Measure the Construct, not the Proxy

 

An idea is not a variable. How you choose to measure it can change the story the data tell.

​

Concepts like connection, engagement, religious practice, or community experience can rarely be captured with a single survey question. Each construct can be represented by several plausible measures—and those measures do not always produce the same pattern. Use the tool to choose a construct, then switch among different ways of measuring it. Watch how the age-group rankings shift as the operational definition changes.

 

What this demonstrates
Construct validity • Measurement choices • Operationalization • Comparing indicators

​

Data: 2025 Greater Boston Jewish Community Study
Created by Daniel Parmer, PhD

 

About this experiment:
The app uses several survey measures that could reasonably represent broader concepts such as connection, organizational engagement, religious practice, and community experience.

​

For example, “connection” might be measured through a sense of belonging, the importance of Jewish community, social networks, religious practice, or ties to Jewish organizations. “Religious practice” might be represented by Shabbat observance, religious-service attendance, or celebrating Jewish holidays.

 

The bars show the percentage of respondents in each age group who meet the selected measure. As you change measures, the bars move and re-rank to show how the apparent pattern can change even though the underlying respondents have not. Choose a construct, then choose a measure to see how operational decisions shape the conclusions we draw from the same dataset.

 

Want to see another experiment?