Chapter 4: Appendix 4A
Fundamentals of Statistics for Human Resource Selection
After studying this material, you should be able to:
• Discuss various levels of measurement
• Discuss the measures of central tendency
• Discuss the importance of normal distribution
• Outline the steps in correlation and regression analysis
The Reality of Statistics
• Provides you with information power
• You gain the ability to:
– see latent relationships unseen by others
– predict events
• Your “people’s job” can be done better!
• You may even have fun! (once you understand statistics better)
The Bottom Line
• In today’s highly competitive, fastpaced world, the ability to analyze data gives
you a decisive advantage.
• To do your job well, whether as a manager or change agent, you must have some
basic statistics skills.
1. Levels of Measurement
• Nominal: Numbers denote only group memberships
(e.g., 1 = Male; 2 = Female)
• Ordinal: Rank order is present (e.g., 1 = Best worker;
2 = Second best worker; 3 = Next best worker; and so on)
• Interval: Many of the rating scales belong to this level
(e.g., 5 = Excellent; 4 = Very good; … 1 = Poor). The scale has no zero point, but
we can add and subtract scores.
• Ratio: The scale has a true zero point, which means that we can multiply, divide,
etc. (e.g., weight, distance, etc., are measured at ratio level)
Why is it important to know about measurement level?
1. Level of data analysis: The data analytical tools that you use depend on the
measurement level. With interval and ratio levels of measurement, you can do more
sophisticated data analysis.
2. Designing tests and questionnaires: You can always downgrade a data measurement
level. That is, you can convert a ratio to an interval or an ordinal to a nominal measure;
but, you can’t do the reverse!
• This means that measurement has to be carefully planned. 2. Measures of Central Tendency
1. Mode: Most frequent item in a given set of data
2. Median: Middle item in a distribution
3. Mean (average): Sum of all items divided by number of items
Points to Remember
1. A mean need not be actually a member of the data set (e.g., the mean of 1 and 99
is 50). In many instances, the mean does not need to be representative of the data.
(It hides more than it reveals.)
2. A median may often be a more representative measure, especially when talking
about income, age, etc.
3. The existence of two modes (a bimodal distribution) may be an indication of
• Representing an entire distribution with a single number such as a mean or
median can be problematic since individual items show considerable variation
from this measure; hence, a number of indices are computed.
– Range: Distance between highest and lowest number in the data set
– Variance: Sum of the squared deviations about the mean divided by the
number of observations.
– Standard deviation: Square root of variance; perhaps the most important