WEAT: Web Enabled Analysis Tool
The BRFSS, the world’s largest telephone survey, tracks health risks in the United States. Information from the
survey is used to improve the health of the American people. |
Data Analysis FAQ
What is cross tabulation?
A cross tabulation, or "crosstab," produces frequencies or percentages for one or more variables, in one or more tables. For example, one can use the cross tabulation procedure in the BRFSS to generate a table showing numbers and percentages of respondents with diabetes by age group. A general formula for cross tabulation can be depicted as A x B, where A is the dependent variable or outcome (e.g., diabetes) and B is the independent or exposure variable (e.g., age).
What is a table?
Cross tabulation produces one or more tables, with each cell of the table containing frequencies and percentages.
What are frequencies?
Frequencies are counts, or percentages derived from counts. In the BRFSS, frequencies represent numbers of persons reporting certain health behaviors or demographic characteristics. You can produce frequencies on a single variable by selecting it as your row variable. The output is also called a one-way frequency table.
What is logistic regression?
Logistic regression calculates the contribution of one or more predictors on a particular outcome, such as "Risk factor:At risk for binge drinking." The results provide a predictive model and can be converted to log odds. The basic logistic formula using one predictor is depicted in the form Y= exp(a + B1X1)/1 + exp(a + B1X1).
What is a dependent variable?
A dependent variable is affected by one or more independent variables. For example, general health is known to be affected by age. Therefore, health may be a dependent variable in some analyses.
What is an independent variable?
An independent variable is one that affects a dependent variable. For example, age is known to affect general health. Therefore, age may be an independent variable in some analyses.
What is a control variable?
Control variables help refine cross tabulations by splitting data into separate tables. For example, health is affected by age, but the effects of aging may differ by the amount of physical activity the respondent has. In the BRFSS dataset, adding a control for exercise would produce separate tables containing Health x Age frequencies for those who exercised in the past 30 days, and those who did not. The general form of such analyses can be depicted as A x B x C, where A is the dependent variable (health), B is the independent variable (age), and C is the control variable (exercise).
What BRFSS data are available for analysis?
At this time, the WEAT database contains only national BRFSS data from 2005. National data from other years and even state-level data may be added in the future, depending on the usage of the tool.
What does sample size tell me?
Sample size reports the number of individuals with the conditions specified in each cell of a table.
What does the weighted sample size represent?
The weighted sample size uses data from the sample of respondents and estimates the number of adults in the US population with the conditions specified in a table.
What does cell percentage tell me?
The cell percentage reports the proportion of individuals matching conditions depicted in each cell of a table.
What does the row percentage tell me?
The row percentage that appears in each cell of a table reports the proportion of individuals with that condition. When row percentages are added across the columns, they will total 100%. This gives you the distribution of individuals with that condition across the "outcome" variable.
What does the column percentage tell me?
The column percentage that appears in each cell of a table reports the proportion of individuals with that row's value. When rows are added down the columns, they will total 100%.
What does "Total Number Missing" value mean?
The total number missing represents the number of "cases" or respondents missing answers to one or more variables. For example, if 10 individuals did not report educational attainment, and another 20 did not report diabetes status, and yet another 3 reported neither diabetes status nor educational attainment, then a cross tabulation of Education x Diabetes would have 27 missing cases.
What is a standard error?
A standard error is the estimated standard deviation of a population estimate. In the WEAT, standard errors are calculated appropriately for the complex survey design of the BRFSS.
Back to top
|
