“The tree of nonsense is watered with error, and from its branches swing the pumpkins of disaster.” – Nick Harkaway
Our motto, collect better quality data more efficiently, is not an idle affectation.
We really do put a premium on data quality. We understand that important decisions are based on the data we help collect, and that those decisions are as good as the quality of it.
Here are 5 of the means by which we consistently deliver better quality data.
1. Automatic validation
Who is cleaning data is 2017?
Of course there is no denying that data cleaning is a critical step in data processing. Data collected in the field by human beings will always have errors to be cleaned.
But it is also true that the vast majority of common errors can be arrested even before they happen.
Collecting data on women of reproductive age? Easy. Limit the age variable to between 15 and 49 years.
Want to ensure that all members of a household are enumerated? Set the length of the household roster to validate against the number of total household members.
Hoji lets you set validations rules like these when creating your digital survey. During data entry, these rules are enforced strictly (or loosely, if so configured) to ensure only valid data makes it to your database.
The truth is that even the best enumerator slackens a little sometimes. You want to rest assured that on those occasions, the validity of your data is not compromised.
2. Automatic skip logic
Just like validation, another way dirty data gets into your database is through skip logic violation. An inattentive enumerator records that a non-smoker smokes 5 cigarettes a day. Or that a respondents who reports favoring Brand X has 2 reasons why they favor a competing Brand Y.
None of this data make any sense, so you have to rely on your data analysis team to detect and deal with the inconsistencies. Yet, this kind of illogical information is even harder to catch than validation errors, increasing the risk that it ends up in your database and, worst of all, in your final report.
Hoji lets you define skip logic at the point of questionnaire digitization, guaranteeing hat enumerators are automatically guided through the correct order questions, and no illogical data enters your database.
3. Zero data entry
This one is easy. The fact that data comes directly from the field and into your database means that there are zero transcription errors.
4. Time and location metadata
Hoji automatically tags every record with time metadata, letting you know exactly when the interview started, when it ended and when it was uploaded. If the interview was updated, Hoji captures the time that happened, too. We can even tell you the exact time when individual questions were entered.
By default, Hoji also tags records with location information, although this can be turned off. That means you can tell exactly where the enumerator was when they created the record, where they were when they completed it, and even the distance in between!
No more relying on expensive supervisors to ensure that your enumerators go where you send them when you send them.
5. Detecting interviewer falsification
Think about what you can do with all the metadata described in #4 above. Someone filling out their questionnaire unrealistically fast? You can tell. Someone filling out their questionnaire at their hotel room? Busted!
There you have it! 5 ways we consistently deliver better quality data. Our motto, I told you, is not an idle affectation.
Have a great week!