Study design report & analytical planning and data analysis Essay Example for Free

Study design report analytical planning and data analysis Essay Part 2 STUDY TWO (Open file 1. 2. sav dataset) RESEARCH QUESTIONS: Factors related to birth weight outcomes specifically for this study: 1. Are there differences between three birthweight groups in maternal lead level? 2. Is maternal age at first childbirth associated with birth weight? BACKGROUND: It has been suggested by previous research findings that maternal blood-lead levels as an environmental factor is an apparent predictor of low birth weight; another possible explanatory factor of low birth weight relates to the maternal age of the mother. This study aimed to explore relationships between maternal blood lead levels, and maternal age (at first childbirth) with any observed differences in birth weight as a pregnancy outcome. METHODS: Life style information, maternal blood lead levels measured by micrograms per decilitre ( µg/dl), age of mother (years) and infant birth weight (grams) was collected from 250 consecutive (first-time) mother-infant pairs from hospital delivery departments in four Brisbane hospitals. Pregnancy outcomes (weight) were verified by consulting medical records. Lead was determined by electro thermal atomic absorption spectrometry (Whole Blood). Birth weight was recoded into three classes: Low birth weight (2500g); Lower end of normal (2500-3000g) and Normal (3000g). Additional notes: Lead exposure adversely affects the cognitive development and behaviour of young children . Regarding ‘LEAD EXPOSURE IN PREGNANCY’, according to the Centre for Disease Control (CDC), the acceptable blood lead level is 10 ug/dL. US Department of Health and Human Services, Agency for Toxic Substances and Disease Registry, 1999. . Write an unambiguous analytical plan to address the research question/s in the box above. Please note that the data required for this question are located on file1. 2. 1. What are the variables involved in answering the research question/s, e. g. independent dependant variables; In the first research question, the independent variable is the maternal lead level, while the dependent variable is birth weight of the infants. In the second research question, the independent variable is the age of the mother at first childbirth and the dependent variable is birth weight. 2. What is their level of measurement (type); All variables considered in this study birth weight of the infants, age of the mother at first childbirth and maternal lead level are ratio variables. Ratio variables are those which are measured on a scale where the distance from one point to another means the same things wherever the measurement is made on the scale. This scale is also characterized to have a true zero point representing the absence of what is being measured (Statistical Glossary, 2007) 3. Write-up the scientific hypotheses that you want to test to address the research question/s; The following hypotheses, stated in the null form, were tested at a 0. 05 level of significance : a. There are no significant differences in the means of the maternal lead levels when the infants are grouped according to birthweight. b. There is no significant association between maternal age at first childbirth and the birthweight of infants. 4. What tables (numerical) and/or graphs would you use to summarise the associations once the data are collected and analysed? Provide ‘dummy’ tables and/or graphs with appropriate labels; The following tables summarize the associations revealed from the test of the first hypothesis. Table 1. Significant Differences in the Maternal Lead Levels When the Infants are Grouped According to Birthweight based on One-Way Analysis of Variance (? = 0. 05). Source of Variation Sum of Squares Degrees of freedom (df) Mean Square F-value p-value Remarks Between groups 42. 492 2 21. 246 4. 350 0. 014 Significant difference in at least one pair of means Within groups 1206. 255 247 4. 884 Total 1248. 747 249 Table 2. Mean differences and p-values in the Maternal Lead Levels Among the Three Groups of Infants using Bonferroni Post-hoc Analysis Infant Groups Mean differences in Maternal Lead Levels (I – J) p-value Remarks I J Low birthweight Lower end of normal birthweight 0. 5952 0. 645 No significant difference Normal birthweight -0. 3895 1. 000 No significant difference. Lower end of normal birthweight Low birthweight 0. 5952 0. 645 No significant difference Normal birthweight -0. 9848 0. 011 Maternal lead level Significantly Higher in normal weight infants Normal birthweight Low birthweight 0. 3895 1. 000 No significant difference Lower end of normal birthweight -0. 9848 0. 011 Maternal lead level Significantly Higher in normal weight infants The following tables summarize the associations revealed from the test of the second hypothesis. Table 3. Relationship Between Maternal Age at Childbirth and Birthweight of Infants (? = 0. 011) Variables Correlation. Coefficient p-value Remarks Maternal age of mothers at first childbith (independent) vs. Birthweight of infants (dependent) 0. 