How to weight data in our tool?
Data Weighting made simple with Sample Weighting.
This video gives you a visual presentation on how to use our tool for weighting your data to fix a mismatch between your sample and the reference population.
Sample Weighting makes it easy in just 5 simple, steps:
- View the representation of the sample
- Calculate the weight factors
- Apply data weights to sample proportions
- Match your population to your sample
- Finishing your research with unbiased results
On this page we’ll show you the necessary steps to fix any imperfections in your sample.
What is Data Weighting and why is it important?
Data weighting is a statistical technique that is used by market and survey researchers to correct survey data. Using this technique researchers can adjust their sample to improve estimates based on the data collected. There are two good reasons why researchers do this: The first is to correct for unequal probabilities that occur during sampling. The second is to bring survey data into balance. This is very useful when certain respondent groups aren’t represented in a sample as they are in the population surveyed.
Representation of the sample
When you have a sample you’ll want to compare it with the population, to see if all groups surveyed are proportionately represented according to their presence in the population. As you can see from our example: The blue group is underrepresented and the pink group is overrepresented. In order to get a proportionally balanced sample you should weight the data.
Calculate the weight factors
If you want a sample that has the desired distribution according to the proportions in the population, first you need to calculate how much weight each group needs to be properly represented in the sample. For this you can use an easy formula: % population / % sample = weight.
Apply data weights to sample proportions
Next, you multiply the sample data with the weights you’ve calculated in step two. Applying the weights will ensure
that your sample proportions match those of the natural population you’ve surveyed.
It's a match!
Now, when the population is compared to the weighted sample the proportions of respondents match.
Made it easy
So far, so good. You might be wondering why you’d need a tool for this.
Now consider a situation where you have more variables besides gender that you want to control for. like age, length and place of birth etc. The problem suddenly becomes complex. Imagine a situation where you have to calculate thousands of samples by hand.
To solve this conundrum. Sample weighting created a web based weighting tool enabling any researcher to easily calculate data weights for their sample, no matter how complex. There is no need to write any syntax or even have programming skills.
All you have do to is provide the data and set the targets.