Relationship between Socioeconomic characteristics and healthy behaviour for the British Columbia population
Zabid Iqbal
2025-12-10
1 Introduction
This report provides an analysis on the relationship between socioeconomic and demographic characteristics and healthy behaviour for British Columbia population using Canadian Community Health Survey 2017-2018 (CCHS) data and applying applied econometric/statistical model(s). The analysis starts with providing a description on the construction of healthy behavior index (a proxy for healthy behavior) and a summary analysis of the index.
2 Healthy Behaviour/How do we define healthy behaviour?
The analysis constructs health behavior index (HBI) to explain healthy behavior of British Columbia population using Canadian Community Health Survey 2017-2018 (CCHS). The existing literature usually use four health behaviour indicators to construct HBI or health behaviours score. The indicators and their details are given below in Table 1. We can see there are four indicators: 1. Physical activity, 2. Smoking status, 3. Drinking status, and 4. Fruit and vegetable consumption. We use first 3 indicators to construct HBI for BC. We exclude fruit and vegetable consumption in our method because the data on this indicator was not available for BC. Fruit and vegetable consumption was optional module in 2017-2018 community health survey data. In constructing HBI, this report first converts each healthy behaviour indicator into a binary variable: poor health behavior (0) and good healthy behavior (1). In the next step, HBI is constructed by adding all the binary indicators as constructed in the first step. This gives us a composite HBI score ranging from 0 to 3. For simplicity and convenience of interpretation, this report categorizes HBI into three groups: 1 (poor healthy behavior), 2 (fair healthy behavior), and 3 (good healthy behavior). Therefore, the constructed and/or transformed HBI varies from 1 (poor behavior) to 3 (good behavior).
| Health.indicator | Negative.Health.Behaviour | Positive.Health.Behaviour |
|---|---|---|
| Physical activity(1) | Less than 150 minutes of activity per week | Physically active 150 minutes or more per week |
| Smoking status | Current daily or occasional smoker | Current non-smoker |
| Drinking status | Heavy drinker(2) in the past 12 months | Not a heavy drinker in the past 12 months |
| Fruit and vegetable consumption | Consumed fruits and vegetables less than five times per day | Consumed fruits and vegetables five or more times per day |
| Note: | ||
|
1.Physically active is defined by the Canadian Physical Activity Guidelines as having at least 150 minutes of moderate to vigorous intensity aerobic physical activity per week (causing the respondent to sweat a little and breathe harder), in bouts of 10 minutes or more. This is based on self-reported physical activity. 2. Heavy drinkers are men who consumed 5 or more drinks per occasion, at least once a month in the past year. Women are heavy drinkers if they consumed 4 or more drinks per occasion, at least once a month in the past year. |
Source: https://www150.statcan.gc.ca/n1/pub/82-625-x/2018001/article/54975-eng.htm https://www150.statcan.gc.ca/n1/pub/82-229-x/2009001/deter/int3-eng.htm
Figures 1 and 2 show the percentage and frequency of the categories of HBI by health regions of BC, respectively. From figure 1, I find that the top three regions in terms of percentage of the good healthy behavior maintained by the BC population are North Shore/Coast Gar (45.30%), South Vancouver Island (45.20%) and Vancouver HSDA (45%). The bottom three regions in terms of percentage of the good healthy behavior maintained by the BC population are Northeast HSDA (31.9%), Thompson/Cariboo HSDA (31.9%), and Northwest HSDA (37.2%). This means the percentage distribution of poor healthy behaviour in these three regions is likely to be higher. The two poorest regrion in terms of poor health behavior are two of these three regions. The three poorest regions are:Northeast HSDA (20.60%), kootenay-Boundary HSDA (15.60%) [Figure 1]. The results indicate that people who live in the city area are most likely to maintain a healthy lifestyle.
*Future research: Does Healthy behavior automatically ensure good health - compare rural vs urban?
