two percentages is expressed as percentage points).

Example: The interval between the occurrence of the first and second cancers was 26.5 ± 57.3 months (mean ± SD) in male patients in treatment group 1 (P < 0.05; Student's t test).

Fixing the Problem: Because the standard deviation (SD) exceeds the value of the mean, the range of values for one SD would include negative numbers, a meaningless situation for a time interval. This indicates that these data are not normally distributed, and that the mean and SD are not valid measures of central tendency and dispersion, respectively. The median and interquartile range should be used for such nonparametric data. In addition, because the Student's t test is intended for parametric data, the statistical analysis is invalid.

Example: Enchondromas are typically hypercellular, while normal hyaline cartilage shows an even distribution of chondrocytes that are hypocellular with abundant cytoplasm.

Fixing the Problem: The term "hypocellular" is usually associated with an abnormal decrease in the number of cells present and would not be used to describe normal hyaline cartilage. Further, chondrocytes, being cells, cannot be hypo- or hyper-cellular, but tissue such as cartilage can be described as such.

Example: The average symptom score decreased gradually from 9.6 to 5.0 during the 8-week treatment period.

Fixing the Problem: According to the methods section, symptom scores were assessed twice, once before and once after, the 8-week treatment period; the term "gradually" cannot be used unless there were multiple, distinct measurements obtained.

Example: This study demonstrates that the use of drug X shortens the infectious course.

Fixing the Problem: Because this was an uncontrolled study, this statement cannot be made without prior reference to historical control data describing the natural course of this infection.

Example: Three of the four cases presented with pathologic fractures.

Fixing the Problem: Because the abstract of this paper states that all patients had pathologic fractures, there is an apparent inconsistency in the data presentation. There is also a problem of usage; "cases" cannot "present with fractures," only patients can present with symptoms.

Example: The frequency distributions were comparable throughout each observation period.

Fixing the Problem: "Comparable" means only that the distributions were able to be compared, and does not suggest that they were similar.