Multiple Health Risk Perception and Information Processing Among African Americans and Whites Living in Poverty
Abstract
We investigated the risk-information-processing behaviors of people living at or near the poverty line. Because significant gaps in health and communication exist among high- and low-income groups, increasing the information seeking and knowledge of poor individuals may help them better understand risks to their health and increase their engagement in health-protective behaviors. Most earlier studies assessed only a single health risk selected by the researcher, whereas we listed 10 health risks and allowed the respondents to identify the one that they worried about most but took little action to prevent. Using this risk, we tested one pathway inspired by the risk information seeking and processing model to examine predictors of information insufficiency and of systematic processing and extended this pathway to include health-protective action. A phone survey was conducted of African Americans and whites living in the southern United States with an annual income of ≤$35,000 (N= 431). The results supported the model pathway: worry partially mediated the relationship between perceived risk and information insufficiency, which, in turn, increased systematic processing. In addition, systematic processing increased health-protective action. Compared with whites and better educated respondents, African Americans and respondents with little education had significantly higher levels of information insufficiency but higher levels of systematic processing and health-protective action. That systematic processing and knowledge influenced health behavior suggests a potential strategy for reducing health disparities.
1. INTRODUCTION
If any progress is to be made in reducing health disparities, whereby persons at greater social disadvantage—such as those living at or near the poverty line—also experience worse health or greater health risks,(1) a variety of strategies must be explored to increase the health-protective behaviors and decrease the health risks among these vulnerable populations. We investigated the risk-information-processing behaviors of people in poverty, looking specifically at health risks they worried about but did relatively little to prevent. Increasing knowledge and information-seeking behavior among this group may help them to better understand their risks and to engage in health-protective behaviors to reduce these risks.
1.1. Poverty, Health, and Information Processing
Income level is one of the strongest contributors to disparities in health(2) and health knowledge.(3) People who live at or near the poverty line suffer disproportionately from many health risks. Socioeconomic factors, such as income and education, are associated with a higher incidence of many chronic health conditions, including cancer, diabetes, and heart disease.(4-6) People living in poverty are also more vulnerable to risks such as natural disasters and accidents and often are less able to protect themselves against these risks.(7, 8) While many factors are known to contribute to these disparities, risk-related knowledge and information-seeking behavior are important to consider as well. The people who experience health disparities are often the same people who experience communication disparities or gaps in knowledge and access to information.(3, 9) Groups with lower socioeconomic status tend to have less knowledge about a wide array of health issues and topics,(10-13) which may limit their participation in health-protective behaviors. Information may increase one's knowledge of health issues and minimize health risks through participation in screening and protective behaviors, yet people with lower income and education levels generally seek less information.(14, 15) They also do not receive information as quickly,(3, 13, 16) perhaps due to more limited access to communication channels.(17-19) Finally, people in poverty may be less motivated to seek information and more likely to face barriers (e.g., a lack of time or money) that prevent them from gathering information or influence their capacity to do so.(20, 21)
1.2. Theoretical Framework
Communication disparities have been studied primarily by examining differential levels of knowledge, access, and use of communication channels that exist between groups categorized by socioeconomic status. However, the literature suggests other contributing factors to communication inequalities, including risk perceptions, norms for information seeking, desire for information, and ability to process and make sense of information, that mediate the effects of individual characteristics on information seeking.(22, 23) To date no research has been conducted to assess the degree to which low-income individuals process the information they do seek or the effect of this processing on their health behavior. Although deeper information processing is associated with more stable attitudes,(24-26) it is not clear whether it also increases health-protective action that may reduce health disparities.
A number of information-seeking frameworks consider the influence of sociodemographic variables on information-seeking behaviors,(21, 27, 28) but very few consider how these factors affect risk perceptions and, in turn, information seeking and processing. To investigate the links between risk perceptions and information processing we drew from one of these models, the risk information seeking and processing model (RISP).(21) We focused on one critical pathway in the model (Fig. 1): increased perceptions of risk trigger an emotional response, worry, which in turn triggers an assessment of information needs and subsequent information seeking and processing. We also extended this pathway to examine the effect of information processing on health-protective behaviors.
