What you describe sounds like a standard pre-test and post-test analysis. Typically these are done to check the impact of a specific intervention. You first give a pre-test to the target group. Then you make the intervention. (You might have a control group too which may not undergo intervention). After the intervention you give the same test again which is called post-test.

If the performance in post-test is better than pre-test you could say the intervention worked. If it's not the case then probably the intervention was ineffective (as seems to be the case in the paper you're looking for). There are lots papers which do pre-test and post-test analysis.

Here are a couple to get you started:  http://www.tandfonline.com/doi/abs/10.2167/le678.0  http://search.proquest.com/openview/91bc411ad432f947bf80358805ec2d1f/1

Here are a couple more where post-test performance was lower than pre-test:  http://dl.acm.org/citation.cfm?id=971328  https://goo.gl/DIsw8W

What you describe sounds like a standard pre-test and post-test analysis. Typically these are done to check the impact of a specific intervention. You first give a pre-test to the target group. Then you make the intervention. (You might have a control group too which may not undergo intervention). After the intervention you give the same test again which is called post-test.

If the performance in post-test is better than pre-test you could say the intervention worked. If it's not the case then probably the intervention was ineffective (as seems to be the case in the paper you're looking for). There are lots papers which do pre-test and post-test analysis.

Here are a couple to get you started: 

http://www.tandfonline.com/doi/abs/10.2167/le678.0 

http://search.proquest.com/openview/91bc411ad432f947bf80358805ec2d1f/1

Here are a couple more where post-test performance was lower than pre-test: 

http://dl.acm.org/citation.cfm?id=971328 

https://goo.gl/DIsw8W