Analysis Of Psychological Survey Data: Answering Four Research Questions

Association between Low Pre-flood Psychological Score and Living Alone

In this paper, I sought to analyse data on psychological survey. There are four research questions that this study sought to answer. The four research questions are;

1. Is there an association between having a low (below 15) pre-flood psychological score and living alone? If so, what this the nature of the association?
2. Are age, social support score and family functioning score predictors of the pre-flood psychological score? Which of these three variables explains most of the variation in pre-flood psychological score? How does the inclusion of place of residence as a predictor change the fitted model? Using the minimum model, which contains only the significant variables, what is the predicted pre-flood psychological score for a 35-year old male living in a rural area with a social support score of 40 and a family functioning score of 22?
3. Is there a difference in the post-flood psychological score between men according to the level of impact of the 2011 flood? If there is a difference, which groups are different?
4. Is the mean change in psychological score between the pre and post-flood survey the same for men who experienced no or limited flood impact compared to men who experienced moderate/major flood impact?

 Results

Research question 1:

To answer this research question, I had to apply a Chi-Square test of association. The pre-flood psychological score was given as a numerical variable and I had to recode where scores below 15 were recoded as low and scores above 15 were recoded as high. I ended up with two categorical variables making Chi-Square test an ideal test to test for the association.

Brief overview of the statistical methods you used

For this analysis, I used Chi-Square test of association. Also called Pearson’s chi-square test or the chi-square test of independence, is used to discover if there is a relationship between two categorical variables. The null hypothesis for the test is that there is no association between the variables.

Using SPSS I had to run the test and the results are displayed below;

Chi-Square Tests
	Value	df	Asymp. Sig. (2-sided)
Pearson Chi-Square	.100a	1	.752
Continuity Correctionb	.000	1	1.000
Likelihood Ratio	.099	1	.752
Linear-by-Linear Association	.099	1	.753
N of Valid Cases	157
a. 2 cells (50.0%) have expected count less than 5. The minimum expected count is 3.57.
b. Computed only for a 2×2 table

The p-value for the Pearson Chi-Square test is 0.752 (a value greater than 5% level of significance). We therefore fail to reject the null hypothesis and conclude that there is no significant association between having a low (below 15) pre-flood psychological score and living alone.

In this section, I aimed at finding out whether age, social support score and family function score predict the pre-flood psychological score.

Brief overview of the statistical methods you used

For this analysis, I used multiple regression analysis. Regression analysis refers to a set of statistical processes that are used to estimate the relationships among variables. The test includes many techniques for modelling and analysing several variables, when the focus is on the relationship between a dependent variable (also known as response variable) and one or more independent variables (explanatory variables).

Regression Coefficients-model 1
Model	Unstandardized Coefficients	Standardized Coefficients	t	Sig.
B	Std. Error	Beta
(Constant)	14.866	1.261		11.794	.000
Age in years	-.018	.015	-.087	-1.210	.228
Social support scale (pre flood)	.070	.018	.280	3.766	.000
Family functioning scale (pre flood)	-.073	.036	-.149	-2.016	.045
R-Squared = 0.140 F(3, 170) = 9.221, p-value = 0.000

As can be seen in the regression analysis results table above, the value of R-Squared is 0.140; this implies that 14% of the variation in the dependent variable is explained by the three explanatory variables. It can also be seen that two of the three variables are significant in the model. The two significant variables are Family functioning scale (pre flood) and Social support scale (pre flood).

I also sought to find out which of the three variables explains most of the variation in pre-flood psychological score. From the same regression results, it was found that the variable that explains most of the variation in pre-flood psychological score is the Social support scale (pre flood) since it had a larger value for the standardized coefficient.

Next I added place of residence as a predictor into the model to see how it affects the fitted model.

Regression Coefficients-model 2
Model	Unstandardized Coefficients	Standardized Coefficients	t	Sig.
B	Std. Error	Beta
(Constant)	15.077	1.270		11.873	.000
Age in years	-.016	.015	-.079	-1.090	.277
Social support scale (pre flood)	.074	.019	.299	3.858	.000
Family functioning scale (pre flood)	-.064	.038	-.132	-1.713	.089
Living alone?	-.587	.594	-.075	-.990	.324
R-Squared = 0.145 F(4, 168) = 7.130, p-value = 0.000

By adding the variable place of residence into the model, the value of R-squared changed to 0.145; implying that 14.5% of the variation in the dependent variable is explained by the four explanatory variables in the model; this shows a very small change. Also, the added variable (place of residence), was found to be insignificant in the model. However, it should be noted that addition of this variable renders the variable (Family functioning scale) insignificant in the model.

Predictors of Pre-flood Psychological Score

Using the minimum model, which contains only the significant variables, the final regression model is given as;

Where,

 is the dependent variable (Psychological domain (pre flood)) while  is the significant predictor variables which is the Social support scale (pre flood).

