Operationalize your objectives with the variables in your dataset
Paste your objectives and HILO matches them against the variables in your dataset. Each objective comes back with the columns that measure it, the ones that come close and what is lost by using them that way, what is still missing to ask, and a concrete formula when there is data to calculate it with.
- Matches against the whole dataset
- Tells you what's missing, not just what's there
- Every indicator is written with columns that exist
From project objectives to an indicator with a formula
Four steps, and the file goes in just as it came out of the form.
The dataset, with its variable names as they are
Excel, CSV, or the SPSS .dat file. HILO builds an inventory of columns with their type, how many distinct values each one has, and a few examples, because the header alone isn't enough: “P12_3” tells you nothing, and its values “Never / Sometimes / Always” tell you everything.
The objectives exactly as written
One per line, copied from the project document. You don't have to rephrase them or turn them into indicators first: that is exactly what is still to be done, and it is the work HILO does.
Direct, approximate, and what is lost
Each objective comes back with two lists. The variables that measure it as it is stated, and those that measure something related without being the same thing. The distinction matters: an approximate one is useful, but you need to know it is approximate before you put it in the report.
What's missing is stated, not hidden
If no data in the dataset measures an objective, that is declared. And when there is data to calculate with, an indicator is proposed with its formula written over the real columns, not over invented variables.
It doesn't invent a column to cover an objective with no data
Columns are named by their letter and checked against the inventory: if an answer mentions one that doesn't exist, it is discarded and you are told how many were discarded. An objective with no variables is reported for what it is: an objective this dataset can't measure.
The dataset tells you the starting point, not where you need to get to: that is a project decision, not a data point.
Yes. Variables are looked at in batches and then a closing pass is done with everything flagged in view, so a dataset of four hundred columns goes in just like one of thirty. The second pass is the one that decides what is missing, because someone who has seen only part of the dataset can't claim that something isn't there.
That's the usual case. That's why the inventory doesn't go with the header alone: it goes with the variable type, how many distinct values it has, and some examples of those values. A column called “P12_3” whose values are “Never / Sometimes / Always” is understood by its values, not by its name.
Yes, and it is designed to say so. An objective with no associated variable shows up as such, with a short sentence explaining it. If the objective matters, you need to ask about it, and finding that out before you write the report is better than finding out afterwards.
The formula comes written with the letters of the columns in your dataset and stays simple: a proportion, an average, a count. The review is yours: what is proposed is a verifiable starting point, not a result.
No. The file is read in your browser, and what travels for the matching is the column inventory: names, types, and a few example values for categorical variables. No examples are sent from free-text columns, because they are people's answers and aren't needed to know what the columns are about.
A document with each objective, the variables that measure it, the ones that come close with what is lost in each case, what is still to ask, and the proposed indicator with its formula. You download it and paste it into the project document.
Before and after operationalizing
Try it with the dataset you already have
Upload the file, paste three objectives, and see what comes back. If an objective has nothing to measure it with, you'll know in ten minutes.