Risk information seeking and processing pathway tested and proposed relationship with health-protective behavior.
The pathway we tested is an extension of the heuristic-systematic processing model,(29) which proposes two routes for processing information, a deeper systematic processing and a shallower heuristic processing. Systematic processing involves a significant amount of effort gathering and evaluating information, whereas heuristic processing is more limited and relies primarily on knowledge already accessible to the individual.(29) According to the RISP model, information-processing results from information insufficiency, which is a discrepancy between the amount of risk-related information people have and the amount of information they desire. Information insufficiencies, in turn, result from the perception of risk (labeled “perceived hazard characteristics” in the RISP model) and an accompanying affective response to the risk (e.g., worry). Although worry has not been confirmed as a mediator within the RISP framework, perceived risk has been shown to positively predict worry, and worry has been shown to predict information insufficiency.(30, 31)
Individual characteristics (e.g., demographic or sociocultural variables, relevant hazard experience, and political philosophy) are believed to operate in the deep background of this information seeking and processing.(32) In the RISP model and other theoretical frameworks, the effect of individual characteristics on information seeking and processing is proposed to be indirect and mediated by other model variables.(21, 27) In the RISP model, individual characteristics are typically included as control variables. We treated them similarly here but included only one of the model's proposed mediators, perceived risk, which allowed us to test a more parsimonious model. While we recognized that individual characteristics may be working through other proposed mediators proposed by the full RISP model (i.e., informational subjective norms, channel beliefs, and information-gathering capacity), the goal of our study was to gain a greater understanding of the basic relationships between risk perceptions and systematic processing and to extend one pathway inspired by the model to health behavior.
Griffin et al. expected information seeking and processing behaviors to ultimately effect engagement in health-protective behaviors,(21) but this has not been tested. However, previous articles have suggested a link between RISP and preventive behaviors via the theory of planned behavior (TPB), a theory predicting human behavior that proposes that behavioral achievement is the result of intention, behavioral control, informational subjective norms, and attitudes toward the behavior.(21, 33-35) Specifically, three RISP variables (i.e., individual characteristics, information seeking and processing, and perceived risk) are proposed to affect three precursors of behavioral intention in the TPB (i.e., subjective norms, attitudes toward the behavior, and perceived behavioral control), and two TPB variables are included in the RISP model (i.e., informational subjective norms and perceived information-gathering capacity). Several studies have confirmed that systematic processing positively increases attitudes toward behavior and perceptions of risk,(33, 36) but no studies have examined whether these variables affect health behaviors specifically. We examined this relationship in our study.
Although we did not explore all of the proposed mechanisms of information processing and behavior change within the RISP or TPB models, we are the first to examine whether there is a direct relationship between risk-information processing and health-protective behavior among people living in poverty, as other researchers have suggested (e.g., the structural influence model).(37) Communication is believed to be one pathway through which social determinants of health affect health outcomes such that individuals with a greater ability to seek and process information are expected to experience better health outcomes and an increased ability to engage in health-protective behavior.(38, 39)
1.3. Hypotheses
In this study we drew on the RISP model to test the relationships between risk perceptions, information needs and processing, and health-protective behavior among people living in poverty. From a health communication standpoint, we have a primary interest in systematic processing and the extent to which people seek and engage with information. People who engage in more effortful information seeking and processing may develop risk-related thoughts, attitudes, and behaviors that are more stable.(21, 25) We hypothesized that (1) risk perceptions will be positively related to worry, (2) worry will be positively related to information insufficiency, (3) information insufficiency will be positively related to systematic information processing, and (4) systematic processing will be positively related to health-protective action.
2. METHODS
2.1. Survey
Our study was part of a larger investigation of multiple risk perceptions and health-protective behaviors among individuals at or near the poverty line who were living in the southern United States. Our data were drawn from information collected during a 2007 telephone survey administered to individuals with landline telephones in South Carolina, Georgia, Alabama, Mississippi, and Louisiana households by the research firm ICF Macro. The sample had initially been drawn using random digit dialing in ZIP codes with large numbers of households with an annual income of ≤$35,000. Because this technique failed to yield the number of participants we desired, we used targeted lists of households (provided by Info USA) who met the income screening criteria.
