© comScore, Inc. Proprietary and Confidential. 1 Congruent Fusion Process at a Glance Stage 1: Ensure the two universes (PMB and comScore) match Internet.

© comScore, Inc. Proprietary and Confidential.

1

Congruent Fusion Process at a Glance

Stage 1: Ensure the two universes (PMB and comScore) match

Internet users

Adults 18+

Imputation to ensure the matchup (PMB 19.4 mil; comScore 19.9 mil)

Stage 2: Determine the fusion linkage variables

age, gender, region

“subsidiary variables”: household size; household income; presence of children

Stage 3: Use weighted data from both surveys for the fusion “split weight” technique

Readership, website visiting, and product data are preserved

© comScore, Inc. Proprietary and Confidential.

2

Congruent Fusion Example: All Totals To Be Preserved

TOTALS

© comScore, Inc. Proprietary and Confidential.

3

Website respondent #1 is Male and Young, with weight 10 Print respondent D is also Male and Young, but has different weight (15)

Congruent Fusion – Step 1Find Respondents with Matching Demographics

© comScore, Inc. Proprietary and Confidential.

4

Print respondent D is split into one fractional respondent with 10 weight

and a second fractional respondent (DD) with the remaining 5 weight

Congruent Fusion – Step 2To Match Website #1 with Print D:

© comScore, Inc. Proprietary and Confidential.

5

Fused respondent 1-D has information from both databases: Information from website respondent 1 Information from fractional print respondent D

Balance of print respondent D (weight 5) is then matched to someone else in website database with the same demographics (respondent 3)

Congruent Fusion – Step 3

© comScore, Inc. Proprietary and Confidential.

6

And so on…

This “split weight” technique continues for thousands of cases using the power of specially created Congruent Fusion software.

The result…

All respondent weights in both databases are retained

Therefore, all data in both databases are retained