Data Analysis
Allow us to help maximize the use of all the data that you are already collecting for day-to-day needs. Whether your needs are basic, advanced or special, we provide the appropriate solutions.
Data
Analysis
How should these data be analyzed?
Basic Statistical Analyses
Descriptive Analyses
These include sums, averages, counts, medians, minimums, maximums, etcetera, and answer such questions as:
-
What is the average age of my clients?
-
What is the average age of clients by gender?
-
How many clients do I have over the age of 40?
-
How many male vs. female clients are injured (or purchase my services)?
Comparative Statistical Analyses
These analyses determine whether differences between groups are more likely to be real, or just due to chance. They include analyses such as:
-
t-tests for comparison of group means
-
Chi-square tests for comparison of group proportions
-
Correlations between groups
Advanced Statistical Analyses
Multiple regression: Adjusting for groups
These analyses determine if differences between groups are likely to be real, but also adjust for group differences. For example, are males at higher risk compared to females after accounting for differences in age? The analyses include:
-
Linear regression
-
Logistic regression
-
Poisson regression
-
Multi-level regression
Prediction models
These analyses have a different focus than the above analyses. They are most often recognized as developing an algorithm that best predicts a particular result.
-
What are the characteristics of my clients that predict who will purchase my services?
-
Which factor best predicts the outcome? (risk factor analysis)
-
What is the best combination of factors likely to predict the outcome? (predictive modeling)
Other Analyses
Causal models: What are the actual causes of the outcome?
Prediction models predict, but do not examine for causes. Red hair might predict Scottish or Irish ancestry, but dying one’s hair red does not cause one to be Scottish or Irish.
If you want a policy change to be effective, you must affect a cause of the outcome.
Establishing causal relationships through prior knowledge, and testing the relationships using your data, is known as causal inference.
Psychometric analyses for tests and examinations
We provide support for test examination development including forms (paper and scanning, or electronic forms), analyses (scores, variability, pass/fail, examination validity) and processes.