For inclusion in our study, participants were screened based on annual household income ≤$35,000, African American or white race, and age 22–64 years. Of all the participants who met the criteria (N= 431), 70% had a household income of ≤$25,000 a year (N= 302). Among the 1,215 households confirmed as eligible to participate, survey completion was approximately 35%. The final unweighted sample was split nearly equally in terms of race, and the mean age was 48.6 years (SD= 11.26). Seventy-one percent of the sample was female (N= 306). Nearly 78% of participants had a high school diploma (N= 334), and <10% of respondents had a college degree (N= 41).
Once eligibility for inclusion in the study was determined, each participant was given a list of 10 health risks, each one of which the respondent was to rate his or her level of worry on a 5-point scale from 1 (not worried at all) to 5 (extremely worried): diabetes, obesity, cancer, bird flu, getting in a car or truck accident, heart disease, sexually transmitted disease (including HIV/AIDS), getting hurt in a natural disaster, stroke, and arthritis. Each participant was then read back the risks he or she assigned the highest worry ratings to (starting with risks with a worry score of ≥4). Whereas most studies look at only a single risk selected by the researcher, we asked respondents to self-identify out of these high-worry risks the one risk that they did the most to protect themselves against and the one risk they did the least to protect themselves against. Finally, for these two self-identified risks, the participant was asked a series of questions about his or her risk perceptions, information seeking and processing, and health-protective action. In this study we looked specifically at the risk that was identified as high worry but low action because this kind of risk may have the most potential for interventions. The risk that participants identified as their high-worry/low-action risk varied across the sample. Natural disaster was most often chosen by study participants (17.8%; n= 77) and was followed by cancer (14.8%; n= 64), bird flu (14.7%; n= 63), arthritis (13.8%; n= 60), obesity (8.2%; n= 35), car or truck accident (7.8%; n= 33), diabetes (7.2%; n= 31), stroke (6.5%; n= 28), heart disease (6.0%; n= 26), and, finally, sexually transmitted disease (3.2%; n= 14).
2.2. Survey Measures
2.2.1. Worry
As mentioned previously, respondents were initially asked to use a 5-point scale to indicate how concerned or worried they were about each of the 10 health risks at this point in their lives. Only worry scores associated with the risk self-identified as high worry but low action were used here. Worry was measured on a 5-point scale from 1 (not worried at all) to 5 (extremely worried).
2.2.2. Perceived Risk
Earlier versions of the RISP model have called the perception of risk “perceived hazard characteristics.” These characteristics are a compilation of factors such as institutional trust, personal control, risk judgment (including severity and probability), and even attributions for responsibility.(30, 40) Because we were interested in individual-level health risks as opposed to population-based environmental risks, we focused only on the characteristic of risk judgment because it was the most relevant to our current study. As in past RISP studies, severity and susceptibility were used as measures of risk judgment.(30)
Participants scored the perceived severity of their high-worry/low-action risk on a 5-point scale ranging from 1 (not at all serious) to 5 (very serious). Perceived susceptibility was assessed with the item “How likely do you think you are to get [risk]?” and was also scored on a 5-point scale, from 1 (not likely at all) to 5 (very likely). Individuals who indicated that they already had that condition were dropped from all analyses, as were participants who were missing severity and/or susceptibility scores (n= 34). Factor analysis using principal component analysis confirmed that the two measures were one-dimensional. For this reason, we summed the severity and susceptibility scores to provide an overall perceived risk score that ranged from 2 to 10.
