Know-How Exchange : Research/Metrics

I'm assuming that your 10-point scale is made up of something like 10 is highly satisfied and 1 is highly dissatisfied - similar to your present survey. If this is the case, then you can recast your older data in a five point scale by combining the results from the top two categories, third and fourth, fifth and sixth, seventh and eighth, and ninth and tenth. That gives you a pseudo-five point scale to compare with your current survey. This works if the questions are matched up for content and wording is similar.

I hope this helps.

Wayde

In the absence of other data, I agree that you should compare the top-2 boxes of the 5-point scale with the top-4 boxes of the 10-point scale.

However, you may care most about how employee satisfaction affects a performance metric such as absenteism, turnover or productivity. For customer satisfaction, research has shown that customers who check the very top box are much more (5 times more) likely to stay loyal to the company. Employee satisfaction and retention may show a similar relationship.

Hence my recommendation: check the relation between satisfaction and your performance metric. It may be that top 3 boxes of the 10 point scale correspond to the top 2 boxes of the 5 point scale in their relation with performance

Hi,
Basic thing you can do with this is to use 2 boxes of 10 point scale with 1 box of 5 point scale. It will work for getting average and static data to you.
Or else you can correspond to the features in the box. Combine factors which are similar to the other one.

Tanuja Thapliyal
Spreading Smiles :)

Hi!

First I think that you are better of f now using the 5 point scale. I find that for satisfaction it lends to easier data analysis.

In general I would suggest you don't compare the old data and new data. There is a different psychology involved with fewer options.

This would be my suggestion (bear in mind I don't know your industry or even if this is a service or product) so you might want to modify:

Old Scale => New Scale
1 => 1
2 => 1
3 => 1
4 => 2
5 => 2
6 => 4
7 => 4
8 => 5
9 => 5
10 => 5

I would not assign anything to 3, becuase on the 10 point scale you did not actually give them a perfectly neutral choice.

The best thing to do is to chart this data after you have translated it and ensure you are getting the same trends in satisfaction. It's not going to be perfect, but it should be close.

Cheers,

-Christian Vanek
CTO, SurveyGizmo

As Wayde says above, you can use the combination methodology for comparing. Keep in mind, though, that it may make a difference if you've anchored both scales, and if each point in between has been named as well. For example, the comparison will work best if the old scale was anchored with extremely satisfied to extremely dissatisfied. In addition, you might want to take the top-3 box from the old 10-point system to compare with top box in the new, as well as top-2 box from the old. Track both measurements for a few time periods to see if there is a significant difference.

Wayde, Koen, Christian and Joy have all made great points - however you do this, make sure you annotate the results and reports clearly so any reader understands the difference between the 2007 and 2008 methodologies.