by
William J. Bailey, M.P.H., C.P.P.
Associate Professor of Applied Health Science, and
Executive Director, Indiana Prevention Resource Center
Indiana University
When large corporations consider adding new retail locations, they access sophisticated data sets to provide detailed information about everything from potential markets to workforce availability. They need to know about their potential customer base -- the range of income levels, lifestyles, family size, leisure habits, and countless other arcane bits of trivia are used to create a profile of each community the corporation is considering.
##
##
Every time you make a purchase at a grocery store, discount store, or drug store, barcode data on your purchase is entered into a database that businesses will use to draw conclusions about your community. Corporate market analysts study the magazines you subscribe to, the television shows you watch, the movies you attend, and the types of cars you drive, and draw conclusions about the types of people who live in your neighborhood and town. Do you buy generic detergent and clip coupons? How many disposable diapers do your children consume, and how does that compare with children in another nearby town? Do you buy soft drinks for consumption at home, or do your purchase them at restaurants? Do you buy 12 ounce cans or 2 liter bottles? Corporate America studies the habits of its customers to predict what they will do and what they will buy in the future.
##
Tobacco and alcohol companies utilize extensive market research prior to the release of a new brand or product -- or even a new advertising campaign. They carefully study data on which "market segments" of the population are most likely to be interested in their products, and then develop strategies to target these market segments. One cigarette brand may target young women from blue-collar families, while another may target the more upscale young debutante. Cigarette manufacturers may use one brand to attract "new smokers," and then switch them to another of their brands after the smokers are hooked. A wine company may use one set of ads in women's magazines, and a completely different strategy when marketing to men.
Until recently, drug abuse prevention professionals were at a disadvantage. Their "opponents" -- the big tobacco and alcohol marketers -- had access to extensive and expensive marketing information that could be used to target specific segments of the population. Cigarette ads could create an aura of sophistication and allure that preventionists found hard to match. Fifth grade boys knew the names of more brands of beer than they did of U.S. presidents. Middle school girls knew the advertising slogans for "women's cigarettes," and collected proofs of purchase to redeem for promotional items adorned with cigarette logos. Massive databases of market information were used to identify prospective smokers and drinkers, and advertising and marketing campaigns were designed to attract them to new vices. Convenience store chains used demographic maps and household purchase data to locate new outlets at optimal locations -- often resulting in neighborhoods where alcohol and tobacco outlet density created a constant stream of temptation for the youth who lived there. Billboard companies located signs in the same neighborhoods, again based upon market research, and youth were lured by outlets and ads at every turn.
Most prevention programs were content to use "instincts" and good intentions to plan prevention campaigns to counter these environmental traps -- and most often lost the battle. The best prevention could do was to complain about the amount of money that big business had to spend to target our youth. Corporate America was using sophisticated marketing data that cost hundreds of thousands of dollars. The odds were stacked against the "good guys."
Demographic Analysis Workstation in the IPRC's Office of Prevention Technology
Prevention programs can utilize demographic and psychographic data to segment their target population into "market segments," and then develop strategies to target (reach and influence the behavior of) those segments.
The Indiana Prevention Resource Center has developed a Demographic Analysis Workstation used to conduct demographic and market research to support prevention planning efforts in Indiana. IPRC uses the PCensus desktop demographics software package created by Tetrad Computer Applications, and has obtained numerous data files to cover the entire state of Indiana. IPRC has all of the Indiana data files from the 1990 census, as well as psychodemographic profiles, consumer and household purchase data, 1998 and 2003 population estimates, and daytime population and employment data files for every census tract and census block in the State of Indiana.
Using the PCensus software, users can search for data for specific geographic areas that are predefined or are defined using map coordinates. Various counties, cities or towns, census tracts or zip codes can be compared with each other. The data can be imported into a Geographic Information System, such as MapInfo or can be exported into spreadsheets or databases for further calculations. Demographic information from the U.S. Census can be combined with psychodemographic profiles, household and consumer expenditure information, population projections, and other data from outside vendors such as National Decision Systems. The PCensus software allows for user-friendly manipulation of multiple databases, and the companion PSearch software allows for quick searching of the databases to find areas with the highest or lowest values on a particular variable, or to rank order areas from top to bottom.