2.2.3. Information Insufficiency
Information insufficiency refers to the amount of knowledge people desire about a risk, relative to their current knowledge. Study participants were asked to self-report current knowledge of their high-worry/low-action risk on a 5-point scale ranging from 1 (don't know anything) to 5 (know everything possible). Similarly, participants self-reported desired knowledge of that risk (i.e., information insufficiency) on a scale ranging from 1 (nothing more) to 5 (everything possible). These measures of current knowledge and desired knowledge were entered separately in regression analysis, with the former entered first to account for variance not explained by current knowledge.4
2.2.4. Systematic Processing
Information processing of a participant's high-worry/low-action risk was measured using Trumbo's scales of heuristic and systematic processing.(36) Heuristic processing was assessed using items (e.g., “Based on what I know I can make a decision about what I should do about [risk]”) that exhibited very low reliability (α= 0.52), so we did not include heuristic processing in our study. To measure systematic processing of a participant's high-worry/low-action risk, respondents were asked to indicate their level of agreement with the statements: “I try to get as much information as I can about the risk,”“I believe the more opinions I can get about the risk the better off I will be,” and “When I see information about the risk I am likely to stop and read or think about it.” Each item was scored on a 5-point scale ranging from 1 (strongly disagree) to 5 (strongly agree). Factor analysis with principal component analysis confirmed the unidimensionality of the three items. Because the second item was weakly correlated with the factor, we dropped this item and summed the scores to produce a single score for systematic processing (possible range, 2–10) (unweighted α= 0.70).
2.2.5. Health-Protective Action
Health-protective action regarding a participant's high-worry/low-action risk was assessed by asking: “How would you describe how much you are doing to protect yourself from [risk]?” The action level was measured on a 5-point scale, from 1 (nothing) to 5 (everything).
2.2.6. Individual Characteristics
Four individual characteristics were measured: age (continuous variable), race (1 = white; 0 = African American), sex (1 = female; 0 = male), and education. Education was measured by asking respondents the highest grade or year of school they had completed; this 6-point scale ranged from 1 (never attended) to 6 (≥4 years of college).
2.3. Data Analysis
Our data set was weighted based on the 2006 U.S. Census Bureau Current Population Survey, Annual Social and Economic Supplement, to match state, age, and sex population totals for African-American and white adults in the five-state area who were between the ages of 22 and 64 years and residing in a household with a combined income of ≤$35,000. Hierarchical multiple Tobit regressions were conducted using Stata software version 10.1. Control variables (i.e., individual characteristics and knowledge) were entered first so that the subsequent variables predicted any variance not accounted for by these two variables.(40)
We used Tobit regression because the dependent variables (i.e., worry, desired information, systematic processing, and health-protective action) were censored. That is, even though these variables could be observed over their entire range of values, a high percentage of the responses clustered at the high end of the scales. Tobit regression accounts for this censoring and provides a more consistent estimate of the parameters.(41) Because this method uses a maximum likelihood method of estimation rather than ordinary least squares, we used likelihood ratio χ2 and pseudo R2 values to assess model fit. The McKelvey-Zavonia pseudo R2 value was used because it has been shown to be the best estimator of explained variance.(42) The Tobit coefficients can be further decomposed into estimates of marginal effect (ME) and linear effect (LE) to provide additional information about the relationships between the independent and dependent variables above or below their upper or lower value thresholds. In this study, ME is the probability of a dependent variable score being below its upper threshold per one-unit change in the independent variable. LE refers to the change in the dependent variable below its upper threshold per one-unit change in the independent variable.(43, 44)
3. RESULTS
Descriptive statistics were calculated for the six model variables; the percentage of observations at the upper thresholds is reported for the four dependent variables (Table I). Before we conducted our data analysis, weighted model and demographic variables were correlated using Spearman correlations to screen for multicollinearity. Correlations between variables were modest, with the absolute values ranging from <0.01 to 0.44.
| Variable | Minimum | Maximum | N* | M | SD | Censored |
|---|---|---|---|---|---|---|
| Worry | 1 | 5 | 431 | 3.67 | 1.10 | 29% at 5 |
| Perceived risk | 2 | 10 | 397 | 8.26 | 1.76 | – |
| Information insufficiency | ||||||
| Current knowledge | 1 | 5 | 426 | 2.79 | 1.03 | – |
| Desired knowledge | 1 | 5 | 428 | 3.70 | 1.41 | 44% at 5 |
| Systematic processing | 2 | 10 | 426 | 8.03 | 2.21 | 36% at 10 |
| Health-protective action | 1 | 5 | 337 | 3.47 | 1.52 | 42% at 5 |
- The total number of responses (N) differs slightly for each variable due to missing data.
