Twin+Handwriting+8

1.)  The main idea of the Twin Handwriting paper would be if you can tell who a person is by their handwriting. They took several pairs of twins and compared their handwriting by an automatic handwriting verification system. Results of the verification tests show that the handwriting of twins is less discriminable than that of nontwins  .   2.)The experiment involved 206 pairs of twins, 1236 twin tests and 1648 non-twin tests (as a control group). The goal was to establish whether or not the documents were both written by one individual. Of the 1236 tests, 824 tests had differing textual content and 412 had the same. Likewise, non-twins were tested on identical and different texts. Twins handwriting was found to be more similar than that of non-twins.

3.) This experiment was done on the handwriting of twins. There were 206 pairs of twins with 1236 tests performed; therefore I would say that this is decent science. I think they there could have been more twins simply for verification of the results of the tests. Also, there was also a control group which makes it pretty good science.

4.) // They found that it was harder to tell twins hand writing samples from each other than it was for non-twins, with an error rate of 12.91% for twins and 3.7% for non-twins. Also that it was even harder to tell identical from fraternal twin’s hand writing samples . //

 5) Interesting Facts: -The automatic handwriting verification system is comprised of three types of features- macro features at the document level, micro features at the character level, and style features from bi-grams and words. These features contribute to the final result to provide a confidence measure of whether two documents under consideration are from the same or different writers.

-The handwriting of identical twins is twice as likely to be similar than the handwriting of fraternal twins.  -The experiment used the Cedar-Fox system to analyze the documents. It examined word and line segments, investigated specific handwriting characteristics and provided character and word recognition. This system could be operated independent of human assistance which allowed it to maintain “word truth,” an absence of human bias.

- Telling the difference between the writing of twins is harder than telling the difference between non-twins, this is probably due to similar genetic influence and very similar environments.

-The error rate can be reduced from 13.22% to 6.23% by rejecting 38% of the cases for twins and reduced from 4.24% to 1.04% by rejecting 18% of the cases for non-twins. 