Every ten years, the U.S. Census Bureau conducts an exhaustive census of the population and housing patterns in the United States. This "decennial" census provides a wealth of information that can be valuable in planning prevention programs. The greatest advantage to using U.S. Census data is its completeness. An attempt is made to enumerate every individual in the country, and data are actually collected on nearly 97% of the population. Basic information is collected from all respondents (generating data on about 975 variables) and more detailed information (the "long form") is collected from a large sample of the respondents (generating data on about 3,200 variables). The detailed information from the long form is extrapolated out to estimate the responses for the entire population. The basic (short form) data are contained in the census bureau's Summary Tape File STF-1A, while the detailed (long form) data are contained in the census bureau's Summary Tape File STF-3A. The Indiana Prevention Resource Center (IPRC) has purchased all of the Indiana data files from STF-1A and STF-3A, as well as a reconfigured detailed data set based upon zip codes, rather than census boundaries (STF-3B).
 |
The census provides information about the size of the population, their ages and educational levels, employment patterns, ethnicity, incomes, and household characteristics. Information about family size, the number of parents present in a household, parent education level, parent income, ethnic origin, number of youth living in the household, and countless other variables can be extracted from the census files. Data are reported by "census tract" or "census block," which are small geographic areas that, in Indiana, average about 200 to 2,000 residents. These tracts and blocks can be aggregated to match neighborhood, school district, voting precinct, city, township, or county boundaries. |
PCensus allows data calculation on small neighborhoods, such as this one in Indianapolis
While information is not provided on an "individual household" basis, it is available on an aggregate basis for each small tract or block. In other words, census data will identify the number or percentage of individuals or households meeting a certain classification criteria who live in a defined geographic area -- not the names or identities of those individuals or households.
The 3,200+ STF-3A Variables include such common items as: Population: total number of persons, by age, by gender, by race, detailed, by age and gender, by age, gender, and race, by marital status, by educational attainment, by language spoken, and by disability; Households: total number of households, household size and relationships within households, number of persons in families, age of household members, poverty status of household, poverty status by household type, and poverty status by age of householder; Income: household income, per capita income, aggregate household income, income by race and gender, employment, self-employment, social security income, and public assistance income; Housing: number of housing units, occupancy status, urban and rural status, tenure of household, number of rooms, persons per room, ownership, renter status, percent of income spent on housing, and mortgage or rent payments.
STF-3A also includes arcane data, such as source of water and sewage disposal arrangements, plumbing facilities, means of transportation to work, time spent commuting to work, time individual leaves for work, and number of vehicles available to the household.
Desktop demographic software, such as PCensus, allows a user to locate, manipulate, and aggregate such data, and to export it into spreadsheets or databases for manipulation. It can quickly locate data on variables of interest, and create tables useful in comparing data for two or more geographic areas. Prevention planners can use these data to identify target populations and study their characteristics. Programs can be tailored to meet the specific needs of the target audience, based upon this information (such as literacy levels, language of choice, ethnicity, and family structure). Examples can be tailored to be more culturally sensitive and relevant to the experiences of the target population.
SAMPLE DATA FROM 1990 CENSUS |
||||||
Category |
All of |
Kennedy |
Nixon |
Ford |
Carter |
Reagan |
| Total Population Living in specified area |
5,554,159 |
112,111 |
100,111 |
11,850 |
32,536 |
51,967 |
| Total population under age 6 living in specified area |
479,416 |
12,342 |
10,671 |
859 |
2,002 |
1,993 |
| Total Population Living below federal poverty level |
573,632 |
4,310 |
14,830 |
2,802 |
1,925 |
3,920 |
| Youth population under age 6 living below federal poverty level | 78,780 |
810 |
1,320 |
325 |
801 |
1,060 |
Data shown are hypothetical values shown for illustration purposes only. |
||||||
| EXAMPLES OF HOW CENSUS DATA CAN BE USED IN PREVENTION |
A local high school prevention organization wanted to develop a peer mentoring program for eighth graders from low-income, single parent families who would be entering the school during the next year. They needed information about the number of such youth living within their school boundaries, and the neighborhoods where such youth lived, so they could plan an outreach program over the summer. Census data allowed them to identify the census blocks in which such youth live. |
A parent group wanted to prepare a mailing for other parents. They needed to know how many parents could read English, how many were illiterate, and if there was a need for materials prepared in other languages. Census data provided a profile of the literacy and English language skills of people living in various neighborhoods in the city. |
A local coordinating council was considering conducting a telephone survey of community needs. They wanted to know the percentage of households in their country that did not have telephones. Census data provided this information, as well as information on family size and income levels. |
As extensive as the U.S. Census data are, there are many questions that are not answered by these data. Market research firms have developed a wide range of additional databases for targeted marketing that are useful in prevention planning. Unlike the U.S. Census data, however, these data do not come from an actual enumeration of every household in the country. They come from sophisticated sampling of representative households, and statistical projections that precisely estimate the responses that would have come from a complete enumeration. Extensive data for large area data sets, such as national-, regional-, state-, and county-level data are combined with more limited data from small area data sets, such as city-, town-, school district-, zip code-, neighborhood-, and census tract- and block-level data to create very accurate estimates of the values for the population of each small area.