3.1 Predictors of Worry
Controlling for individual characteristics, our study results supported the hypothesis that perceived risk positively increased worry (p < 0.01; ME =−3%; LE = 0.10; upper threshold = 5) (Table II). Worry was also significantly higher among females, younger respondents, and persons with less education.
| Independent variable | Worry | Information Insufficiency (Desired Knowledge) | Systematic Processing | Health-Protective Action | ||||
|---|---|---|---|---|---|---|---|---|
| Tobit | MEa(LE)b | Tobit | ME(LE) | Tobit | ME(LE) | Tobit | ME(LE) | |
| Individual characteristics | ||||||||
| Racec | −0.02 | 0%(−0.01) | −0.58* | 11%(−0.28) | −0.65* | 8%(−0.37) | −1.59** | 25%(−1.02) |
| Sexd | −0.36* | 7%(−0.23) | −0.28 | 5%(−0.14) | −0.45 | 5%(−0.26) | −0.49** | 7%(−0.33) |
| Age | −0.01* | 0%(−0.01) | 0.01 | 0%(0.00) | 0.04** | −1%(0.02) | 0.04** | −1%(0.02) |
| Education | −0.25** | 5%(−0.16) | −0.37** | 7%(−0.19) | −0.47** | 6%(−0.27) | −0.18* | 3%(−0.12) |
| Δ pseudo R2 | 0.09 | 0.11 | 0.17 | 0.24 | ||||
| Current knowledge | 0.02 | 0%(0.01) | 0.78** | −9%(0.44) | 0.43** | −6%(0.29) | ||
| Desired knowledge | 0.76** | −9%(0.44) | 0.03 | 0%(0.02) | ||||
| Δ pseudo R2 | 0.00 | 0.20 | 0.11 | |||||
| Worry | 0.30** | −6%(0.15) | 0.32* | −4%(0.18) | 0.03 | 0%(0.02) | ||
| Δ pseudo R2 | 0.04 | 0.01 | 0.00 | |||||
| Perceived risk | 0.16** | −3%(0.10) | 0.16** | −3%(0.08) | 0.10 | −1%(0.06) | −0.01 | 0%(−0.01) |
| Δ pseudo R2 | 0.05 | 0.04 | 0.01 | 0.06 | ||||
| Systematic processing | 0.23** | −3%(0.15) | ||||||
| Δ pseudo R2 | 0.04 | |||||||
| Likelihood ratio χ2 | 56.33** | 76.29** | 178.84** | 171.69** | ||||
| Pseudo R2 | 0.14 | 0.19 | 0.39 | 0.45 | ||||
- *p < 0.05, **p < 0.01.
- aMarginal effect (ME) is the probability that a dependent variable score is below the upper threshold with a one-unit change in the independent variable.
- bLinear effect (LE) is the change in the dependent variable below its upper threshold with a one-unit change in the independent variable.
- cAfrican American = 0, White = 1.
- dFemale = 0, Male = 1.
3.2 Predictors of Information Insufficiency
When we controlled for individual characteristics and current knowledge, our results confirmed our second hypothesis that worry would positively increase information insufficiency (measured as desired knowledge) (p < 0.01; ME =−6%; LE = 0.15; upper threshold = 5) (Table II). Perceived risk was also shown to positively predict information insufficiency (p < 0.01; ME =−3%; LE = 0.08; upper threshold = 5). African Americans and persons with less education had significantly higher levels of information insufficiency.
That worry and perceived risk were both significant in the model tested suggest that worry may partially mediate the relationship between perceived risk and information insufficiency. To determine whether this was the case, we ran a third regression as suggested by Baron and Kenny(45) to determine whether perceived risk was also a significant predictor of information insufficiency. Controlling for individual characteristics and knowledge, our results showed that perceived risk was indeed a significant predictor of information insufficiency (Tobit = 0.19, p < 0.01; ME =−4%, LE = 0.10; upper threshold = 5). Because worry and perceived risk were significant predictors of information insufficiency, both independently and in combination, we concluded that worry acted as a partial mediator of the relationship between perceived risk and information insufficiency.
