Systems, methods, and articles of manufacture to measure online audiences
Methods and apparatus to monitor media content at a content display site are described. An example method includes assigning a first set of weights to each of a plurality of panelists based on a set of collected characteristics for each person of a subset of unmeasured persons at unmeasured locations, the subset of unmeasured persons being the unmeasured person for which the collected characteristics are know selecting a subset of the panelists based on the first set of weights, each panelist of the subset of panelists selected to be representative of an unmeasured person of the unmeasured persons, re-weighting the subset of the panelists based on estimated characteristics of the unmeasured persons at the unmeasured locations to generate a second set of weights, generating a virtual panel.
1. An apparatus comprising:
a daemon to:
obtain measured panel data from one or more metering devices via one or more wireless communications, the panel data including accesses to web pages from first panelists; and
obtain census-based page views via one or more wireless communications, the census-based page views including accesses to web pages from unmeasured locations;
a negative binomial distribution calculator to:
calculate page view demographic distributions based on the measured panel data associated with the first panelists;
distribute the census-based page views according to corresponding demographic distributions; and
estimate a first audience estimate associated with the unmeasured locations based on the distributed census-based page views and the measured panel data;
a virtual panel generator to:
generate a virtual panel based on the first audience estimate to enable more accurate estimation of traffic from the unmeasured locations by weighting second panelists associated with the unmeasured locations and reweighting a subset of the second panelists; and
estimate a second audience estimate associated with the unmeasured locations based on the virtual panel, the second audience estimate including demographic characteristics; and
a hybrid online audience measurement system to calculate an online audience based on a measured location audience and the second audience estimate.
2. The apparatus of claim 1 , wherein the demographic distributions are associated with at least one of (a) day of the week and (b) day part of the day of the week.
3. The apparatus of claim 1 , wherein each page view is associated with a Uniform Resource Locator (URL) pattern.
4. The apparatus of claim 1 , further including an activity fulfiller to determine additional activity not represented by the virtual panel, and wherein the hybrid online audience measurement system is to calculate the online audience also based on the additional activity.
5. The apparatus of claim 1 , wherein the virtual panel generator is to weight the second panelists associated with the unmeasured locations based on a set of known collected characteristics of unmeasured persons at the unmeasured locations.
6. The apparatus of claim 1 , wherein the first audience estimate is determined by evaluating a negative binomial distribution.
7. The apparatus of claim 6 , wherein the negative binomial distribution calculator is further to generate census sessions based on the page views and the measured panel data, and wherein the negative binomial distribution calculator evaluates the negative binomial distribution based on negative binomial distribution parameters determined from the census sessions.
8. A method for calculating an online audience, the method including:
obtaining measured panel data from one or more metering devices via one or more wireless communications, the panel data including accesses to web pages from first panelists;
obtaining census-based page views via one or more wireless communications, the census-based page views including accesses to web pages from unmeasured locations;
calculating, by executing an instruction with one or more processors, page view demographic distributions based on the measured panel data associated with the first panelists;
distributing, by executing an instruction with the one or more processors, the census-based page views according to corresponding demographic distributions;
estimating, by executing an instruction with the one or more processors, a first audience estimate associated with the unmeasured locations based on the distributed census-based page views and the measured panel data;
generating, by executing an instruction with the one or more processors, a virtual panel to enable more accurate estimation of traffic from the unmeasured locations, the generation of the virtual panel based on the first audience estimate including weighting second panelists associated with the unmeasured locations and reweighting a subset of the second panelists;
estimating, by executing an instruction with the one or more processors, a second audience estimate associated with the unmeasured locations based on the virtual panel, the second audience estimate including demographic characteristics; and
calculating, by executing an instruction with the one or more processors, the online audience based on a measured location audience and the second audience estimate.
9. The method of claim 8 , wherein the demographic distributions are associated with at least one of (a) day of the week and (b) day part of the day of the week.
10. The method of claim 8 , wherein each page view is associated with a Uniform Resource Locator (URL) pattern.
11. The method of claim 8 , further including determining additional activity not represented by the virtual panel, and wherein calculating the online audience is also based on the additional activity.
12. The method of claim 8 , wherein weighting the second panelists associated with the unmeasured locations is based on a set of known collected characteristics of unmeasured persons at the unmeasured locations.
13. The method of claim 8 , wherein the first audience estimate is determined by evaluating a negative binomial distribution.
14. The method of claim 13 , further including generating census sessions based on the page views and the measured panel data, and wherein the evaluating the negative binomial distribution is based on negative binomial distribution parameters determined from the census sessions.
15. A computer readable medium comprising instructions that, when executed, cause a processor to:
access measured panel data from one or more metering devices via one or more wireless communications, the panel data including accesses to web pages from first panelists;
access census-based page views via one or more wireless communications, the census-based page views including accesses to web pages from unmeasured locations;
calculate page view demographic distributions based on the measured panel data associated with the first panelists;
distribute the census-based page views according to corresponding demographic distributions;
estimate a first audience estimate associated with the unmeasured locations based on the distributed census-based page views and the measured panel data;
generate a virtual panel based on the first audience estimate to enable more accurate estimation of traffic from the unmeasured locations by weighting second panelists associated with the unmeasured locations and reweighting a subset of the second panelists;
estimate a second audience estimate associated with the unmeasured locations based on the virtual panel, the second audience estimate including demographic characteristics; and
calculate an online audience based on a measured location audience and the second audience estimate.
16. The computer readable medium of claim 15 , wherein the demographic distributions are associated with at least one of (a) day of the week and (b) day part of the day of the week.
17. The computer readable medium of claim 15 , wherein each page view is associated with a Uniform Resource Locator (URL) pattern.
18. The computer readable medium of claim 15 , further including instructions that, when executed, cause the processor to determine additional activity not represented by the virtual panel, and wherein calculating an online audience is also based on the additional activity.
19. The computer readable medium of claim 15 , wherein weighting the second panelists associated with the unmeasured locations is based on a set of known collected characteristics of unmeasured persons at the unmeasured locations.
20. The computer readable medium of claim 15 , wherein the first audience estimate is determined by evaluating a negative binomial distribution.