Three major market research firms, National Decision Systems, Claritas, and Spectra Marketing joined together to combine their unique data sets and proprietary sampling techniques to create a precision estimation methodology known as the VNU-Precision Marketing Group (VNU-PMG) methodology. The combined data and estimation experiences of these firms has created a very precise system for market analysis.
While the U.S. Census provides extensive information about populations and households, it produces little useful information about lifestyles and values. Two households that appear identical in census data (i.e. two-parent, two young adolescents, similar income and education levels, etc.) can have very different lifestyles and values. One could have traditional "white picket fence" values and well mannered children -- like the Ward and June Cleaver household -- while the other might have more hedonistic values and their children may be less supervised -- more like the Homer and Marge Simpson household.
Psychodemographics is the study of lifestyles and values in various demographic groups. Market research firms have developed "lifestyle profiles" that describe the lifestyles and values of different types of households. More than a dozen different psychodemographic systems have been developed for use by businesses in examining their potential customer bases. Each of these systems uses forty to sixty different "lifestyle profiles" to describe the characteristics of household types. The IPRC uses the Microvision 50 system, developed by National Decision Systems.
The Microvision 50 system classifies households into 50 unique lifestyle profiles that identify "market segments." Each of these market segments "consists of households that are at similar points in the life cycle and share common interests, purchasing patterns, financial behavior and needs for products and services." Microvision 50 uses catchy, short descriptive labels, such as "Movers and Shakers," "Bedrock America," and "American Classics".
SAMPLE PSYCHODEMOGRAPHIC DATA FROM MICROVISION 50 |
||||||
Category |
All of |
Reagan |
City of |
Southside |
North Campus |
Suburb |
| Percent "Movers and Shakers" | 12.3% |
15.7% |
18.3% |
28.9% |
14.3% |
20.8% |
| Percent "Books & New Recruits" | 4.3% |
7.8% |
10.1% |
12.3% |
18.2% |
9.1% |
| Percent "Comfortable Times" | 6.8% |
8.8% |
14.8% |
4.3% |
15.2% |
8.1% |
| Percent "University USA" | 4.3% |
27.1% |
17.1% |
20.8% |
48.8% |
14.6% |
Data shown are hypothetical values shown for illustration purposes only. |
||||||
EXAMPLES OF HOW PSYCHODEMOGRAPHIC PROFILES CAN BE USED IN PREVENTION |
A community prevention coalition is planning a recruitment campaign for parents to volunteer with programs for young teens. Using psychdemographic profiles, they can identify neighborhoods where parent involvement is highly valued. They also can identify census blocks where civic-minded senior citizens live, to recruit as mentors for the teen programs. |
A local school corporation is looking for an elementary school to pilot test a violence prevention program with parents who have firearms in their homes. Using psychodemographics, they can identify neighborhoods where hunting is a valued. Combined with information from household expenditures, they should be able to identify schools where the greatest number of children are exposed to guns in their households. |
Daytime Population, Commuting Patterns, and Employment by Industry
Every morning, a large proportion of Hoosiers get in their cars and commute to work. Many cross county lines to reach their jobs; others commute from suburbs to the city. Using 1990 Census data to estimate populations has its limitations when trying to determine the size of a daytime population. Workplace prevention programs, in particular, need a better source of information about the number of people who are present in a given geographic area during the day. Prevention programs for youth also need this information, however. Youth programs often depend upon parent involvement in recruiting youth, arranging transportation for youth, and in participating in parent components of the program. Information about the number of adult family members present in your community during the day, commuting patterns, and employment by industry can provide information vital to program planners.
