Simmons Insights

Sample (Unweighted) and Weighted numbers

Here's another example of how to read the Sample and Weighted numbers, for a question examining the number of children in a household:

• The total survey respondents were 11,864. 
  • Of those, 1,874 respondents had 1 child in the household,
  • 1,474 respondents had 2 children in the household,
  • 7,503 respondents had no children in the household,
  • Etc. 
• If we project those representative numbers to the entire population, we would expect 241,691,000 adults over age 18 in the entire contiguous U.S. (Remember, add 000 to the number in the box)
  • Of those, we would expect 35,718,000 people to have 1 child in the household,
  • 30,183,000 people to have 2 children in the household,
  • 157,245,000 people to have no children in the household,
  • Etc. 

Vertical and Horizontal percentages

Here's a sample crosstab, studying number of children in the household compared to how many of them buy a large number of batteries (20-29) per year:

For vertical %: start with the vertical (column) heading:

"Of the people who ___[vertical column]___, ____% of them are ___[horizontal row]___." 

• Of the people who bought 20-29 batteries in the last year,
  • 13.9% of them had one child in the household. 
  • 14.7% of them had two children in the household
  • 6.2% of them had three children in the household

The four-children row has asterisks next to it, so we won't look too closely at that row. See the Index video for what the asterisks mean.

For horizontal %: start with the horizontal (row) heading:

"Of the people who ___[horizontal row]___, ____% of them are ___[vertical column]___." 

• Of the total survey respondents, 8.5% of them bought 20-29 batteries in the past month.
• Of the people who have one child in the household, 8% of them bought 20-29 batteries in the past month.
• Of the people who have two children in the household, 10% of them bought 20-29 batteries in the past month. 
• Of the people who have three children in the household, 9.5% of them bought 20-29 batteries in the past month. 

The four-children row has asterisks next to it, so we won't look too closely at that row. See the Index video for what the asterisks mean.

Index numbers

-- beware of the asterisks.

Here's another example of how to read the Index numbers, for a question examining the likelihood that a larger number of children in a household = a larger number (20-29) of battery purchases:

The average for the Total row is set at 100. That's the average for the entire U.S. population, with or without any number of children.  

• Respondents with 1 child have an Index of 94.
  • That's 6 below 100, so we could say they are 6% less likely to buy 20-29 batteries per year. But that isn't very far off of the national average (100), so I personally wouldn't make strong marketing recommendations based off of something that could be within the margin of error for that survey. (Simmons doesn't really tell us what the margins of errors are for its survey, so that's why I use caution with anything that's too close to the average.)
• Respondents with 2 children in the household have an index of 118 (or 18% more likely, hence the green up arrow), and respondents with 3 children in the household have an index of 112 (or 12% more likely, hence the green up arrow).
  • That's a bit farther off of the average (100), so it is probably a more useful insight!
• Respondents with 4 children in the household have an index of 81 (or 19% less likely, hence the red down arrow), and respondents with 5+ children in the household have an index of 89 (or 11% less likely, hence the red down arrow).
  • BUT! Those boxes also have asterisks next to the numbers. That tells me the sample size for this respondent group was below 30. That makes the numbers much less reliable and more easily skewed by just a few unusual responses. 

REMEMBER: 

* Sample size is between 31-60, use with caution

** Sample size is below 30, use with extreme caution