Kaylie Analysis
Notes:
1: I am generally going to include more plots than you need. You can pick out the ones you want, but keep it consistent with the methods you use.
2: Test outputs are generally not included (i.e. copy pasted) into your results section. It’s better to pick out the values you need and report them in-line with the text. I report the full information here for completeness.
Findings for Question 1
Jr.Metacognititve Awareness Inventory
Descriptive statistics for control group of scored data
## MA pretest MR pretest MA posttest MR posttest
## Min. :21.0 Min. :18.00 Min. :23.00 Min. :16.0
## 1st Qu.:30.0 1st Qu.:26.00 1st Qu.:31.50 1st Qu.:25.5
## Median :31.0 Median :30.00 Median :35.00 Median :29.0
## Mean :32.0 Mean :29.41 Mean :33.74 Mean :29.0
## 3rd Qu.:35.5 3rd Qu.:33.50 3rd Qu.:36.00 3rd Qu.:32.5
## Max. :40.0 Max. :39.00 Max. :41.00 Max. :38.0Descriptive statistics for treatment group of scored data
## MA pretest MR pretest MA posttest MR posttest
## Min. :21.0 Min. :18.00 Min. :23.00 Min. :16.0
## 1st Qu.:30.0 1st Qu.:26.00 1st Qu.:31.50 1st Qu.:25.5
## Median :31.0 Median :30.00 Median :35.00 Median :29.0
## Mean :32.0 Mean :29.41 Mean :33.74 Mean :29.0
## 3rd Qu.:35.5 3rd Qu.:33.50 3rd Qu.:36.00 3rd Qu.:32.5
## Max. :40.0 Max. :39.00 Max. :41.00 Max. :38.0Histograms for the scored values of the treatment group, pre-test and post-test
Histograms for the scored values of the control group, pre-test and post-test
Boxplots for the control and treatment groups on scores of Metacognitive Regulation
Boxplots for the control and treatment groups on scores of Metacognitive Awareness
Since the control group took the same test twice, we can look at the correlation between those tests as a measure of test-retest reliability. A strong correlation indicates the test is consistent with what is measuring. A significant correlation between pre and post test groups in the control condition for both Metacognitive Awareness scores (r = .79,p < .05) and Metacognitive Regulation scores (r = .78,p < .05) was found.
##
## Pearson's product-moment correlation
##
## data: JrMaiK$MA.Sc and JrMaiK$postMA
## t = 6.3889, df = 25, p-value = 1.091e-06
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## 0.5815220 0.8986023
## sample estimates:
## cor
## 0.7875075##
## Pearson's product-moment correlation
##
## data: JrMaiK$MR.Sc and JrMaiK$postMR
## t = 6.1339, df = 25, p-value = 2.061e-06
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## 0.5600654 0.8922961
## sample estimates:
## cor
## 0.7751105A scatterplot shows the relationship between the pre and post tests for each condition.
Chi Square test of independence for Control group
The number of students whose post-test answers matched or differered from the post test is displayed in the figure above. Each line represents a pretest option: Never, Seldom, Sometimes, Often, and Always. The x axis represents the number of students who answered in the post-test given they had answered with the pre-test level matching the line. For example, the Always line starts at 0, and climbs to 38. This means that there were 0 student that selected Always on the pre-test who also selected Never on the post-test. This also means that there were 38 students who answered Always in both the pre and the post test. In fact, for every condition except for Seldom, the mode of the post-test responses matched that of the pre-test. Those that reported Seldom in the pretest have a mode of Sometimes in the post test.
RepeatedMeasureTable## MAQ_KPostTable
## MAQ_KPreTable Never Seldom Sometimes Often Always
## Never 4 4 3 3 2
## Seldom 1 6 14 8 2
## Sometimes 3 11 21 18 12
## Often 0 2 16 30 16
## Always 0 1 5 23 38##
## Pearson's Chi-squared test
##
## data: contingencyTable
## X-squared = 6.1328, df = 4, p-value = 0.1894Chi Square test of independence for Treatment group
The treatment condition appears to have a general trend in the positive direction. Most of the students that reported Never in the pre-test codition report Seldom in the post test condition. Every other pretest condition, with the exception of the Always condition, has a mode in the post test condition that is one higher than would be expected.
RepeatedMeasureTable## MAQ_KPostTable
## MAQ_KPreTable Never Seldom Sometimes Often Always
## Never 1 12 8 5 9
## Seldom 2 6 8 4 4
## Sometimes 3 9 16 20 11
## Often 3 4 10 16 17
## Always 1 2 13 21 29#construct contingency table
contingencyTable <- rbind(table(MAQ_KPreTable),table(MAQ_KPostTable))
rownames(contingencyTable) <- c("Pretest","Posttest")
#run chi square test of independence (does not meet assumptions!)
chisq.test(contingencyTable)##
## Pearson's Chi-squared test
##
## data: contingencyTable
## X-squared = 17.775, df = 4, p-value = 0.001366Since the Chi-Square test of independence found no signficance in the control condition (\(\chi^2\) = 6.133, df = 4, P = .189), then we can assume there are no statistically significant differences between the pre-test and post-test. Differences between the pre-test and post-test of the Treatment condition were found to be statistically significant (\(\chi^2\) = 17.775, df = 4, P = .001).
Disclaimer: Keep in mind that this is NOT the correct test to use in this situation. If you do not score the data and run mixed ANOVA then you can still run non-parametric tests, but this test is still not appropriate for this data set. Our two conditions (pre and post test) are within group and are ordinal scaled, the Chi Square test of independence tests Between groups and reduces that ordinality to categorical. It will most likely miss differences that are present in the data and your interpretation will suffer from it. A more appropriate non-paramteric test is the Generalized McNemar (Stuart-Maxwell Test), the Wilcoxon Signed-Rank Test, or an Ordinal Logistic Regression model.