3.3 Predictors of Systematic Processing
We assessed whether information insufficiency increased systematic processing when knowledge, risk perceptions, worry, and individual characteristics were controlled for. The results supported our third hypothesis that information insufficiency was a positive predictor of systematic processing (p < 0.01; ME =−9%; LE = 0.44; upper threshold = 10) (Table II). Current knowledge was also a positive predictor of systematic processing (p < 0.01; ME =−9%; LE = 0.44; upper threshold = 10). These two independent variables accounted for more than half of the total pseudo R2 value. Moreover, worry positively increased systematic processing (p < 0.05; ME =−4%; LE = 0.18; upper threshold = 10). Systematic processing was significantly higher among African Americans, older participants, and less educated participants.
3.4. Predictors of Health-Protective Action
Finally, with individual characteristics and indirect RISP predictors (i.e., knowledge, worry, perceived risk, and information insufficiency) being controlled for, our study results showed that systematic processing significantly predicted health-protective action (p < 0.01; ME =−3%; LE = 0.15; upper threshold = 5) (Table II), which supported our fourth hypothesis. Current knowledge also positively increased health-protective action (p <0.01; ME =−6%; LE = 0.29; upper threshold = 5). Being older, African American, less educated, or female significantly increased health-protective action. Perhaps the most striking results were the MEs and LEs for race. Compared with African Americans, whites engaged in fewer preventive efforts. Whites were 25% more likely to report an action score <5, and their scores were 1.02 lower below that upper threshold. This finding is likely due to the fact that 55% of African Americans versus 29% of whites had reported they were doing the most they could to protect themselves.
4. DISCUSSION
In this study we investigated the information processing and health behaviors of individuals living in poverty, looking specifically at the category of risks they worry about but take relatively little action to protect themselves against. Rather than examining only one specific risk that may or may not be relevant to participants, we allowed them to self-identify their highest worry but lowest action risk. Because the goal of our study was to examine the relationship between variables such as risk perceptions, information processing, and health-protective behavior, the risk itself was less important. Although we sacrificed the ability to report on specific risks, this approach ensured that we examined only the risks that were salient to these low-income respondents and thus allowed us to better understand the factors that motivate them to take health-protective action. While this approach does not take into account the psychometric properties of the risks themselves (i.e., that some risks are inherently more “risky” than others), it does recognize and support previous research showing that individuals often vary in their perceptions and characterization of these risks.(46-48)
We investigated the hypothesis that risk perceptions increase worry, which, in turn, triggers feelings of information insufficiency and subsequently systematic processing. This study represents the first test of this critical pathway within the RISP model in the context of personal health behaviors, and exclusively with people living in poverty. It is also the first study to extend this pathway to health-protective behavior. Because communication and health inequalities are related,(9) we expected systematic processing to increase the likelihood of health-protective action against perceived risks.
The results of our assessment of the relationships within this critical pathway were largely consistent with results from previous research. In our study, worry was a direct predictor of information insufficiency and a mediator of the relationship between perceived risk and information insufficiency, which supports past research on the RISP model.(30, 34) Thus, emotional responses to a risk, and not simply risk alone, appear to influence people's information-related perceptions and behaviors and should be considered when prevention messages for low-income audiences are developed. As in previous studies,(31, 49) we also found that information insufficiency was a direct predictor of systematic processing in addition to current knowledge. Although this finding further demonstrates the notion that information processing is driven by a need for more information, our results also suggest that some people may seek and process information systematically even when knowledge is adequate, which could be a reflection of participants’ level of health consciousness or health orientation.(50)
When we extended the RISP pathway to protective behavior, systematic processing increased the likelihood of health-protective action along with knowledge. Our findings support the idea that deeper information processing is associated with not only more stable attitudes(24-26) but also health-protective behaviors, even among people in poverty who may have more limited resources and access to health care. Additionally, our results emphasize the importance of eliminating communication disparities and increasing knowledge about health risks because they may increase healthy behaviors and, in turn, reduce health disparities.