Figure 1: Percentage of Health Behavior index (HBI) categories by Health Region of BC
Figure 2: Frequency of Health Behavior index (HBI) by Health Region of BC
3 Methodology to explain relationship between socioeconomic characteristics and healthy behaviour
We estimate the relationship between socioeconomic/demographic characteristics and healthy behaviour using an ordered logit regression. This is because the dependent variable healthy behaviour index (HBI) that we constructed has three categories and is ordered from 1 (poor behavior) to 3 (good behavior). The form of the ordered logit model in our case is as follows. Assume Y be HBI ordinal outcome with J (three) categories. Then \({P(Y \leq j)}\) is the cumulative probability of Y less than or equal to a specific category j=1, 2 (J-1). The odds of being less than or equal to a particular category j can be defined as \(\frac{P(Y \leq j)}{P(Y > j)}\) for j=1, 2 (J-1). The log odds is also known as the logit, can be written as \(\log \left( \frac{P(Y \leq j)}{P(Y > j)}\right)\) = \(logit [P(Y \leq j)]\). The ordered logit regression model can be written as
check this: https://stats.oarc.ucla.edu/other/mult-pkg/faq/ologit/ https://bookdown.org/chua/ber642_advanced_regression/ordinal-logistic-regression.html
\[\begin{equation} logit [P(Y \leq j)] = \beta_0 + \beta_1 X_1 +.....+ \beta_k X_k \tag{1} \end{equation}\]
For simplicity of interpretation of the results, we estimate the probability of objerving outcome j obtained from the logit model. The probability of observing outcome i corresponds to the probability that the estimated linear function, plus random error of the regression, is within the range of the cutpoints estimated for the outcome:
\[\begin{equation} Pr(outcome_j=i) = Pr(c_{i-1} < \beta_1 X_{1j} +.....+ \beta_k X_{kj} + u_j \leq c_i) \tag{2} \end{equation}\]
and the associated marginal effect is
\[\begin{equation} \displaystyle \frac{\partial Pr(outcome_j=i)}{\partial X_k} =\frac{\partial Pr(c_{i-1} < \beta_1 X_{1j} +.....+ \beta_k X_{kj} + u_j \leq c_i)} {\partial X_k} \tag{3} \end{equation}\]
Where:
( Y ) is the ordinal response variable (HBI).
( j ) represents the cut-off points for the ordinal categories.
\(\beta_0\), \(\beta_1\),…., \(\beta_k\) are the coefficients estimated by the model.
\(X_1\),…., \(X_k\) are the predictor variables refers to socioeconomic variables such as sex, marital status, education, aboriginal/first nation, perceived mental health conditions, working status of each respondent, and household income.
\(u_j\) is assumed to be logistically distributed in ordered logit. In either case, we estimate the coefficients \(\beta_1\), . . . , \(\beta_k\) together with the cutpoints \(c_1\), . . . , \(c_{k-1}\), where k is the number of possible outcomes.(source: https://www.stata.com/manuals13/rologit.pdf)
The coefficients (( \(\beta\) )) in equation (1) show the ordered log-odds (logit) regression coefficients, which can be interpreted as the change in the ordered log-odds scale of the dependent or outcome variable from a one-unit increase in the explanatory variable or predictor, holding every other variable in the model are constant.
The marginal effect in equation (3) can be interpreted as one unit increases in raises the probability of being towards healthy behaviour from poor behaviour by some percentage.
4 Results and Discussion
4.1 Interpretation of regression coefficients
Results of the regression are shown in Table 1. The table reports the estimates of both coefficients and odds ratio along with standard errors (S.E) in the parentheses. The Standard interpretation of the ordered logit coefficient is that for a small change in the predictor, the response variable level is expected to change by its respective regression coefficient in the ordered log-odds scale while the other variables in the model are held constant. Below I provide interpretation of coefficients in terms of the changes in coefficient values and in terms of direction (increase or decrease) of changes.
- Marital status of the respondents
The results show that the respondents who are common law, the ordered log-odds of being in good health behaviour category is 0.212 less than who are single/never married (comparison group), given everything else is constant. The ordered log-odds of being in good health behaviour category for widow/divorce/separated is 0.190 less than who are single. These two coefficient estimates are statistically significant at the 5% level of significance as indicated by the p-value. The results also indicate that married respondents has higher log-odds than respondents with single marital status but statistically insignificant.
The results suggest that marital status with single and married respondents are likely to maintain good healthy behavior than respondents with common law and widow/divorce/separated marital status.
- Sex
The ordered logit for male in maintaining good healthy behaviou is 0.560 less than females when the other variables in the model are held constant. The results indicate that females are likely to maintain good healthy behavior than males.
- Age group
The results show that all age group respondents have a higher (positive values) log odds of being practicing good healthy behavior than the age group of 12 to 24 years old (comparison group). Except for the age group 25-34 years, all coefficient estimates are statistically significant at the 1% or 5% level of significance as indicated by the p-values. The results also indicate that as age of the respondents increases the log-odds of being practicing good healthy behavior turns out higher in magnitude compared to comparison age group.
- Education
The results indicate that the higher the level of education the higher is the chance of being practicing good healthy behavior. The magnitude of the coefficient values increases as education level increases. The coefficient estimates for secondary graduation and post secondary certificate are positive and statistically significant at the 1% level of significance as compared to lower education levels.
- Perceived Mental Health
The results indicate that the better the perceived mental condition of the respondent the better chance is to maintain good healthy behaviour and the higher is the log-odds value. The reference/comparison category here is poor perceived mental health. The coefficient estimates are positive and statistically significant. * Home/House ownership
The results indicate that people who own a house has a higher chance of maintaining good healthy behaviour than people who live in a rented house/place. The coefficient estimates statistically significant at the 1% level of significance.
- Household income
The results indicate that the higher the household income the higher is the chance that people will practice good healthy behavior compared to people with lower income (comparison group). The coefficient estimates for all income groups are positive but the estimates for income group of $80,000 or more is statistically significant at the 1% level of significance as compared to no income or lower group. The magnitudes of the log-odds coefficient estimates go up as income goes up.
- Food Security
The results indicate that the respondents who perceived them food secure/moderately food insure are more (positive log-odds coefficient values) likely to maintain good healthy behaviour compared to people suffering with severely insecure. The coefficient estimates are positive and statistically significant at the 1% level of significance.
- Perceived Social Support
The higher values of the perceived social support means the higher is the social connection and support. The results show that the log-odd estimate is positive and statistically significant at the 1% level of significance, indicating the people with higher social support/connection are more likely to maintain good healthy behaviour than people with lower social support/connection.
4.2 Significance of overall model and Threshold value
In the model.1a of Table 1, the likelihood ratio (LR) chi-square test value of 1023.00 with a p-value of 0.000 indicates that the fitted model as a whole is statistically significant, as compared to the null model with all regression coefficients equal zero.
cut1 and cut2 as shown in the bottom of table 1 are the estimated cutpoints on the latent response variable used to differentiate low HBI from fair and good HBI when values of the explanatory variables are evaluated at zero. The results show that the individuals with cut1 value of 0.101 2.75 are classified having “Poor HBI”, those who receive a latent score between 0.10 and 1.95 are classified as having “Fair HBI” and those with greater than 1.95 are classified as having “Good HBI”.
4.3 Predicted probabilities and Marginal Effects
Figures 3 to 21 show predicted probabilities and average marginal effects (AMEs)/changes in predicted probabilities for each category of HBI to changes in a number socioeconomic/demographic variables as we used in our regression models (see table 1 for the relevant variables).
Table 1. Association between socioeconomic/demographic characteristics and healthy behaviour estimates using Ordered logit regression (Dependent variable: healthy behaviour index (HBI))
##
##
## |VARIABLES |Model.1a |Model.1b |Model.2a |Model.2b |
## |:------------------------------------------------------|:---------|:----------|:---------|:----------|
## | |coeff |odds ratio |coff |odds ratio |
## |marital_status = 1, Married |0.113 |1.120 |0.112 |1.118 |
## | |(0.0728) |(0.0816) |(0.0727) |(0.0814) |
## |marital_status = 2, Common-law |-0.212** |0.809** |-0.223** |0.800** |
## | |(0.0887) |(0.0718) |(0.0886) |(0.0709) |
## |marital_status = 3, Widowed/Divorced/Separated |-0.190** |0.827** |-0.197** |0.821** |
## | |(0.0779) |(0.0644) |(0.0778) |(0.0639) |
## |sex = 1, Male |-0.560*** |0.571*** |-0.560*** |0.571*** |
## | |(0.0449) |(0.0256) |(0.0449) |(0.0256) |
## |household size |0.0279 |1.028 |0.0338 |1.034 |
## | |(0.0243) |(0.0250) |(0.0242) |(0.0250) |
## |age_group = 2, 25-34 years |0.0718 |1.074 |0.0546 |1.056 |
## | |(0.0949) |(0.102) |(0.0946) |(0.0999) |
## |age_group = 3, 35-44 years |0.226** |1.254** |0.205** |1.228** |
## | |(0.0994) |(0.125) |(0.0991) |(0.122) |
## |age_group = 4, 45-54 years |0.496*** |1.642*** |0.469*** |1.598*** |
## | |(0.104) |(0.171) |(0.104) |(0.166) |
## |age_group = 5, 55-64 years |0.537*** |1.711*** |0.503*** |1.653*** |
## | |(0.108) |(0.185) |(0.108) |(0.178) |
## |age_group = 6, 65-74 years |0.830*** |2.293*** |0.785*** |2.193*** |
## | |(0.115) |(0.263) |(0.113) |(0.249) |
## |education = 2, Secondary graduation, no post-secondary |0.614*** |1.847*** |0.612*** |1.845*** |
## | |(0.0743) |(0.137) |(0.0742) |(0.137) |
## |education = 3, Post-secondary certificate |1.069*** |2.913*** |1.068*** |2.910*** |
## | |(0.0738) |(0.215) |(0.0737) |(0.215) |
## |mental_health = 1, Fair |0.288* |1.334* |0.292* |1.340* |
## | |(0.168) |(0.225) |(0.168) |(0.226) |
## |mental_health = 2, Good |0.409*** |1.505*** |0.419*** |1.520*** |
## | |(0.158) |(0.238) |(0.158) |(0.240) |
## |mental_health = 3, Very good |0.536*** |1.709*** |0.544*** |1.723*** |
## | |(0.158) |(0.270) |(0.158) |(0.273) |
## |mental_health = 4, Excellent |0.529*** |1.697*** |0.540*** |1.716*** |
## | |(0.161) |(0.273) |(0.161) |(0.276) |
## |house_own = 1, Yes |0.154*** |1.167*** |0.159*** |1.173*** |
## | |(0.0550) |(0.0642) |(0.0550) |(0.0644) |
## |income = 2, $20,000 to $39,999 |0.0938 |1.098 |0.0895 |1.094 |
## | |(0.0946) |(0.104) |(0.0946) |(0.103) |
## |income = 3, $40,000 to $59,999 |0.121 |1.128 |0.113 |1.120 |
## | |(0.0951) |(0.107) |(0.0950) |(0.106) |
## |income = 4, $60,000 to $79,999 |0.154 |1.166 |0.146 |1.158 |
## | |(0.0990) |(0.115) |(0.0989) |(0.115) |
## |income = 5, $80,000 or more |0.257*** |1.293*** |0.240*** |1.271*** |
## | |(0.0911) |(0.118) |(0.0909) |(0.116) |
## |racial_background = 1, Non-white |0.219*** |1.244*** | | |
## | |(0.0723) |(0.0900) | | |
## |1.Immigrant#1b.time_Immigration |0.0741 |1.077 | | |
## | |(0.126) |(0.136) | | |
## |1.Immigrant#2.time_Immigration |0.100 |1.106 |0.0563 |1.058 |
## | |(0.111) |(0.123) |(0.150) |(0.159) |
## |1.Immigrant#3.time_Immigration |0.205*** |1.227*** |0.124 |1.132 |
## | |(0.0718) |(0.0881) |(0.133) |(0.150) |
## |Food_security = 0, Food secure |0.322*** |1.380*** |0.333*** |1.395*** |
## | |(0.118) |(0.163) |(0.118) |(0.165) |
## |Food_security = 1, Moderately food insecure |0.345** |1.411** |0.351** |1.421** |
## | |(0.141) |(0.199) |(0.141) |(0.200) |
## |Perceived social support |0.0171*** |1.017*** |0.0157*** |1.016*** |
## | |(0.00519) |(0.00528) |(0.00516) |(0.00525) |
## |Immigrant = 1, Landed immigrant/non-permanent resident | | |0.167 |1.181 |
## | | | |(0.122) |(0.145) |
## |Constant cut1 |0.101 |1.107 |0.0521 |1.053 |
## | |(0.281) |(0.311) |(0.281) |(0.296) |
## |Constant cut2 |1.948*** |7.016*** |1.897*** |6.669*** |
## | |(0.281) |(1.974) |(0.281) |(1.873) |
## | | | | | |
## |Observations |10,977 |10,977 |10,977 |10,977 |
## |Pseudo R-squared |0.0622 | |0.0617 | |
## |chi-square test |1023 | |1014 | |Note. Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
To summarize, the main findings and implications of the regression analysis are as follows. First, people with higher income/ food secure/better education/own house are more likely to maintain good healthy behavior than otherwise. Second, better-perceived mental health means people of that group would like to maintain good healthy behavior. Third, Males are less likely to maintain good healthy behavior than females are. Fourth, as people get older, they are likely to maintain good healthy behavior. Fifth, people with higher social support are likely to maintain good healthy behavior than people with lower social support.
5 Conclusion and Recommendation
As for the support of public health programs or initiatives that can meaningfully increase the number of healthy non-users of the B.C. public health care system over the next two to five years, we would like to the make following recommendations: i. Need to increase awareness about poor healthy behavior ii. Focus on programs that can improve the mental health condition of the people iii. Invest more in education iv. Increase income generating activities
6 Appendix
6.1 Healthy Behavior index (HBI) and Socioeconomic characteristics
6.1.1 Education level
Figure 3: Predicted probability and average marginal effects for poor healthy behaviour by education level
Figure 4: Predicted probability and average marginal effects for fair healthy behaviour by education level
Figure 5: Predicted probability and average marginal effects for good healthy behaviour by education level
6.1.2 Perceived Mental Health
Figure 6: Predicted probability and average marginal effects for poor healthy behaviour by perceived mental health
Figure 7: Predicted probability and average marginal effects for fair healthy behaviour by perceived mental health
Figure 8: Predicted probability and average marginal effects for good healthy behaviour by perceived mental health
6.1.3 Perceived social support
Figure 9: Predicted probability and average marginal effects for poor healthy behaviour by perceived social support
Figure 10: Predicted probability and average marginal effects for fair healthy behaviour by perceived social support
Figure 11: Predicted probability and average marginal effects for good healthy behaviour by perceived social support
6.1.4 Age Group
Figure 12: Predicted probability and average marginal effects for poor healthy behaviour by age group
Figure 13: Predicted probability and average marginal effects for fair healthy behaviour by age group
Figure 14: Predicted probability and average marginal effects for good healthy behaviour by age group
6.1.5 Racial Background
Figure 15: Predicted probability and average marginal effects for healthy behaviour by racial background
6.1.6 Food Security
Figure 16: Predicted probability and average marginal effects for poor healthy behaviour by food security status
Figure 17: Predicted probability and average marginal effects for fair healthy behaviour by food security status
Figure 18: Predicted probability and average marginal effects for good healthy behaviour by food security status
6.1.7 Income
Figure 19: Predicted probability and average marginal effects for poor healthy behaviour by income
Figure 20: Predicted probability and average marginal effects for fair healthy behaviour by income
Figure 21: Predicted probability and average marginal effects for good healthy behaviour by income