BRS - Using Data

BRS - What Do I Do With The Data?

It is important to plan the data collection and analysis PRIOR to beginning any study. Collecting "post-hoc" data and analyzing it is prone to bias and error.

The steps here are fairly straightforward:

1. List the exposures of interest (the independent variables) and categorize each by the type of data collected.

2. List the outcomes of interest (the dependent variables) and categorize each by the type of data collected.

3. Decide which variables you will report as descriptive data (frequencies, etc.)

4. Decide which comparisons and inferential statistics you will use.

   1. When comparing averages of continuous data, use PARAMETRIC STATISTICS (e.g. a T-test).

   2. When comparing proportions of categorical data (nominal, ordinal, or dichotomous), use NON-PARAMETRIC STATISTICS (e.g. a chi-square)

   3. When analyzing the effects of confounding variables, interacting variables, or multiple exposures, use MULTIVARIABLE STATISTICS (e.g. regression)

Using a data table is the best way to organize your data collection and analysis.

For a case-control study of the association of cell phone use and brain tumors, you might construct the following table:

Data Element

Type of Data

cell phone type

Nominal

duration of cell phone use

Continuous

brain tumor diagnosis (+/-)

Dichotomous

brain tumor stage

Ordinal

From this data, you could report the frequencies of cell phone types, average duration of cell phone use in the groups, and the stages of tumor in the case group.

For the inferential statistics, you can use non-parametric statistics to correlate duration of cellphone use and diagnosis of brain tumor.

There is MUCH more to this that is presented here. For additional detail, you're welcome to check out other study design/data planning sources. But for our course, just know the detail above.