Ensuring the Accuracy of Data Estimates in MFour Studio

This article will explain how behavioral data is adjusted to provide accurate representation with the US population.

Studio offers a user-friendly platform for exploring, categorizing, and visualizing over 550 million daily app, web, and location journeys throughout the US. This provides unparalleled insights into consumer behavior and trends.

To ensure the data gathered from our consumers is representative of the US population, weighting IS applied to the consumer database. Weighting is often used in survey research to correct for imbalances and to account for any biases that may be present in the data. By applying weighting, we can ensure that the population is representative and that the data estimates provided are accurate and reliable. 

Weighting Explained

The weighting formula utilized in the Journeys Marketplace takes into consideration the following variables:

• Age
• Gender
• Ethnicity
• Education
• Operating System (iOS or Android)
• Region/State

Traditional weighting involves each survey respondent being assigned an adjustment weight, with underrepresented groups receiving a weight greater than 1 and overrepresented groups receiving a weight less than 1. The weighted values are then used in the calculation of means, totals, and percentages. The weighting applied in Studio uses a similar approach but is dynamic as it factors in other components specific to MFour Studio’s behavioral data parameters. 

In Studio, a daily weighting level is used to ensure that the survey responses are evenly distributed across the days of data collection, accounting for any daily variations in representation. 

In addition to the daily weighting, two types of weights are applied to provide a final weight for each respondent: 

• Design weight
• Demographic weight

Design weight is the average number of units in the population that each sampled unit represents, which is the inverse of its inclusion probability (π) in the sample. For Studio, as the sample is taken on a daily basis, the design weight is determined accordingly. 

Here is the formula for calculating Design Weight: 

Design Weight Formula= 1/ (Daily Active User by OS) * US Population by OS

Example:  Calculating Design Weight for B&M Android Users on 2023/03/01

Daily Active B&M Android Users =  47,000

Android Ownership among US Population: 44%

Total US Population: 257,518,311 (derived)

US Population for Android: 257518311 * 0.44 = 113308056 

Design Weight: 1/47,000 * 11330805 =  2410.8

Demographic Weight is used to rebalance the proportion of demographic components, such as age, gender, and race/ethnicity within each state to match the US population. This has to be calculated separately for each US state. 

Here is the formula for calculating Demographic Weight: 

Demographic Weight = Population distribution by state & OS % / Sample distribution by state and OS %

Example: Calculated Age Range Distribution for Android Users from California on 2023/03/01

Panelist ID	6836453
Age Range	25-34 years
Gender	Female
Ethnicity	Caucasian
Region	West
US Census Division	Pacific
Education	Some college, A.A. degree or technical college
State	CA
Zipcode	93422
DMA	Santabarbara-sanmar-sanluob

Demographic Weight =  

Age Weight * Ethnicity Weight * Gender Weight * Education Weight * DMA Weight

1.18*0.553*1.427*0.841*0.874 = 0.684

Upon calculating the design and demographic weights, both are combined to produce the Final Weight which is applied to each respondent’s daily visit to a specific venue, app or website. 

Final Weight Formula

Weight = Design Weight  *  Demographic Weight 

These weighted results for consumers are displayed the the Panelist Weights table, which is further categorized by venue, app, or website.

By applying weighting, MFour Studio is able to provide a more accurate representation of the US population, delivering valuable insights and data that businesses can use to make informed decisions with confidence.