« Termination As a Beginning | Main | Terrorist Attacks ---- Are Your Employees Covered? »

12/29/2015

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

So, we are talking about a general industry survey, either for a particular industry with companies in different sizes or for a region with even differing industries). The most important contributor for data accuracy is undoubtedly the job matching exercise. You might not have control as a participant, but whoever is overseeing the survey can do a lot to mitigate the anomalies. It goes without saying that this would be easier for a long running survey than one that just started a year or two ago.

In addition, most surveys would allow you to produce a custom report with a selected peer group. This would have to adhere to certain rules to avoid any litigation. Overtime, you would be able to achieve a certain level of comfort vis-à-vis the accuracy of the data.

I think, that one of the major fallacies with these surveys is the focus on actual pay rather than on pay policies. Actual pay fluctuates and more so if you are looking at a specific job with fewer occupants. I would prefer to know what is the average basic and total pay policy for a certain job or grade in addition to the actual pay. Comparing the two would provide a higher level of accuracy.

We could also minimize the inaccuracies if we were to clearly define what is included in total pay and what is not to be included.

At the end, you are right. As much as we could do individually to improve the accuracy of the data, we must always take with a grain of salt. It will never be 100% accurate as long as it is subject to human interpretation.

Hi Chuck!
Here's a few other issues with surveys...
* Often the least experienced compensation person is filling out the survey
* Companies dropping out or joining the survey can impact individual jobs with smaller sample sizes
* Some surveys report only actual data while others provide target and actual
* Some surveys leave out important information such as LTI data
* Practitioners may interpret LTI data as part of TDC for all when only a small percentage of participants are actually eligible or receive it

...are there other challenges we can think of?

Chuck spoke well on a popular topic. Here's another take on survey implications: http://www.compensationcafe.com/2015/01/why-pay-surveys-rarely-agree.html. Seems like there are more ways to mislead than there are to properly enlighten, perhaps because the anxious demands for absolute certainty are impossible to satisfy. In the real world, statistical reliability involves standard error. And the findings may be irrelevant, too.

Errata: a final period disables the link I posted above. Remove that period, and it works again.

The comments to this entry are closed.