Market research firms have developed comprehensive databases of businesses and daytime employee populations. Most of these databases utilize a system of Standard Industrial Classification (SIC) codes, which identify the nature and type of businesses. Knowing the nature of the employer's business can be important in determining the times at which commuting occur. For example, many offices operate from 8:00 a.m. to 5:00 p.m., while industrial plants may have a day shift that works from 7:00 a.m. to 3:00 p.m., and a "second shift" that works from 3:00 p.m. to 11:00 p.m. Planning after school prevention programs to keep youth supervised until adult family members get home requires information about when the family members are likely to get off work. The IPRC has obtained a data set from National Decision Systems that allows us to utilize the PCensus desktop demographics software to estimate the daytime employee populations and employment patterns within various geographic areas from the census block level on up.
SAMPLE DAYTIME POPULATION DATA |
||||||
Category |
All of |
Metro |
City of |
Zip Code |
Zip Code |
Zip Code |
| Adults over the age of 18 1998 Population Projection |
3,185,999 |
512,936 |
135,131 |
4,901 |
12,820 |
9,967 |
| Adults over the age of 18 Daytime Population Estimate |
3,170,999 |
321,111 |
65,196 |
450 |
905 |
1,060 |
| Data shown are hypothetical values shown for illustration purposes only. | ||||||
EXAMPLES OF HOW DAYTIME POPULATION CAN BE USED IN PREVENTION |
A community prevention coalition was concerned about the large number of new alcohol outlets in one part of their county. They wanted to know if the number of restaurants with alcoholic beverage licenses was disproportionately high for the natural demand. The restaurant industry argued that new restaurants were needed to serve the people working in that area. Using daytime population counts, as well as normal census counts, they were able to compare the number of licenses per capita, to see if there were parts of their county where alcohol outlets were too dense. They can use full color maps to identify "high density per capita" neighborhoods. Statistical information, such as this, can improve their chances, if they seek to have alcoholic beverage control boards restrict new outlets in neighborhoods that already have too many alcohol outlets. |
A parent education program wanted to reach more parents with their programming, and decided to offer "brown bag lunch" programs on parenting skills. Using daytime population statistics and mapping them by census block, they could identify the parts of their city where the largest number of people worked, and this suggested convenient lunchtime program locations. |
Have you ever tried to find local statistics on alcohol consumption for use in a grant application? If you search for government statistics in Indiana, good luck! Local alcohol sales do not need to be reported to the State, so there is no way to extract anything close to alcohol consumption statistics from excise tax records or license applications, or anywhere else. The same thing goes for tobacco. Tobacco sales and excise taxes are not reported to the State on a county-by-county, or other local basis. Although a few counties may have data from local prevalence surveys, meaningful comparisons of use rates between and among counties or other local entities are not possible from government-collected statistics.
Market research firms, however, have access to a number of sources of information about household and consumer expenditures. Using data from household surveys and diaries, sales figures reported by wholesalers and manufacturers, and barcode scanning data from retailers, and other sources, market research firms can precisely estimate such expenditures at the census tract amd census block level. Once available at this level, desktop demographic software can aggregate the data into larger geographic areas, such as neighborhoods, zip code boundaries, city and town limits, and school districts. This would allow a community to compare alcohol consumption patterns between and among various parts of a city or county.
This type of data is expensive to collect and also expensive to purchase. The IPRC has purchased data files on 100 specific consumer commodity groups, for all geographic areas in Indiana (about 20% of all available data), including all forms of alcohol and tobacco products, over-the-counter and prescription drugs, health care, and expenditures for education and recreation. This allows the IPRC to provide relevant data to prevention programs, and still control the cost of the data files.
Consumer and household spending on alcohol and tobacco products is currently the only indicator of alcohol and tobacco use prevalence that is available for ALL counties and geographic areas in Indiana. Data can be mapped using a Geographic Information System, and the location of alcohol-specific incidents (ie. impaired driving arrests) can be superimposed over the consumption data.
