Displayr contains all the tools required to weight your sample to represent the population and to use this weight appropriately in your analysis and reporting.
Displayr supports cell weighting. For example, if you know that 1.2% of your population are males aged 18 to 34 living in Texas, you can weight your data so that all your analyses reflect this.
You can also construct weights even when you do not know the interlocked targets (rim weighting). This means if you don’t know how many males aged 18 to 34 live in Texas, but you do know how many males are in the population, how many people are aged 18 to 34, and how many are Texans, you can still create a weight.
Traditional weighting software only deals with categorical data, whereas Displayr allows you to weight categorical and numeric data (e.g., market share, average purchase consumption). This is done through the modern technique of calibration.
You can specify maximum and minimum values for weights (capping) to avoid generating extreme weights.
Sometimes studies have existing weights designed to rectify known non-representativeness (e.g., caused by stratification or differential response rates). These design weights can be incorporated when creating new weights to address the overall representativeness. You can also easily create expansion weights to make the weighted sample add up to the population.
And with multiple weights, you can specify the weight for either all analyses, or subsets of analyses (e.g., create one weight for your occasion data, and another for your household data).
The statistical tests on tables and generalized linear models (e.g., regression, driver analysis, logistic regression) address the weights via Taylor Series Linearization (i.e., they do not confuse the weighted sample size with the actual sample size).
Weights will automatically update when you revise the data, whether filtering, data cleaning, or adding in a new wave of data.
Displayr is a general-purpose app that does everything from crosstabs to text coding to advanced analysis to dashboards, driver analysis, and segmentation.
Once you have weights, you can easily use them in all your subsequent analyses and reporting. There’s no need to use one package for creating weights and another package for analyzing them.
Research Analyst, dunnhumby
SQL, databases, Excel, CSV, text, SPSS, survey platforms, APIs, integrations, & more.
Summary tables, crosstabs, pivot tables, regression, text analysis, segmentation, machine learning, & more.
Data visualization, interactive data apps, dashboards, presentations, PowerPoint, Excel, PDF, web pages, & more.
AI automatically identifies and categorizes themes within your text data, providing deeper insights.
Understand and analyze complex emotions like frustration and sadness, helping you understand customer motives.
Extract key entities like names, places, and organizations to enrich your analysis.
Fine-tune and adjust categories to match your specific needs and preferences.
Create stunning word clouds, charts, and dashboards that help tell the story behind your text.
Analyze text data in any language, with true native language support to a global audience.
Analyze large volumes of text to gauge positive, negative, or neutral sentiments.
Extract insights with unrivalled accuracy, utilizing NLP to reduce manual effort and free up time.
Displayr helps Vennli complete projects 5x faster
Data weighting (also known as sample balancing, post-survey adjustment, raking, or poststratification) is a technique that adjusts the results of a survey to bring them into line with the target population. A common use case of weighting would be to adjust a sample that does not accurately represent the gender breakdown of a specific population, i.e., the survey results are made up of 35% females, but the actual population contains 51% females.
Weighting survey data removes discrepancies in the results. This allows researchers to conduct surveys on samples that do not perfectly represent a desired demographic and still deliver accurate findings.
Weighted data refers to survey results that have been adjusted using weighting techniques. There are certain cases – such as hierarchical cluster analysis and distance calculations – where weights should be ignored, as the calculations are based on the differences between individual cases.
Rim weighting, also known as raking, is when categorical adjustment is applied to two or more variables. ‘Rim’ refers to the numbers being on the edge of the table. It iteratively adjusts weights to align the sample proportions with population benchmarks for each variable.
There are a number of different best practices to keep in mind when weighting survey data. Some of the key things to keep an eye out for include:
Yes, you can certainly weight your survey data in Excel. To weight data in Excel:
However, weighting data in Excel can sometimes be a complicated process. Displayr simplifies survey data weight, reducing errors and saving you time.
© 2024 Displayr Pty Ltd. All rights reserved.