Our results also point to important individual differences among people living in poverty in systematic processing and health-protective action. Low-income African Americans and low-income persons with less education were significantly more likely than their white and more educated counterparts to seek and systematically process information. These findings surprised us for several reasons. First, previous research has generally found individuals with lower socioeconomic status to be less likely and less motivated to seek information(14, 20) and that African Americans tend to seek less health information than whites do.(51, 52) While our results refute these commonly held expectations, they may also reflect communication inequalities in terms of access to health information or health knowledge between high and low socioeconomic groups.(53) Our findings might indicate that those with less knowledge must put forth more effort to obtain needed information, but we cannot be certain because our study design excluded the possibility of a higher income group for comparison. Our study showed that African Americans were more likely to experience information insufficiency, which is consistent with other literature showing that minority groups may lack knowledge about health conditions and what steps they should take to protect themselves.(12, 54-56) Our results are also intriguing because they refute the notion that people with less income are less likely to seek and systematically process the information they desire. However, these findings must be tempered by the fact that three proposed mediators of the relationship between individual characteristics and systematic processing (i.e., information-seeking norms, channel beliefs, and perceived gathering capacity) in the RISP model were not included in our analysis and might account for these relationships.
In our study, African Americans, persons with less education, women, and older individuals were also more likely to engage in health-protective action. This finding implies that individuals who are more likely to systematically process information are also more likely to engage in health-protective action. While the literature has shown that being female and being older are positive predictors of adherence to multiple health-related behaviors,(57) which is consistent with our results, it has also been shown that African Americans are less likely to engage in a variety of health behaviors,(58, 59) contrary to our findings. A possible explanation for this discrepancy is that African Americans’ perceptions of action were higher than their actual behaviors would warrant, or that whites’ perceptions of action were lower than their actual behaviors. Because we had asked the study participants to self-report how much they were doing to protect themselves, rather than observing them or having them record their behaviors, our scores may reflect a self-report bias or limited knowledge of the participants about how much action is needed to protect against some risks.
4.1. Limitations
Having a large sample of low-income African-American and white individuals is a major strength of our study, but it is also a limitation because we did not include a comparison group with a higher income. It is therefore not possible to make comparison statements about low- versus high-income groups. Given our emphasis on individuals residing in the southern United States, we also did not look at regional differences among participants nor compare our results to other regions of the country. Additionally, because we explored multiple risks and let the participants identify the risk they were threatened by but took little action to prevent, we were unable to draw conclusions about specific risks. Future studies should examine whether these results are similar or different for specific risks.
Although we confirmed one critical path in the RISP model, our investigation did not include the three variables proposed to mediate the relationship between individual characteristics and systematic processing (i.e., information-seeking norms, channel beliefs, and perceived gathering capacity). This is a limitation because individual characteristics may be working through these variables to affect outcomes such as information insufficiency and systematic processing. Future efforts to understand risk information seeking and processing among people living in poverty must include these mediators to fully test the RISP model and to gain a better understanding of the factors that influence information needs, processing, and behavior.
5. CONCLUSION
This study is the first to look specifically at the health information seeking and processing behaviors of low-income individuals to better understand the links between information processing and health-protective action. Our results support the relationships between cognitive and emotional responses to risk, information insufficiency, and systematic processing that are described in the RISP model. They also illustrate the significant influence of information processing and knowledge on health behavior and a potential pathway for reducing health disparities. Furthermore, these results demonstrate that we must look closely at the ways in which social and individual determinants (including and beyond race, age, education, and sex) influence information seeking and processing, much as they do in health. To design more effective communication strategies to increase individuals’ knowledge and to reduce communication and health disparities, we must continue to explore the mechanisms by which these social determinants affect information seeking.
Footnotes
ACKNOWLEDGMENTS
This research was conducted by the Southern Center for Communication Health and Poverty (Vicki Freimuth, Principal Investigator) and funded by the Centers for Disease Control and Prevention as a Center of Excellence in Health Marketing and Health Communication (5PO1CD000242-03). Shelly Hovick is supported by the Kellogg Health Scholars Program, under a grant (P0117943) from the W.K. Kellogg Foundation to the Center for Advancing Health.
