CREDIBILITY

Header: Rahul / Saurabh / Swati / Sonal / Yogesh

Date: 09-12-06

Credibility

Our claim/argument:

"Our raw score/percentile score is a fairly good indicator of the employability of a candidate"

But, as of now, we do not have any independent/external proof for this claim. Till such time, our claim will remain "suspect/debatable/arguable".

A distinct possibility (to prove our claim), will emerge during next 6/12 months.

As follows:

Let us say, during next 6 months, some 5000 interviews get conducted using our IIT (-only those where PEN is used; not temporary).

For each of these 5000 interviews, we would have captured/stored

"Interview Score %"

For each of this candidate interviewed, we would also have in our database,

• Raw Score
• Percentile

on interview-date

Header: [2/2]

Now, with this data, we will plot following SCATTERGRAM:

(Graph Description):

• Y-axis: Raw Score (or Percentile?) — a measure of our software’s assessment.
• X-axis: Interview Score % $\rightarrow$ (—a measure of independent/external assessment of candidates, carried out by HR mgrs).
• Visuals: A scatter plot with points clustered around a "Line of Best Fit", bounded by a "BAND OF DISPERSAL".

If the observed "Band of Dispersal" is quite narrow (around "Line of Best Fit"), then there is a high correlation between:

• Raw Score & Interview Score % } High "Co-efficient of Correlation"

In which case, our software's logic stands proved — Our claim stands vindicated! Our credibility will go up dramatically! Once we capture the data, plotting is no big deal.

(Signed)