PHR/SPHR Study Group Module I
So, we left the discussion of Descriptive Statistics to blogging! We commonly use statistics in HR as you will see through this blog (I hope). Statistics provides an indicator of variation around "central tendency" values.
Blog back and tell me what you think that means.
To understand Descriptive Statistics we need to look at a few words.
Range is the distance between highest and lowest scores. Percentile is the specific point that has a given percentage of cases below it and the standard deviation is how much scores are spread out around a mean.
In what aspect of HR do we commonly look at range?
Measures of Association - Correlation - shows the relationship between two variables.
What are variables?
Measures of Association - Regression - refers to the effect of one variable to the other. If two variables are related and one changes, the second will change. This relationship is represented by a regression equation. Regression equation is used to construct a regression line (usually a straight line) on a scattergram. (You can find a sample of a scattergram in most Excel programs).
When might HR use a regression analysis?
Inferential Statistics - form a conclusion by studying a sample of the population. Population is the entire group (all employees). A sample would be part of the employees (20 randomly chosen employees).
Normal distribution is the expected distribution given a random sampling of a large population.
For what purpose would we (HR) use inferential statistics?
Reliability - Is the information reliable? is the measuring instrument consistent? Are the results consistent? For example, two tests are given to an applicants. Applicant A passes both tests at 95%. Applicant B passes one at 75% and the other at 95%. Applicant A test scores are more consistent and therefore more reliable.
Validity - Is the information valid? Did it measure what was intended to measure? A reliable instrument is not always valid. A valid instrument is always reliable.
Give an example of something we do in HR that is reliable but may not be valid.
Instructions for blogging: Just click on the "submit a comment" link below! Share your answers!
It seems like Central Tendency is where most of a group have the same tendencies--most people have a certain score, and there are few who are outside of the norm. For example when we train classes we track the scores to see where the "middle" is, the median, mode or average. When performance reviews are written, we track where each employee falls within the categories, i.e. quality of work, quantity of work, attendance, etc. Each department can see their middle, or performers in the central tendency, then those above and below. This tracking helps the production department supervisors to develop standards over time.
When I think of range, I think of salary or wage range. Of course test scores makes me think of ranges.
Variables are characteristics or factors that change an outcome of a situation, I think. Age, gender, time of day are things that vary. So if I were measuring turnover, it would look different at different times of the year. We are very slow in March and so turnover is higher, and very busy in January, and we might have lower turnover. The variable is the time of year. In my turnover example, regression analysis might be applicable. Because January is a very busy month, sometimes turnover is created because people get too much overtime. This would be interesting to see on a scatter gram.
We might use inferential statistics in HR to see if people are using a benefit or maybe a skill taught. For example, if we've taught someone how to lift properly, we could observe a sample of the population. For example, we have 50 trained people. We observe half of them, or 25 people. 20 of them are lifting properly. So we conclude that 80% have learned and can demonstrate how to lift safely.
Something we do in HR that is reliable but not always valid could be essay testing for promotions. It could be a reliable test if people answer the "right way" according to correct answer samples, however each rater could see the answers differently and score differently.
I think I am on the right track with these but please, let me know if I am not!
This makes sense to me. It also gave me "real" examples of how we would use these concepts. Thanks.