18 0. 004 There is significant slight correlation between maternal age of mothers at first childbirth and the birthweight of infants. 1 Hypothesis was tested using 0. 01 level of significance. Results of the test of hypothesis that â€Å"There is no significant association between maternal age at first childbirth and the birthweight of infants† is shown in Table 3. 5. Provide a statement of the statistical test(s) that will be appropriate to test the hypotheses; To test the hypothesis that â€Å"There are no significant differences in the means of the maternal lead levels when the infants are grouped according to birthweight†, the best statistical treatment to use is one way Analysis of Variance ANOVA at a 0. 05 level of significance. When the computed p-value, which for this study is 0. 014 is less than the level of significance, which is 0. 05, there is significant difference in at least one pair of means. To evaluate, which of the means differed, a post-hoc analysis is conducted. Since there are only three groups, the conservative Bonferroni multiple comparisons test is applied. P-values from the post hoc analysis indicate which of the pairs of mean differed. To determine which of the two means that differed is actually higher, either the descriptives which shows the means of the groups or the mean difference in post-hoc analysis results may be used. To test the hypothesis that â€Å"There is no significant association between maternal age at first childbirth and the birthweight of infants† correlation analysis is employed at a 0. 01 level of significance. The resulting value of the correlation coefficient was interpreted using the following table : Table 4. Interpretation of Pearson’s Coefficient of Correlation (Monzon-Ybanez, 1997) Coefficient Range Interpretation 0. 00  ± 0. 20  ± 0. 20  ± 0. 40  ± 0. 40  ± 0. 70  ± 0. 70  ± 0. 90  ± 0. 90  ± 1. 00 Slight correlation; almost negligible relationship Low correlation; small relationship Moderate correlation; relationship substantial High correlation; marked relationship Very high correlation; Very dependable relationship Correlation is significant if the p-value generated is less than the specified level of significance. 6. Provide a list of assumptions that will need to be met to apply the test(s) validly; To validly apply ANOVA, the following assumptions must be satisfied : the distribution of the data to be analysed should be normal and there should be homogeneity of variance. Normality may be simply checked using Q-Q plots or with statistical tests like Kolmogorov-Smirnov or Shapiro Wilk. Homogeneity of variance is verified using Levene’s test (Becker, 1999). In a correlation analysis, relationship between the independent and dependent variables are supposed to be concurrent, or in other words, both variables are in the same time frame (Jensen, 2005). 7. Provide a statement of the levels of significance, that you will use to test each hypothesis; The hypothesis that â€Å"There are no significant differences in the means of the maternal lead levels when the infants are grouped according to birthweight† was tested at a level of significance (? ) of 0. 05. The hypothesis that â€Å"There is no significant association between maternal age at first childbirth and the birthweight of infants† was tested at a level of significance (? ) of 0. 01. 8. Finally, provide a sentence or two that you would present to the researchers to explain the results at both a descriptive (univariate) and inference (bivariate) level. Results of a one-way Analysis of Variance revealed a significant difference in the maternal lead levels among the three groups of infants grouped according to birth weight, with an F-value of 4. 350 and a p-value of 0. 014. Post hoc analysis via Bonferroni multiple comparisons test suggested that the mean of maternal lead levels of the group of infants with normal birthweight (4. 4052  ± 2. 37333) are significantly higher (p=0. 011) than the mean of maternal lead levels of the group of infants at the lower end of the normal birthweight (3. 4214  ± 1. 55170). No significant differences were noted on the other pairs of means. On the other hand, a correlation analysis performed at 0. 01 level of significance implied significant slight correlation (Pearson r = 0. 18, p-value=0. 004) between maternal age at first childbirth (32. 11  ± 5. 274) and birthweight of infants (3204. 47  ±620. 986). References Becker, L. (1999, July 7). Explore: Assumption testing for ANOVA.Retrieved April 11, 2008, from The University of Colorado at Colorado Springs: http://web. uccs. edu/ lbecker/spss80/explore2. htm. Jensen, A. (2005, August 10). Correlation Analysis. Retrieved April 10, 2008, from California State University, Sacramento: http://www. csus. edu/indiv/j/jensena/ mgmt105/correl01. htm. Monzon-Ybanez, L. (1997). Basic Statistics. Quezon City, Philippines : Phoenix Press, Inc. Statistical Glossary. (2007). Retrieved April 10, 2008, from http://www. statistics. com/resources/glossary/r/ratioscale. php. Statistical Package for the Social Sciences. (2006). Version 11. 0. [CD-ROM] Chicago, Illinois.