So the predicted pre-flood psychological score for a 35-year old male living in a rural area with a social support score of 40 and a family functioning score of 22 is given as follows;

Hence the predicted pre-flood psychological score for the given values is 44.677.

In this section, I sought to test whether there a difference in the post-flood psychological score between men according to the level of impact of the 2011 flood.

Brief overview of the statistical methods you used

For this analysis, I used analysis of variance (ANOVA) test. ANOVA refers to a statistical model that is used to analyse the differences among group means and their associated procedures for variables with more than two factors.  This research question involves one dependent variable and an independent variable with three factors hence ANOVA test was ideal for use.

ANOVA
Psychological domain (post flood)
	Sum of Squares	df	Mean Square
Between Groups	52.820	2	26.410
Within Groups	407.796	113	3.609
Total	460.616	115

The p-value as can be seen from the above table is 0.001 (a value less than 5% level of significance), we therefore reject the null hypothesis and conclude that there are differences in the mean post-flood psychological score between men according to the level of impact of the 2011 flood.

I conducted a post-hoc analysis using LSD to Bonferroni where we found out that the average post-flood psychological scores was significantly higher in the no impact condition (M = 15.69, SD = 2.01) than in the moderate/major impact condition (M = 14.66, SD = 2.00), p = .001. There was however no significant difference in the mean post-flood psychological scores between the other groups.

This section sought to answer the last research question. The question I sought to answer was whether the mean change in psychological score between the pre and post-flood survey the same for men who experienced no or limited flood impact compared to men who experienced moderate/major flood impact. I used an independent t-test to answer this.

Brief overview of the statistical methods you used

For this analysis, I used an independent samples t-test. Also known as student’s t-test, the test refers to inferential statistical test that determines whether there is a statistically significant difference between the means in two unrelated groups. This research question involves one dependent variable and an independent variable with two unrelated (independent) factors hence independent t-test test was ideal for use.

Mean change in psychological score between the pre and post-flood

Group Statistics
Impact of the floods for you in terms of the property you were living in	N	Mean	Std. Deviation	Std. Error Mean
No or limited flood impact	63	-.3993	2.39085	.30122
Moderate/major flood impact	52	.7770	1.57561	.21850

I performed an independent samples t-test to compare the mean change in psychological score between men who experienced moderate/major flood impact and those who experienced no or limited flood impact. Results showed that the mean change in psychological score between the pre and post-flood survey was significantly different for men who experienced no or limited flood impact compared to men who experienced moderate/major flood impact (p-value = 0.03). Among the men who experienced no or limited flood impact, the mean change in in psychological score between the pre and post-flood survey was -0.3993 while the mean change in in psychological score between the pre and post-flood survey for those who experienced moderate/major flood impact was 0.7770.

Conclusion

This study sought to investigate four research questions. Four different statistical tests were employed to analyse the research questions. For the first research question, I used Chi-Square test of association where I found out that there is no significant association between having a low (below 15) pre-flood psychological score and living alone. For the second research question, I multiple regression analysis. The third research question applied ANOVA test while the last part employed independent t-test.

References

Cornwell, E. Y. & Waite, L. J., 2009. Social disconnectedness, perceived isolation, and health among older adults. Journal of Health and Social Behavior, 50(1), pp. 31-48.

Cox, D. R., 2006. Principles of statistical inference.

Dean, A., Kolody, B., Wood, P. & Matt, G. E., 1992. The influence of living alone on depression in elderly persons. Journal of Aging and Health, 4(1), p. 3–18.

Fay, M. P. & Proschan, M. A., 2010. Wilcoxon–Mann–Whitney or t-test? On assumptions for hypothesis tests and multiple interpretations of decision rules. Volume 4, p. 1–39.

Gee, E. M., 2000. Living arrangements and quality of life among Chinese Canadian elders. Social Indicators Research, 51(3), p. 309–329.

Greenwood, P. E. & Nikulin, M. S., 1996. A guide to chi-squared testing.

Mellor, D. et al., 2008. Need for belonging, relationship satisfaction, loneliness, and life satisfaction. Personality and Individual Differences, 45(3), p. 213–218.

Mui, A. C., 1996. Depression among elderly Chinese immigrants: an exploratory study. 41(6), pp. 633-645.

Perlman, D. & Peplau, L. A., 21-56. Towards a social psychology of loneliness, in Personal Relationships in Disorder.

Yang, K. & Victor, C. R., 2008. The prevalence of and risk factors for loneliness among older people in China. Ageing and Society, 28(3), p. 305–327.

Zimmerman, D. W., 2007. A Note on Interpretation of the Paired-Samples t Test. Journal of Educational and Behavioral Statistics, 22(3), p. 349–360.