Hi I am carrying out a study investigating the association between age and password strategies among online users. I have collected my data, and inputted it on SPSS, however, I am not sure which SPSS test would be best suited to analyse my data. Any suggestions would be greatly appreciated.
Thanks
Assuming your strategies are nominal / qualitatively differnce, you should conduct some sort of chi-square (nominal x interval/ratio) test for significant associations.
Hope this helps,
Matt
The answer is X2 test according to the table in attachment ( still have no reference for this table).
Your predictor variable Age is on an interval scale, your dependent variable password Strategies is nominal or maybe ordinal (if you consider one strategy better than the other). Maybe you could calculate the Spearman rank-order correlation coefficient if the latter is the case. Otherwise, it depends upon how much categories of strategies you have. If only 2: the point-beserial correlations, if more than 2: some measure of association. I'am not quite sure which one, but maybe this can help: http://uregina.ca/~gingrich/ch11a.pdf
Assuming that the strategies are nominal, and that each respondent has been assigned to one of the categories, you could use the Kruskal-Wallis test (Analyse -Nonparametric tests - Legacy dialogs - K independent samples) with Age as the Test variable and the Strategies as the Grouping variable. This would test whether the median ages of people who use Strategy A, Strategy B, Strategy C ... etc. are significantly different. If the ages are also categorised (18-20, 21-23 etc) then a chi-square test should do the trick (Analyse - Descriptive statistics - Crosstabs) choose Age for the rows and Strategies for the columns, tick Clustered bar charts, and under Statistics choose Chi-Square. For more precise instructions, you will have to provide more information about your variables.
Thanks for your answers. Yes the ages are grouped in categoreis (decades) and the startegies used are upper + lower case letters, numerical values, special characters, same passwords for multiple accounts, and influencers of the password ex special event. The possible responses on the questionnaire were yes/no... So I guess a chi square could do the job?
Yes, a chi-square test may be done separately for each strategy vs age, but that does not give you an overall view for all the strategies jointly. An informative graphic display could be a "clustered bar chart for separate variables" (the clusters being the age groups and the separate variables being the strategies). Correspondence Analysis seems to me to be the ultimate answer, if you can manage that. CHAID might also be a solution if you have a very large sample (at least several hundred). You might obtain some insightful answers - good luck.
If you can, use the Age independent variable as an interval scale. A lot of info is lost by categorizing Age. Then run a simple logistic regression, as illustrated in:
https://www.southampton.ac.uk/passs/neighbourhood_policing_awareness/multivariate_analysis/simple_logistic_regression_categorical.page
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS