Use data to ground your thinking
So you've collected outcomes data. Fantastic! Now it's time for context. Analyze that data to extract insights that you can use to improve outcomes. Then share those insights with others with clear and compelling reports.
Sometimes the simple act of collecting and comparing numbers is all you need. At other times you might want to evaluate an existing program or predict the outcomes of a possible new one, which calls for more complex analysis.
Sophisticated software for data analysis and visualization is readily available, but you don't have to start there. See if the software you're already using can help you do the following.
Describe data with statistics
Though descriptive statistics are relatively simple to collect, they can be useful. These include:
- Totals. You might add up the amount of money you received from donors annually over the last five years and look for increases or decreases over that period.
- Mean. You might total the amount of money you received from donors last year. Then divide that total by the number of donors. The result is the mean, or average donor contribution.
- Median. The median is computed by arranging measurements in order from highest to lowest and finding the one in the middle. The median is especially meaningful when your data is skewed — say, when one or two donors contribute many times more than the rest.
- Mode. This is the most common value in your data — for example, the county where most of your clients live.
- Standard deviation. Low standard deviation means that most of your data points cluster closely around the mean — for instance, that most graduates of a workforce development program are now earning about the same average hourly wage. In contrast, high standard deviation means that your data points are widely spread out — for example, when the highest earners make many times more per hour than the lowest earners.
Use data to make inferences
In addition to describing your existing data, you might want to draw wider conclusions from it. Some applications apply statistical formulas to help you make such inferences.
Correlation indicates a relationship between two sets of data. For example, you might that find people who spent more hours in your workforce development program now earn, on average, more than participants who spent fewer hours in the program.
Regression involves plotting numbers on a graph and finding a line that indicates strong trends in the data. Based on regression analysis, you might predict how much graduates will earn based on the number of hours spent in your workforce development program.
Remember that making inferences isn't the same as linking causes and effects. Correlation and regression can overlook relevant factors that you were unable to measure. For example, graduates who earn more might also have worked in their field before entering your program. This added experience might help them command higher wages.
Report data with visuals
Your organization's management team and database manager might be able to understand the implications of raw data. However, other stakeholders — board members, donors, volunteers — may want to see data reported in visual form. Some options include:
- Bar charts
- Pie charts
- Line graphs
- Geographic maps
- Heat maps
- Scatter plots
Any of these visuals can be included as gauges on a dashboard. A dashboard displays performance indicators in real time and is updated each time new data is collected. Many customer relationship management applications generate dashboards based on performance indicators that you choose.
Color coding the dashboard helps translate data into insights and action. For example, green shows good progress in meeting performance targets. Yellow indicate trends to watch, and red denotes items that need immediate attention.
You can create a dashboard for each aspect of your organization's activity. A finance dashboard could show days of cash on hand and net surplus or deficit compared to budget. A program dashboard could show the number of first-time participants during each quarter of the year in comparison to your annual target.
Be clear about what data can't do for you
Not everything that matters can be measured. For example, the quality of relationships between your staff members, management team and board of directors or trustees is crucial to your organization's success. Yet this factor is hard to quantify. Data can ground your thinking, but the next decision is still up to you.