They show different things; neither is "superior" to the other. Their appropriateness depends on what you are trying to show.
It sounds like you are looking at a given population, and you want to know the difference between estimating its standard deviation, and, say, its interquartile range, which each measure the population dispersion.
The standard deviation is one of the moments of a population. If you have an idea of a population shape, then mean and standard deviation might be enough information for your purposes, but you may also want to estimate skewness and kurtosis as well.
[You might also be interested in Chebyshev's Inequality.]
[If you have estimates of mean and standard deviation, then you can estimate confidence intervals for the estimated mean by estimating its standard error and making use of the Central Limit Theorem to assume a normal distribution of the mean estimates with a large enough sample size, or by using other distribution-types.]
If you lack information about the shape of your population distribution, then instead of a standard deviation, you may just want to rank your data from smallest to largest, and look at the "interquartile range." That means considering "percentiles."
The standard deviation is best associated with the mean.
The interquartile range is best associated with the median.
Cheers
Morning, Abdul-Kadri,
I think these two statistics are a bit different. It depends on what is the focus: independent or your dependent variable. SD might show you in how much homogenious your variable is. It can happen, for example, that males and females are equal in something based on the average scores. However, if the standard deviation is high, then you can cay that male behave different from females, since one of the groups shows a much higher dispersion.
Variance explained is a measure which shows in how much you independent variable explains the dependent variable. It is a relationship showing that the change in independent varaible is concerned with a related change in the dependent variable. SD in this case might have less influence on the interpretation of your results.
If you mean no the variance explained but the variance in your indepdendent vairable than yes, it will show the quality of collected data as James wrote. It can indicate that the data are collected wrong or that there is a big discrepancy in the population, that the variable is not normally distributed, or even, that the population is defined wrong. In this event variance and SD indicate almost the same.
Best,
Eugene
I strongly agree with James R Knaub ,Stephen Politzer Ahles and Yemen Bogodistov
The variance is in squared units whereas the standard deviation is in the original units. For example, the variable Age will be age squared and not easy to interpretation wheres
Standard deviation is in the original units. The range is easy to interpret although it ignores in between values
- Badges
- Science topic
- Similar topics
- Psychometrics
- Measurement