Weighting.

Weighting allows you to address any sample bias in your Glow surveys due to response volumes not matching desired proportions (e.g. nationally representative sample). This helps researchers eliminate bias that occurs when the data derived from the survey does not represent the target population accurately and could impact on the decisions made.

You can weight your data in analysis using any filters you've created in your dashboard.

E.g. if you captured 45% males and 55% females in field, and you want to weight this to be 50% males, 50% females, you can set Weight for Males at 1.11 (50/45) and Females at 0.91 (50/55) in your dashboard's weighting scheme.

Setting Up Weighting

To weight your data

Click the Weight icon (top left under the dashboard dropdown)

Tick "Enable Weighting"
Click "Add Rule"
Select Filter, e.g. Male
Set Weight, e.g. 1.11

Repeat steps 3 to 5 for each Filter you want to weight by.

Example: Gender Weighting Scheme

Anything not listed as a grouping, i.e. nonbinary would be weighted as 1.

If you wanted to remove a group, you could set it's weight as 0.

When you're happy with your Weighting Scheme, click Apply.

The Weight icon will highlight in orange indicating weighting has been applied to this dashboard.

Note, you'll need to have already created filters for the variables you want to weight by.

To create filters e.g. for "Male"

Go to the Filters tab
Click Create Filter
Name your Filter
Click the Condition dropdown
1. Scroll down to your question, e.g. Q2. Do you identify as... OR
2. Use the Search bar to find your question, e.g. search "Q2" or "identify"
Choose your logic condition, e.g. Any of > Male
Click Done

Creating Interlocked Weights

Interlocked Weights

In Weighting V1, to set Interlocked Weights, e.g. Gender x Generation, just create Interlocked Filters!

Example: Gender (2) x Generation (4)

Create interlocked filters by combining the Gender question with the Generation tags using "AND" in your filter logic

Once you've created 8 interlocked Gender x Generation filters, then use these filters in the same way to set weight.

Q: But how do we know what weight to apply to each?

1: Our data shows a Gender x Generation distribution of...

2: But ABS data shows a Gender x Generation distribution of...

3: Therefore, each Filter needs a Weight of: ABS % / Actual Data % = Weight Value

Example: for Male Gen Z (18-24): ABS 5.9% / Actual Data 6.5% = Weight of 0.898 (3dp)

4: Which when applied, aligns OUR distribution to the ABS distribution 🎉

You can do the same thing with as many interlocked filters as you want, e.g. gender x age x region

Weighting V1 does not yet allow for non-interlocked weights, e.g. gender x age weighted independently from state. Therefore, when using Weighting V1, only use mutually exclusive groupings (if you don't, it'll use the 1st weight qualified for).

Where Weighting Applied

Every Question in your Dashboard with a Table & Chart will have weighting applied.

However, the following question types will not have weighting applied.

Text questions with Word Clouds and Full Answer Lists
Location question with Heatmap/Clusters
Image Upload question with Collage

Every Value in your View will have weighting applied:

Example: (using 1dp)

1a: Unweighted Counts & Bases

1b: Weighted Counts & Bases

Weighted Row Total (n) - far right column of table
Weighted Column Total (n) - bottom row of table
Weighted Total Total (n) - bottom right of table

The Column %, Row %, Total %, and Sum, and Average values will of course also be weighted.

The Chart will reflect the weighted data in the same way the Table does

Different Weights for Different Dashboards

Each of your dashboards could have a different weighting scheme applied to them

For example, you could have...

an unweighted dashboard
a gender weighted dashboard
an age weighted dashboard
a gender x age weighted dashboard

Weighting in Shared Dashboards

Weighting enabled in your dashboard is also applied in any dashboard shared with the share icon.

Your shared dashboard users will see the weight icon highlighted in orange (top left) to show that weighting has been applied to their shared data and can click that icon to see what Weighting Scheme has been applied.

Your client can edit & apply a different weight scheme but just can't save the custom weights in their shared dashboard. When they refresh their shared dashboard, it will revert to the original dashboard with the original weighting scheme you shared with them (i.e. whatever's saved in your subscriber dashboard)

Statistical Tests (Sig-Test & Margin of Error) and Weighting

In Weighting V1, weighting is incompatible with significance testing & margin of error and therefore when 1 is used, the other is disabled

I.e. you can ONLY run significance testing & margin of error on an unweighted dashboard.

Weighting in Response Exports

If you export the response to csv, xlsx, or spss-xlsx, this is how it will appear in your exports.

CSV - new "Weighting Scheme" added to Column J after "Respondent Fingerprint" in Column I but before the 1st question in Column K.

XLSX

new "Weighting Scheme" added to Column J after "Respondent Fingerprint" in Column I but before the 1st question in Column K in the "Numeric Data" tab.
no change in the "Data Map" tab

SPSS-XLSX

new "Weighting_Scheme" added to Column J after "Respondent Fingerprint" in Column I but before the 1st question in Column K in the "Numeric Data" tab (same as above).
new Variable: "Weighting_Scheme" in the "Variable Labels" tab.
no change in "Value Labels" tab.
no change in "Import Syntax" tab.

If you have multiple dashboards, the weighting in your response exports will be for the CURRENT dashboard selected.