Bipolar disorder individuals often make highly lethal suicide attempts therefore early identification of high-lethality suicide attempters is critical. In this study, BD patients were classified as high- or low- lethality attempters. High lethality attempts required inpatient medical treatment. A logistics model and permutation analyses examined the correlations between lethality, number and order of attempts. Researchers found that multiple attempters showed no pattern in increasing or decreasing their lethality attempts later on, but individuals who were disposed to early high-lethality attempts indicated this pattern of recurrence of high-lethality attempts later. In other words, risk for high-lethality suicide attempts was greatest amongst those who were more resolved to die and who had made a greater number of suicide attempts. However, attempters of low-lethality attempts does not indicate that these individuals would not try high-lethality attempts later on. Are procedures parametric or non-parametric? Why

Parametric tests require a normal distribution of the studied variable, the drawing of independent samples, and at least one variable being studied to be interval. They often also require a larger sample size than nonparametric tests do.

In this case, the procedures were largely parametric although non-parametric tests were included to deal with nonparametric variables that were ordinal, categorical, and continuous. The study, as a whole, could be classified as parametric since the sample size was large (n=146), and interval data was used. The fact that the mean and SD were employed also indicates parametric perimeters. The two groups were well matched indicating no outliers. The t-test used for comparing the two groups is parametric although the chi-squared test is not (and indeed was used for categorical variables) as is the Wilcoxon matched pair variable used, as in this case, for continuous variables. Spearman's test used for ordinal data and existent here too, is another nonparametric procedure.

What types of statistic tests were used? What do the results of these tests tell you about the data?

The statistics tests used were the t-test to make comparisons between the two groups in terms of clinical, diagnostic, and demographic characteristics; the Wilcoxon test that was used to make comparison when continuous variables were involved, the chi-squared test that sought to find a relationship between a dependent and independent variable and used for categorical data; permutation analysis where 2000 random permutations of the data were generated in which the order of the lethality for suicide attempts for each individual was scrambled; and correlation coefficient that tested whether lethality increased with each subsequent attempt.

The results of the tests (t-test, Wilcoxon, chi) tell us that subjects with high-lethality attempts were generally older and more likely to be married than the low-lethality attempters but showed no difference on gender of other demographic variables. High-lethality attempted were also more likely to be inpatients at the time of entry to the study than low-lethality attempters and had a greater number of prior psychiatric hospitalizations than low-lethality attempters. High lethality-attempters were also more seriously intent on suicide than low-lethality attempters were and had made a greater number of previous attempts. Permutation analysis / Spearman failed to find a pattern of increasing lethality with each subsequent attempt. However, it was discovered that multiple attempt individuals who had made early attempts at high lethality were more likely later to make subsequent attempts at high lethality.

Were hypothesis-testing errors present? What are the consequences for the study if a Type 1 or Type II error was made?

The study did not involve any prior hypothesis. Researchers were testing to identity clinical and behavioral characteristic of high-lethality suicide attempts in order to target at risk individuals with Bipolar Disorder. A Type 1 error in this case may be the postulation that certain characteristics signify a tendency to high lethality when in fact no such tendency exists. This would be less dangerous than were the reverse to be the case i.e. were the Type II error to be committed where researchers failed to identify characterizes pointing to high-lethality or rejected existence of these characteristics when in fact they did exist. With the possibility of a Type II error slipping through, many more high-lethal suicide attempters would remain unimpeded to continue their high-level suicide attempts. The study was through in collecting all pertinent information in order to assess possible correlations and in order to investigate related data. The clinical and diagnostic measures were also reliable and current. Their sources of data specifically the Columbia Suicide History form was authoritative too.

Are there additional studies in the same field that corroborate results of selected studies?

There are several studies in the field that corroborate researchers'...

Two example are that of Raja and Azzoni (2004) and Fagiolini et al. (2004).
Article 2. Parental communication style's impact on children's attitudes towards obesity and food advertising, Yu (2011).

Summary of article

Yu (2011) sought to investigate the relationship between obesity level of children and parental influence (specifically their styles of communication when watching TV) and to see whether this influenced the children's attitude towards advertised snack / fast-food ads. Researchers hypothesized that since children (7-12 years old) generally spend more time with parents than with anyone else, parental influence (specifically their styles of communication when watching TV) may have a major impact in influencing their diet as well as in influencing their attitudes towards TV advertised snack / fast-food ads. Parents of three elementary schools in the U.S. were contacted for this study. Surveys solicited basic demographic information as well as children's understanding of TV commercials before questioning their attitude toward TV advertised snack / fast-food ads. Parents were then questioned on the amount of time they spent watching TV with children, verbal interaction with their children and control over their children whilst watching TV. Researchers discovered that parental styles / communication significantly influenced children's obesity level and their attitude towards TV advertised snack / fast-food ads.

Are procedures parametric or non-parametric? Why?

Parametric tests require a normal distribution of the studied variable, the drawing of independent samples, and at least one variable being studied to be interval. They often also require a larger sample size than nonparametric tests do. Nonparametric tests are often used on nominal or ordinal data that are not normally distributed.

The procedure here is nonparametric: it is a survey that largely deals with nominal, ordinal and categorical data (in its open-ended questions). The chi-square goodness of fit test used to check whether the frequency of the most popular answer was statistically significant is a nonparametric test

A Type I error would not have been so egregious as a Type II error. With the Type I error, the researchers would merely have been wrong in their conclusion, namely that parental style and communication would not have impacted children's attitude to TV commercial and level of obesity (as Yu (2011) suggested. A Type II error on the other hand would have meant that another variable that Yu either overlooked or dismissed is responsible for level of obesity. This is more important since knowledge of this or more influencing variables may help researchers and involved individuals devise interventions for preventing ads dealing with these variables hence more effectively addressing childhood obesity.

What types of statistic tests were used? What do the results of these tests tell you about the data?

Appropriate scales were used to assess response. The chi-square goodness of fit test was used in order to check whether the frequency of the most popular answer was statistically significant. The test showed that most children equivocated the purpose of TV commercials to promoting and selling products. Regression analysis was also used. The hierarchical level was employed in order to prioritize data in a stepwise regression. Regression analysis showed that all three hypotheses - the amount of time parents spent watching TV with their children; the level of verbal interaction with children, as well as parental level of control over children-- had a significant positive influence on children's attachment to TV snack / fast food advertising.

Were hypothesis-testing errors present? What are the consequences for the study if a Type 1 or Type II error was made?

Researchers assumed that three factors influenced children's attitude towards TV snack / fast food advertising: the amount of time parents spent watching TV with children; their level of verbal interaction with children whilst watching TV; and parental level of control over children's TV viewing. To adequately and thoroughly test each of these hypotheses, therefore, each of these three factors have to be correctly measured. Errors are conceivably present in that surveys relied on self-report and both children and parents could conceivably err particularly in parent's assessment of their level of control over content of children's TV viewing time. The survey may also have led to other errors including misunderstanding or misinterpretation of questions; some questions being incomplete or incorrectly filled out whether intentionally or unintentionally, and researcher misinterpreting response. Accuracy would have been better achieved were objective investigation of parental interaction to have occurred.

Are there additional studies in the same field that corroborate results of selected studies?

Yu (2011) himself quotes various studies that are similar to his own. Livingston and Helsper (2006) and…