SAMPLE HOUSEHOLD CONSUMER EXPENDITURE DATA |
||||||
Category |
All of |
Campus |
Downtown |
Westside |
Fraternity |
East
Campus |
| Total purchases of beer for consumption at home |
$925,985,999 |
$8,212,111 |
$100,111 |
$850,321 |
$912,500 |
$91,967 |
| Per household purchases
of beer for consumption at home |
$301 |
$780 |
$422 |
$450 |
$1,900 |
$860 |
Data shown are hypothetical values shown for illustration purposes only. |
||||||
HOW HOUSEHOLD CONSUMER EXPENDITURES CAN BE USED IN PREVENTION PLANNING |
An impaired driving prevention program wanted to compare alcohol consumption patterns in various neighborhoods, to help them determine priorities for future programming. They created a color map that showed differences among census blocks in expenditures for alcoholic beverages per household. They further refined their maps by noting differences between alcohol purchased for in-home consumption, versus purchases for on-site consumption. |
A for-profit after school prevention program wants to identify potential locations for expansion programs. They integrate data on the number of youth in their target age range with household expenditures for fees for recreational lessons and participant fees, to identify neighborhoods where families are accustomed to paying for youth's enrollment in such programs. |
A youth-serving organization is planning a capital campaign to fund a new indoor swimming pool. They use household expenditure records to identify neighborhoods where families have above-average expenditures on both recreation and on charitable contributions. Their fundraising efforts can then be targeted in these neighborhoods. |
One of the big disadvantages in using 1990 Census data for population statistics in 1998 is the "age" of the data. Since the 1990 Census, several Indiana communities have experienced military base closures and plant shutdowns. Other communities, particularly in suburban areas around large cities, have seen their populations increase dramatically. Data that are eight and one-half years old may be inadequate for planning purposes in communities that have experienced significant population changes.
Market research firms use a wide variety of other data to modify the population figures from the 1990 Census. Using data on building permits, employment and unemployment data, real estate transactions, apartment vacancy rates, birth and death records, and similar data sources, the Precision Marketing Group, including National Decision Systems, has produced very accurate demographic updates and population projections for 1998, as well as estimates of expected growth by the year 2003. The IPRC had purchased these data sets for all geographic areas in Indiana, from census blocks on up, which can be used by prevention program planners to better estimate the size and demographics of their target populations.
SAMPLE DATA FROM POPULATION PROJECTIONS DATABASE |
||||||
Category |
All of |
Carter |
City of |
City of |
Town of |
City of |
| 1980 Census | 5,185,999 |
212,331 |
100,111 |
2,850 |
12,500 |
1,967 |
| 1990 Census | 5,554,156 |
218,130 |
103,118 |
2,350 |
13,041 |
1,988 |
| 1998 Population Estimates | 5,870,319 |
224,731 |
104,773 |
2,458 |
12,920 |
2,060 |
| 2003 Population Estimates | 5,982,909 |
232,114 |
109,351 |
2,850 |
12,995 |
2,067 |
Data shown are hypothetical values shown for illustration purposes only. |
||||||
EXAMPLES OF HOW POPULATION ESTIMATES CAN BE USED IN PREVENTION |
| A local parents' group is trying to start a new teen center in a rapidly growing suburban community. They know that the population in their county has nearly doubled since the 1990 census, and they need better estimates of the current population of teen-agers, and how many teens will be living in the community in five years, to more accurately establish the need for such a center. Using 1998 population estimates, they can map the changes in teen population density, neighborhood-by-neighborhood since 1980, and establish estimates for the year 2003. |
A community prevention coalition is planning a mass distribution of educational materials on underage drinking prevention for parents of teens. Due to the closing of a military base and a large factory, there have been major changes in the mixture of people living in neighborhoods in their community. Using 1998 population estimates, they can more accurately identify the neighborhoods where the greatest numbers of high school students live. |
A local parks department is planning new programs for middle school youth. They want to know the impact that two new housing developments and a large apartment complex on the south side of their town have had on the population of 12 through 14 year olds. Using color maps, they can chart the changes in young adolescent population from 1990 to 1998, to use in estimating the demand for their new programs. They also can integrate household expenditure data to see if these families are accustomed to paying fees for their youth's recreational activities. |
Source: