Only 39 cases needed to be top coded to 20 servings. The individual fruit and vegetable items for all 39 cases were examined carefully to make sure that they consisted of believable number of servings, albeit large numbers reported, and all 39 were accepted to remain part of the data file for analysis.The General RDD sample in File-1 was weighted to 1990 Census data for California adults using race, age group, and gender dimensions to emulate the standardized weighting of the CDPS surveys—all of which were weighted to the 1990 Census. Weighting within each of the three race/ethnic groups within each month’s sample, the ANOVA model was used to examine the year-month interaction term and variation among months for White, African American, and Latino groups. This last analysis produced the conclusions for this study with regard to seasonality in fruit and vegetable intake.As part of a nested design, three brief questions were asked to a randomized subset of survey respondents to see how well these might perform compared to the longer, more complex set of CDPS recall questions. This short form of the intake questions are referred to in this study as the “SF3.” For this analysis of the SF3,25 liter pot plastic approximately 57% of the File-1 sample was used due to prior random allocation.
These respondents were administered the SF3 questions ahead of the usual CDPS dietary intake questions. The SF3 captured the following: the number of servings of fruit not including 100% fruit juices ; the number of servings of 100% fruit juice ; the number of servings of vegetables ; and, the total number of servings of fruit and vegetables . Analyses were performed comparing each respondent’s SF3 answers and the respondent’s corresponding number of servings calculated from the CDPS intake responses. The initial analytical step examined Pearson correlations to see if they were positive and relatively high for this type of measurement . If true, results were then examined using a paired t-test to see if there was any consistent bias in the relationship between the SF3 and the CDPS questions. Paired t-tests were performed using a two-tailed significance level of 0.05. Looking at the 6,650 cases in File-1 and the 4,031 low-income cases in File-2 , the year-month interaction is significant for the total servings of fruit and vegetables for File-1 and File-2 . In File-1 this is not the case for servings of fruit . The observed pattern of servings of fruit intake seen in Exhibit 7 shows the month of July with the highest intake at 2.5 servings and November with the lowest intake at 2.0 servings.
Since the year-month interaction is significant for servings of vegetables , a direct interpretation of the monthly variation is not possible. In File-2 there is significant year-month interaction for both servings of fruit and vegetables . These findings indicate that the patterns of intake across the 12 months is different for each of the two years of data collection , and thus it is not feasible to combine the monthly samples from these two years to interpret variation among the months using all the cases in each file. The General RDD component of File-1 , a sample that does not have any race/ethnic over-samples included and thus very much like the CDPS general population sample, has no significant year-month interaction for the total servings of fruit and vegetables. This is also true, as might be expected, for the servings of fruit and the servings of vegetables, separately. Thus, it is the Targeted RDD component of File-1 consisting of the “all-incomes” over-samples of Latinos and African Americans where significant interaction is found for total fruit and vegetables , total fruit , and total vegetables . Given the findings for the General RDD component, the pattern of intake across the 12 months is not significantly different for each of the two years of data collection, and therefore, it is possible to combine the monthly samples from these two years to interpret variation among the months. However, the significant year-month interaction for the Targeted RDD component precludes a direct interpretation for the “all-incomes” oversamples of Latinos and African Americans.
The results for variation among months in the General RDD sample suggest that there is no significant difference among months for the total servings of fruit and vegetables. However, the p-value is exactly 0.05 and is right on the boarder of being statistically significant. Variation among months for the total servings of fruit and for the total servings of vegetables are each not significant. Exhibit 8 shows the average number of servings per month for each of the twelve months of the year for total fruit and vegetables, for fruit, and for vegetables in the General RDD sample after combining the data for Year-1 and Year-2. These results and monthly patterns in the General RDD sample are virtually unchanged even after controlling for race, ethnicity, and gender . Since the sample is designed to examine fruit and vegetable intake for Whites, African Americans, and Latinos, results are generated for each in both File-1 and File-2 . Exhibit 9 shows the results of the ANOVA Ftests, looking first if any year-month interaction exits, and if none exists, looking for significant variation among months as an indication of seasonal effects. No seasonality is found for Whites when looking at all income levels for total fruit and vegetables, or for fruit and vegetables separately. This is true for low-income Whites for servings of fruit and for servings of vegetables, but significant year-month interaction for the total number of servings of fruit and vegetables precludes any direct interpretation of monthly variation. No seasonality is found for African Americans when looking at all income levels for the total of fruit and vegetables or for fruit and vegetables separately . This finding is also true for low-income African American, with the exception of significant monthly variation for servings of fruit . It is not possible to interpret directly the results either for Latinos of all-incomes or for File-2 low-income Latinos, since there is significant year-month interaction for the total of fruit and vegetables as well as for fruit and vegetables separately.
An examination of the monthly pattern for Years 1 and 2 separately shows each year to be very different; however, the pattern within each year is similar for the Latinos in File-1 and in File-2. The very different patterns for each of the years are the cause for the significant year-month interaction. To demonstrate this, the pattern for the total number of servings of fruit and vegetables for Year 1 and Year 2 and for File-1 and File-2 Latinos is shown in Exhibit 10. Observe that March and April, June and July, and September, October, and November appear different between the two years. There is no explanation for these different patterns. When looking only at the low-income cases found in File-1, the picture changes slightly . For low-income Whites, no year-month interaction is now detected as it was shown when looking at all income Whites. This is also true for servings of fruit for lowincome African Americans. or African Americans , they may be harder to detect given the smaller sample sizes of these low-income groups. A year-month interaction for total fruit and vegetables is also not detected. For both Whites and for African Americans , no variation among months is found for the total servings of fruit and vegetables or for fruit and vegetables separately. For Latinos,25 litre plant pot with a large sample size of 1,040, year-month interaction for total fruit and vegetables is not detected and no significant variation among months is found for the total servings of fruit and vegetables . However, significant variation among months is found for servings of fruit. Exhibit 12 shows the observed highest months of fruit intake to beFebruary and June with the lowest observed months to be March, May, and November. It is not possible to easily determine variation among months for the servings of vegetables due to significant year-month interaction . To gain additional insight into the low-income Latino patterns, the data are examined by the two levels of Latino acculturation: high and low. Exhibit 13 shows that there is no significant variation among months for Latinos with high acculturation whether the data are examined for all incomes , or for those with low-income. Among Latinos with low acculturation, however, it is not possible to easily interpret the data for the all-incomes groups due to significant year-month interaction . For the low-income, low-acculturation Latinos, there is no significant variation among months for total fruit and vegetables . For servings of fruit, both the all-incomes and low-income Latinos show significant variation among months if they are low acculturated.
Exhibit 14 shows the observed average servings of fruit per month for low-acculturation Latinos in the all-incomes group and the low-income group . The months of February and June show the highest averages, while March, May, and November show the lowest averages of fruit intake. These observations for individual months for both lowincome acculturation groups are the same as seen in Exhibit 11 for all low-income Latinos regardless of acculturation level. The CDPS data for the eight bi-annual surveys conducted since 1989 are each weighted to the 1990 Census for California. The cases are weighted for gender, age group, and race dimensions. Since the General RDD sample in this study is similar to the CDPS general population sample, i.e., it includes the entire adult population , this study’s General RDD sample data have been weighted using the identical weighting approach used in the CDPS. In Exhibit 15, using the weighted General RDD sample data , no two groups are statistically different from each other. In an examination of the variation among months, Exhibit 18 shows the results for the overall population of all adults as well as the results for Whites, African Americans, and Latinos. Among all adults, variation among months is found to be significant for the total servings of fruit and vegetables consumed. However, this variation is not significant either for servings of fruit alone or for servings of vegetables alone. Exhibit 19 shows the monthly averages across the year. It is found that the average number of servings for total fruit and vegetables in the month of July is significantly higher than the number of servings in January , a difference detectable due to the relatively large sample size. However, July is found to be not significantly higher than any other month in the year. There is no significant variation among months found either for servings of fruit or for servings of vegetables for the adult population. For Whites, there are no significant differences in the variation of number of servings among months for the total of fruit and vegetables or for either fruit or vegetables alone. The observed monthly patterns for Whites are shown in Exhibit 20. African Americans show significant variation among months for the number of servings of total fruit and vegetables and for servings of vegetables . Despite the pattern for servings of fruit being very close to that of vegetables, no significant difference is detected for servings of fruit . Exhibit 21 shows the wide variation observed among the months of the year. Even with the smaller African American sample, significant differences between individual months for total servings of fruit and vegetables are seen, with the month of December higher than June and January and none of the other months being different from each other. Although the variation among months for servings of vegetables is found to be significant, the follow-up procedure could not detect any significant differences between individual months. The small AfricanAmerican sample, 177 cases distributed over 12 months , needs to be considered when interpreting these findings. Latinos show no significant differences in the variation of number of servings among months for the total of fruit and vegetables or for either fruit or vegetables alone. The observed monthly patterns for Latinos are shown in Exhibit 22. The large sample sizes used in this study reveal that patterns across the months of the year can differ from year to year, as indicated by the significant year-month interactions found for the overall samples of both File-1 and File-2.